1 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3 * predict.c (maybe_hot_frequency_p): Do not use cfun.
5 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
10 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
12 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
13 enable generating XXBRH if the value is in a vector register.
14 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
15 value is in a vector register.
16 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
17 register to register bswap64's instead of doing the GPR sequence
18 used on previous machines.
19 (bswapdi2_xxbrd): New insn.
20 (bswapdi2_reg): Disallow on ISA 3.0.
21 (register to register bswap64 splitter): Do not split the insn on
22 ISA 3.0 systems that use XXBRD.
24 2017-11-10 Martin Sebor <msebor@redhat.com>
27 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
29 (find_subframework_file): Same.
31 2017-11-10 Jeff Law <law@redhat.com>
33 * vr-values.h (VR_INITIALIZER): Move #define here.
34 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
35 * Makefile.in (OBJS): Add tree-evrp.o
36 * tree-vrp.h (assert_info): Move structure definition here.
37 (set_value_range_to_varying): Prototype.
38 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
39 (infer_value_range, register_edge_assert_for): Likewise.
40 (stmt_interesting_for_vrp): Likewise.
41 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
42 (set_value_range_to_varying): No longer static.
43 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
44 (infer_value_range, register_edge_assert_for): Likewise.
46 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
48 * auto-profile.c (afdo_indirect_call): Drop frequency.
49 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
50 (cgraph_node::create_edge): Drop frequency argument.
51 (cgraph_node::create_indirect_edge): Drop frequency argument.
52 (cgraph_edge::make_speculative): Drop frequency arguments.
53 (cgraph_edge::resolve_speculation): Do not update frequencies
54 (cgraph_edge::dump_edge_flags): Do not dump frequency.
55 (cgraph_node::dump): Check consistency in IPA mode.
56 (cgraph_edge::maybe_hot_p): Use IPA counter.
57 (cgraph_edge::verify_count_and_frequency): Rename to ...
58 (cgraph_edge::verify_count): ... this one; drop frequency checking.
59 (cgraph_node::verify_node): Update.
60 * cgraph.h (struct cgraph_edge): Drop frequency.
61 (cgraph_edge::frequency): New function.
62 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
64 (cgraph_edge::rebuild_edges): Likewise.
65 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
66 (duplicate_thunk_for_node): Do not pass frequency.
67 (cgraph_node::create_clone): Scale only counts.
68 (cgraph_node::create_virtual_clone): Do not pass frequency.
69 (cgraph_node::create_edge_including_clones): Do not pass frequency.
70 (cgraph_node::create_version_clone): Do not pass frequency.
71 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
72 (cgraph_node::expand_thunk): Do not pass frequency.
73 (cgraph_node::create_wrapper): Do not pass frequency.
74 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
76 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
77 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
78 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
79 (gather_caller_stats): Use frequency function.
80 (ipcp_cloning_candidate_p): Use frequency function.
81 (ipcp_propagate_stage): Use frequency function.
82 (get_info_about_necessary_edges): Use frequency function.
83 (update_profiling_info): Update only IPA profile.
84 (update_specialized_profile): Use frequency functoin.
85 (perhaps_add_new_callers): Update only IPA profile.
86 * ipa-devirt.c (ipa_devirt): Use IPA profile.
87 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
88 (dump_ipa_call_summary): Use frequency function.
89 (estimate_edge_size_and_time): Use frequency function.
90 (ipa_merge_fn_summary_after_inlining): Use frequency function.
91 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
92 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
93 (update_noncloned_counts): ... ths one; scale counts only.
94 (clone_inlined_nodes): Do not scale frequency.
95 (inline_call): Do not pass frequency.
96 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
97 (compute_inlined_call_time): Use IPA profile.
98 (want_inline_small_function_p): Use IPA profile.
99 (want_inline_self_recursive_call_p): Use IPA profile.
100 (edge_badness): Use IPA profile.
101 (lookup_recursive_calls): Use IPA profile.
102 (recursive_inlining): Do not pass frequency.
103 (resolve_noninline_speculation): Do not update frequency.
104 (inline_small_functions): Collect max of IPA profile.
105 (dump_overall_stats): Dump IPA porfile.
106 (dump_inline_stats): Dump IPA porfile.
107 (ipa_inline): Collect IPA stats.
108 * ipa-inline.h (clone_inlined_nodes): Update prototype.
109 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
110 (ipa_propagate_frequency): Use frequency function.
111 (ipa_profile): Cleanup.
112 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
113 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
114 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
115 (input_node): Do not stream frequency.
116 (input_edge): Do not stream frequency.
117 (merge_profile_summaries): Scale only IPA profiles.
118 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
119 * predict.c (drop_profile): Do not recompute frequency.
120 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
121 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
122 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
123 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
124 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
125 * tree-inline.c (copy_bb): Do not scale frequency.
126 (expand_call_inline): Do not scale frequency.
127 (tree_function_versioning): Do not scale frequency.
128 * ubsan.c (ubsan_create_edge): Do not pass frequency.
130 2017-11-10 Julia Koval <julia.koval@intel.com>
132 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8,
133 _mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8,
134 _mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8,
135 _mm256_maskz_gf2p8affine_epi64_epi8,
136 _mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8
137 _mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
138 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi,
139 __builtin_ia32_vgf2p8affineqb_v32qi,
140 __builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
141 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
143 2017-11-10 Tamar Christina <tamar.christina@arm.com>
147 (arm_option_override): Refactor.
148 (arm_option_reconfigure_globals): New.
149 (arm_options_perform_arch_sanity_checks): New.
150 * config/arm/arm-protos.h (arm_option_reconfigure_globals): New prototype.
151 (arm_options_perform_arch_sanity_checks): Likewise
153 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
155 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
157 * rs6000/rs6000.md ("type" attr): Add qmul.
158 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
159 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
160 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
162 2017-11-10 Martin Sebor <msebor@redhat.com>
165 * builtins.c (compute_objsize): Handle arrays that
166 compute_builtin_object_size likes to fail for. Make extern.
167 * builtins.h (compute_objsize): Declare.
168 (check_strncpy_sizes): New function.
169 (expand_builtin_strncpy): Call check_strncpy_sizes.
170 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
171 -Wstringop-truncation.
172 (gimple_fold_builtin_strncat): Same.
173 * gimple.c (gimple_build_call_from_tree): Set call location.
174 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
175 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
176 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
177 (handle_builtin_strlen): Use strlen_to_stridx.
178 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
180 Use strlen_to_stridx.
181 (pass_strlen::execute): Release strlen_to_stridx.
182 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
183 (-Wstringop-truncation): Document new option.
185 2017-11-10 Martin Liska <mliska@suse.cz>
187 PR gcov-profile/82702
188 * gcov.c (main): Handle intermediate files in a different
190 (get_gcov_intermediate_filename): New function.
191 (output_gcov_file): Remove support of intermediate files.
192 (generate_results): Allocate intermediate file.
193 (release_structures): Clean-up properly fn_end.
194 (output_intermediate_file): Start iterating with line 1.
196 2017-11-10 Jakub Jelinek <jakub@redhat.com>
198 PR tree-optimization/82929
199 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
200 ops_swapped_p non-static data member.
201 (store_immediate_info::store_immediate_info): Clear it.
202 (imm_store_chain_info::coalesce_immediate_stores): If swapping
203 ops set ops_swapped_p.
204 (count_multiple_uses): Handle ops_swapped_p.
206 2017-11-10 Martin Liska <mliska@suse.cz>
208 * coverage.c (coverage_init): Stream information about
209 support of has_unexecuted_blocks.
210 * doc/gcov.texi: Document that.
211 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
212 * gcov.c (read_graph_file): Likewise.
213 (output_line_beginning): Fix a small issue with
216 2017-11-10 Bin Cheng <bin.cheng@arm.com>
218 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
219 reference of trivial component.
221 2017-11-10 Jakub Jelinek <jakub@redhat.com>
224 * gimple-ssa-store-merging.c
225 (pass_store_merging::terminate_all_aliasing_chains): For
226 gimple_store_p stmts also call refs_output_dependent_p.
228 PR rtl-optimization/82913
229 * compare-elim.c (try_merge_compare): Punt if def_insn is not
232 2017-11-09 Jeff Law <law@redhat.com>
234 * vr-values.h: New file with vr_values class.
235 * tree-vrp.c: Include vr-values.h
236 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
237 data objects into the vr_values class.
238 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
239 (get_value_range): Make it a member function within vr_values class.
240 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
241 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
242 (op_with_constant_singleton_value_range): Likewise.
243 (extract_range_for_var_from_comparison_expr): Likewise.
244 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
245 (extract_range_from_binary_expr): Likewise.
246 (extract_range_from_unary_expr): Likewise.
247 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
248 (check_for_binary_op_overflow, extract_range_basic): Likewise.
249 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
250 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
251 (compare_name_with_value, compare_names): Likewise.
252 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
253 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
254 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
255 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
256 (extract_range_from_phi_node): Likewise.
257 (simplify_truth_ops_using_ranges): Likewise.
258 (simplify_div_or_mod_using_ranges): Likewise.
259 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
260 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
261 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
262 (simplify_float_conversion_using_ranges): Likewise.
263 (simplify_internal_call_using_ranges): Likewise.
264 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
265 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
266 poitner via x_vr_values for calls into gimple folder.
267 (vrp_initialize_lattice): Make this the vr_values ctor.
268 (vrp_free_lattice): Make this the vr_values dtor.
269 (set_vr_value): New function.
270 (class vrp_prop): Add vr_values data member. Add various member
271 functions as well as member functions that delegate to vr_values.
272 (check_array_ref): Make a member function within vrp_prop class.
273 (search_for_addr_array, vrp_initialize): Likewise.
274 (vrp_finalize): Likewise. Revamp to avoid direct access to
275 vr_value, values_propagated, etc.
276 (check_array_bounds): Extract vrp_prop class instance pointer from
277 walk info structure. Use it to call member functions.
278 (check_all_array_refs): Make a member function within vrp_prop class.
279 Smuggle class instance pointer via walk info structure.
280 (x_vr_values): New local static.
281 (vrp_valueize): Use x_vr_values to get class instance.
282 (vr_valueize_1): Likewise.
283 (class vrp_folder): Add vr_values data member. Add various member
284 functions as well as member functions that delegate to vr_values.
285 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
286 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
287 class instance from vr_values. Use it to call member functions.
288 (vrp_dom_walker): Add vr_values data member.
289 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
290 instance via x_vr_values.
291 (identify_jump_threads): Accept vr_values as argument. Store
292 it into the walker structure.
293 (evrp_dom_walker): Add vr_values class data member. Add various
295 (evrp_dom_walker::try_find_new_range): Use vr_values data
296 member to access the memory allocator.
297 (evrp_dom_walker::before_dom_children): Store vr_values class
298 instance into the vrp_folder class.
299 (evrp_dom_walker::push_value_range): Rework to avoid direct
300 access to num_vr_values and vr_value.
301 (evrp_dom_walker::pop_value_range): Likewise.
302 (execute_early_vrp): Remove call to vrp_initialize_lattice.
303 Use vr_values to get to dump_all_value_ranges member function.
304 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
305 and simplify_cond_using_ranges_2 via vrp_prop class instance.
306 Pass vr_values class instance down to identify_jump_threads.
307 Remove call to vrp_free_lattice.
308 (debug_all_value_ranges): Remove.
310 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
311 (vrp_folder): Likewise.
313 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
314 Get it from the existing bitmap instead.
315 (vrp_intersect_ranges_1): Likewise.
317 2017-11-09 Jakub Jelinek <jakub@redhat.com>
319 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
321 (store_immediate_info::store_immediate_info): Add bitnotp argument,
323 (imm_store_chain_info::coalesce_immediate_stores): Break group
324 if bit_not_p is different.
325 (count_multiple_uses, split_group,
326 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
327 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
328 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
329 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
330 to store_immediate_info ctor.
332 2017-11-09 Jim Wilson <jimw@sifive.com>
334 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
335 (scan_prog_file): Likewise.
337 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
339 * bb-reorder.c (max_entry_frequency): Remove.
340 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
341 connect_traces, push_to_next_round_p): Remove prototypes.
342 (find_traces_1_round): Use counts only.
343 (push_to_next_round_p): Likewise.
344 (find_traces): Likewise.
345 (rotate_loop): Likewise.
346 (find_traces_1_round): Likewise.
347 (connect_traces): Likewise.
348 (edge_order): Likewise.
350 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
352 * config/arm/arm.c (output_return_instruction): Add comments to
353 indicate requirement for cmse_nonsecure_entry return to account
354 for the size of clearing instruction output here.
355 (thumb_exit): Likewise.
356 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
357 return in hardfloat mode.
359 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
361 * config/rs6000/rs6000.c (machine_function): Add a bool,
362 "toc_is_wrapped_separately".
363 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
364 if it wasn't explicitly set or unset, we are optimizing for speed, and
365 doing separate shrink-wrapping.
366 (rs6000_get_separate_components): Enable the TOC component if
367 saving the TOC register in the prologue.
368 (rs6000_components_for_bb): Handle the TOC component.
369 (rs6000_emit_prologue_components): Store the TOC register where needed.
370 (rs6000_set_handled_components): Mark TOC as handled, if handled.
371 (rs6000_emit_prologue): Don't save the TOC if that is already done.
373 2017-11-09 Martin Jambor <mjambor@suse.cz>
375 * ipa-param-manipulation.c: New file.
376 * ipa-param-manipulation.h: Likewise.
377 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
378 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
379 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
380 (ipa_parm_adjustment): Likewise.
381 (ipa_parm_adjustment_vec): Likewise.
382 (ipa_get_vector_of_formal_parms): Moved declaration to
383 ipa-param-manipulation.h.
384 (ipa_get_vector_of_formal_parm_types): Likewise.
385 (ipa_modify_formal_parameters): Likewise.
386 (ipa_modify_call_arguments): Likewise.
387 (ipa_combine_adjustments): Likewise.
388 (ipa_dump_param_adjustments): Likewise.
389 (ipa_modify_expr): Likewise.
390 (ipa_get_adjustment_candidate): Likewise.
391 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
392 ipa-param-manipulation.c.
393 (ipa_get_vector_of_formal_parm_types): Likewise.
394 (ipa_modify_formal_parameters): Likewise.
395 (ipa_modify_call_arguments): Likewise.
396 (ipa_modify_expr): Likewise.
397 (get_ssa_base_param): Likewise.
398 (ipa_get_adjustment_candidate): Likewise.
399 (index_in_adjustments_multiple_times_p): Likewise.
400 (ipa_combine_adjustments): Likewise.
401 (ipa_dump_param_adjustments): Likewise.
402 * tree-sra.c: Also include ipa-param-manipulation.h
403 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
406 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
407 Alan Hayward <alan.hayward@arm.com>
408 David Sherwood <david.sherwood@arm.com>
410 * doc/sourcebuild.texi (vect_masked_store): Document.
412 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
413 Alan Hayward <alan.hayward@arm.com>
414 David Sherwood <david.sherwood@arm.com>
416 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
418 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
419 Alan Hayward <alan.hayward@arm.com>
420 David Sherwood <david.sherwood@arm.com>
422 * doc/sourcebuild.texi (vect_variable_length): Document.
424 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
425 Alan Hayward <alan.hayward@arm.com>
426 David Sherwood <david.sherwood@arm.com>
428 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
430 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
431 Alan Hayward <alan.hayward@arm.com>
432 David Sherwood <david.sherwood@arm.com>
434 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
436 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
437 Alan Hayward <alan.hayward@arm.com>
438 David Sherwood <david.sherwood@arm.com>
440 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
441 previously undocumented selectors.
442 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
444 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
446 * doc/rtl.texi (const_vector): Say that elements can be
448 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
449 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
450 (gen_vec_duplicate): Use it instead of CONSTANT_P.
451 * optabs.c (expand_vector_broadcast): Likewise.
453 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
454 Alan Hayward <alan.hayward@arm.com>
455 David Sherwood <david.sherwood@arm.com>
457 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
458 scaled index even if the unscaled address was invalid.
459 Don't increase the complexity of using a scale in that case.
461 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
462 Alan Hayward <alan.hayward@arm.com>
463 David Sherwood <david.sherwood@arm.com>
465 * doc/rtl.texi: Rewrite the subreg rules so that they partition
466 the inner register into REGMODE_NATURAL_SIZE bytes rather than
467 UNITS_PER_WORD bytes.
468 * emit-rtl.c (validate_subreg): Divide subregs into blocks
469 based on REGMODE_NATURAL_SIZE of the inner mode.
470 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
471 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
472 * expmed.c (lowpart_bit_field_p): Divide the value up into
473 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
474 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
475 whether something is likely to occupy more than one register.
477 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
480 * ipa-inline-transform.c (update_noncloned_frequencies): Use
481 profile_count::adjust_for_ipa_scaling.
482 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
483 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
485 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
487 2017-11-09 Jakub Jelinek <jakub@redhat.com>
489 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
490 (split_group): Add total_orig and total_new arguments, estimate the
491 number of statements related to the store group without store merging
492 and with store merging.
493 (imm_store_chain_info::output_merged_store): Adjust split_group
494 callers, punt if estimated number of statements with store merging
495 is not smaller than estimated number of statements without it.
497 (handled_load): Remove has_single_use checks.
498 (pass_store_merging::process_store): Likewise.
500 2017-11-09 Richard Biener <rguenther@suse.de>
502 PR tree-optimization/82902
503 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
505 2017-11-09 Martin Liska <mliska@suse.cz>
508 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
511 2017-11-09 Martin Liska <mliska@suse.cz>
513 PR tree-optimization/82669
514 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
516 2017-11-09 Martin Liska <mliska@suse.cz>
518 PR gcov-profile/48463
519 * coverage.c (coverage_begin_function): Output also end locus
520 of a function and information whether the function is
522 * gcov-dump.c (tag_function): Parse and print the information.
523 * gcov.c (INCLUDE_MAP): Add include.
524 (INCLUDE_SET): Likewise.
525 (struct line_info): Move earlier in the source file because
526 of vector<line_info> in function_info structure.
527 (line_info::line_info): Likewise.
528 (line_info::has_block): Likewise.
529 (struct source_info): Add new member index.
530 (source_info::get_functions_at_location): New function.
531 (function_info::group_line_p): New function.
532 (output_intermediate_line): New function.
533 (output_intermediate_file): Use the mentioned function.
534 (struct function_start): New.
535 (struct function_start_pair_hash): Likewise.
536 (process_file): Add code that identifies group functions.
537 Assign lines either to global or function scope.
538 (generate_results): Skip artificial functions.
539 (find_source): Assign index for each source file.
540 (read_graph_file): Read new flag artificial and end_line.
541 (add_line_counts): Assign it either to global of function scope.
542 (accumulate_line_counts): Isolate core of the function to
543 accumulate_line_info and call it for both function and global
545 (accumulate_line_info): New function.
546 (output_line_beginning): Fix GNU coding style.
547 (print_source_line): New function.
548 (output_line_details): Likewise.
549 (output_function_details): Likewise.
550 (output_lines): Iterate both source (global) scope and function
552 (struct function_line_start_cmp): New class.
553 * doc/gcov.texi: Reflect changes in documentation.
555 2017-11-09 Jakub Jelinek <jakub@redhat.com>
558 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
559 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
560 and similarly for not instead of neg.
562 2017-11-08 Andi Kleen <ak@linux.intel.com>
564 * config/i386/i386.opt: Add -mforce-indirect-call.
565 * config/i386/predicates.md: Check for flag_force_indirect_call.
566 * doc/invoke.texi: Document -mforce-indirect-call
568 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
570 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
572 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
573 config/riscv/riscv.c (predict.h): New include.
574 (riscv_slow_unaligned_access_p): No longer static.
575 (riscv_block_move_straight): Add require.
576 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
578 2017-11-08 Jakub Jelinek <jakub@redhat.com>
581 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
582 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
583 nonimmediate_operand predicate for operand 1 instead of
586 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
588 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
590 * config/aarch64/constraints.md (Uml): New constraint.
591 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
594 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
596 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
597 of two vec_duplicates into a vec_concat.
599 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
601 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
602 Simplify vec_merge of vec_duplicate and vec_concat.
603 * config/aarch64/constraints.md (Utq): New constraint.
604 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
607 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
609 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
610 Simplify vec_merge of vec_duplicate and const_vector.
611 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
613 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
614 mode iterator. Update predicate on operand 1 to
615 handle non-const_vec constants. Delete constraints.
616 (*aarch64_combinez_be<mode>): Likewise for operand 2.
618 2017-11-08 Jakub Jelinek <jakub@redhat.com>
620 PR tree-optimization/78821
621 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
623 (store_operand_info::store_operand_info): Initialize it to false.
624 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
625 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
626 store in the group, and if chain_info is non-NULL, to ignore altogether
628 (compatible_load_p): Fail if bit_not_p does not match.
629 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
630 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
631 (pass_store_merging::process_store): Adjust
632 terminate_all_aliasing_chains calls to pass NULL in all current spots,
633 call terminate_all_aliasing_chains newly when adding a store into
634 a chain with non-NULL chain_info.
636 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
638 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
640 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
642 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
644 (aarch64_layout_frame): Initialise emit_frame_chain.
645 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
646 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
648 2017-11-08 Martin Liska <mliska@suse.cz>
650 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
651 of gimple_call_internal_p.
653 2017-11-07 Tom de Vries <tom@codesourcery.com>
655 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
657 2017-11-07 Tom de Vries <tom@codesourcery.com>
659 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
662 2017-11-08 Martin Liska <mliska@suse.cz>
665 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
667 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
669 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
671 2017-11-07 Jakub Jelinek <jakub@redhat.com>
674 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
675 (*cmp<mode>_ccz_1): New insn with $k alternative.
678 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
679 RTX_COMM_COMPARE as commutative as well.
680 (ix86_binary_operator_ok): Formatting fix.
681 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
682 *<code><mode>3<mask_name><round_saeonly_name>,
683 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
684 *<s>mul<mode>3_highpart<mask_name>,
685 *vec_widen_umult_even_v16si<mask_name>,
686 *vec_widen_umult_even_v8si<mask_name>,
687 *vec_widen_umult_even_v4si<mask_name>,
688 *vec_widen_smult_even_v16si<mask_name>,
689 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
690 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
691 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
692 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
693 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
694 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
695 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
696 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
697 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
698 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
699 ix86_binary_operator_ok. Formatting fixes.
700 (*<plusminus_insn><mode>3<mask_name><round_name>,
701 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
704 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
706 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
707 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
708 eq and ne if TARGET_ISEL.
709 (cmp): New code_iterator.
710 (UNS, UNSU_, UNSIK): New code_attrs.
711 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
712 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
714 ("eq<mode>3"): ... this.
715 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
717 ("ne<mode>3"): ... this.
719 2017-11-07 Julia Koval <julia.koval@intel.com>
722 * common/config/i386/i386-common.c
723 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
724 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
725 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
726 * config/i386/i386.opt: Ditto.
727 * config/i386/i386.c (ix86_target_string): Ditto.
728 (ix86_option_override_internal): Ditto.
729 (ix86_init_mpx_builtins): Move MPX to args2.
730 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
731 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
732 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
733 __builtin_ia32_vgf2p8affineinvqb_v32qi,
734 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
735 __builtin_ia32_vgf2p8affineinvqb_v16qi,
736 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
738 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
741 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
742 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
743 (zero-extendsidi peephole2): Remove peephole.
745 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
748 * stor-layout.c: Include attribs.h.
749 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
750 explicit lookup of "aligned" attribute.
752 2017-11-07 Andrew Waterman <andrew@sifive.com>
754 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
755 (riscv_expand_block_move): Likewise.
756 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
758 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
759 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
760 * config/riscv/riscv.c (riscv_block_move_straight): New function.
761 (riscv_adjust_block_mem): Likewise.
762 (riscv_block_move_loop): Likewise.
763 (riscv_expand_block_move): Likewise.
764 gcc/config/riscv/riscv.md (movmemsi): New pattern.
766 2017-11-07 Michael Clark <michaeljclark@mac.com>
768 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
769 (MUSL_DYNAMIC_LINKER): Likewise.
771 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
773 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
774 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
777 2017-11-07 Richard Biener <rguenther@suse.de>
779 * match.pd: Fix build.
781 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
782 Jackson Woodruff <jackson.woodruff@arm.com>
784 PR tree-optimization/71026
785 * match.pd: Canonicalize negate in division.
787 2017-11-07 Sudakshina Das <sudi.das@arm.com>
790 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
792 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
794 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
795 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
797 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
799 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
800 non-scalar integral types.
801 * match.pd (negate_expr_p): Handle MINUS_EXPR.
802 (-(A-B), -(~A)): New transformations.
804 2017-11-07 Tom de Vries <tom@codesourcery.com>
806 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
807 semicolon after "do {} while (0)".
808 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
809 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
810 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
811 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
812 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
814 2017-11-07 Tom de Vries <tom@codesourcery.com>
816 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
817 after "do {} while (0)".
818 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
819 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
820 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
821 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
822 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
824 2017-11-07 Tom de Vries <tom@codesourcery.com>
827 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
829 (arm_rtx_costs_internal): Add missing semicolon after
830 HANDLE_NARROW_SHIFT_ARITH call.
832 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
834 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
835 disable isel if it was not set explicitly.
837 2017-11-06 James Bowman <james.bowman@ftdichip.com>
839 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
841 (add_type_attribute) likewise.
843 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
845 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
846 of struct ix86_frame.
847 (ix86_initial_elimination_offset): Likewise.
848 (ix86_expand_split_stack_prologue): Likewise.
850 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
852 * tree-vrp.h (enum value_range_type): Update stale comment.
854 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
855 Alan Hayward <alan.hayward@arm.com>
856 David Sherwood <david.sherwood@arm.com>
858 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
859 (aarch64_expand_vec_perm_const): Take the number of units too.
860 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
861 (aarch64_expand_vec_perm_const): Likewise.
862 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
863 (vec_perm<mode>): Update accordingly.
865 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
866 Alan Hayward <alan.hayward@arm.com>
867 David Sherwood <david.sherwood@arm.com>
869 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
870 Take the number of units too.
871 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
872 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
873 but check for a vector mode before rather than after the call.
874 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
875 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
876 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
877 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
878 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
879 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
880 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
881 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
882 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
883 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
884 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
885 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
886 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
887 (aarch64_sqdmull2_n<mode>): Update accordingly.
889 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
890 Alan Hayward <alan.hayward@arm.com>
891 David Sherwood <david.sherwood@arm.com>
893 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
894 the number of units too.
895 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
896 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
897 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
898 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
899 (vec_store_lanesxi<mode>): Update accordingly.
901 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
902 Alan Hayward <alan.hayward@arm.com>
903 David Sherwood <david.sherwood@arm.com>
905 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
906 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
907 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
908 of units rather than the mode.
909 * config/aarch64/iterators.md (nunits): New mode attribute.
910 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
911 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
912 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
913 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
914 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
915 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
916 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
917 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
918 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
919 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
920 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
921 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
922 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
923 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
924 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
925 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
926 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
927 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
928 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
929 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
930 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
931 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
932 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
933 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
934 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
935 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
936 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
937 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
938 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
939 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
940 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
941 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
942 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
943 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
944 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
945 (aarch64_simd_vec_setv2di): Likewise.
947 2017-11-06 Carl Love <cel@us.ibm.com>
949 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
951 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
952 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
954 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
956 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
957 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
958 Add power 8 macro expansions.
959 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
960 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
961 power 8 instructions. (VSX_XXBR): Add iterator.
963 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
965 * config/arm/arm.md (predicable_short_it): Change default to "no",
966 improve documentation, remove uses that are identical to the default.
967 (enabled_for_depr_it): Rename to enabled_for_short_it.
968 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
969 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
970 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
971 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
972 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
974 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
977 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
978 float128 helper macros, which are no longer used after deleting
979 the old 'q' built-in functions, and moving the round to odd
980 built-in functions to being special built-in functions.
981 (BU_FLOAT128_2): Likewise.
982 (BU_FLOAT128_1_HW): Likewise.
983 (BU_FLOAT128_2_HW): Likewise.
984 (BU_FLOAT128_3_HW): Likewise.
985 (FABSQ): Delete old 'q' built-in functions.
986 (COPYSIGNQ): Likewise.
987 (SQRTF128_ODD): Move round to odd built-in functions to be
988 special built-in functions, so that we can handle
989 -mabi=ieeelongdouble.
990 (TRUNCF128_ODD): Likewise.
991 (ADDF128_ODD): Likewise.
992 (SUBF128_ODD): Likewise.
993 (MULF128_ODD): Likewise.
994 (DIVF128_ODD): Likewise.
995 (FMAF128_ODD): Likewise.
996 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
997 built-in names to 'f128'.
998 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
999 old 'q' built-in functions, as the machine independent code for
1000 'f128' built-in functions handles this.
1001 (rs6000_expand_builtin): Add expansion for float128 round to odd
1002 functions, keying off on -mabi=ieeelongdouble of whether to use
1003 the KFmode or TFmode variant.
1004 (rs6000_init_builtins): Initialize the _Float128 round to odd
1006 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1007 _Float128 'q' built-in functions are now mapped into the new
1008 'f128' built-in functions.
1010 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1012 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1013 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1015 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1018 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1019 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1020 * toplev.c: Include "ipa-fnsummary.h".
1021 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1023 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1025 PR tree-optimization/82838
1026 * gimple-ssa-store-merging.c
1027 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1028 on a separate gimple_seq which is then appended to seq.
1030 2017-11-06 Jeff Law <law@redhat.com>
1033 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1034 (get_probe_interval): New functions.
1035 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1036 (ix86_adjust_stack_and_probe): Likewise.
1037 (output_adjust_stack_and_probe): Likewise.
1038 (ix86_emit_probe_stack_range): Likewise.
1039 (ix86_expand_prologue): Likewise.
1041 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1043 PR tree-optimization/82816
1044 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1045 if the modes of the two types are the same.
1046 (convert_plusminus_to_widen): Likewise.
1048 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1050 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1052 (usadv16qi): New define_expand.
1053 (usadv8hi): New define_expand.
1055 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1058 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1060 (inline_transform): Likewise.
1061 * predict.c (counts_to_freqs): Remove useless conditional.
1062 * profile-count.h (profile_count::apply_scale): Move sanity check.
1063 * tree-inline.c (copy_bb): Always scale.
1064 (copy_cfg_body): Likewise.
1066 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1069 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1071 (*cmp_ite0): Add enabled_for_depr_it attribute.
1072 (*cmp_ite1): Likewise.
1074 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1076 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1079 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1081 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1082 return the value by pointer.
1083 (extract_range_from_multiplicative_op_1): Update accordingly.
1084 Return as soon as an operation fails.
1086 2017-11-05 Tom de Vries <tom@codesourcery.com>
1089 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1090 (DEF_SANITIZER_BUILTIN): ... here.
1091 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1092 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1094 2017-11-05 Tom de Vries <tom@codesourcery.com>
1097 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1099 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1100 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1101 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1103 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1104 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1105 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1107 2017-11-05 Tom de Vries <tom@codesourcery.com>
1110 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1113 2017-11-04 Michael Clark <michaeljclark@mac.com>
1115 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1116 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1118 (*addsi3_extended): Likewise.
1119 (*addsi3_extended2): Likewise.
1120 (<optab>si3): Likewise.
1121 (<optab>di3): Likewise.
1122 (<optab><mode>3): Likewise.
1123 (<*optabe>si3_internal): Likewise.
1124 (zero_extendqi<SUPERQI:mode>2): Likewise.
1125 (*add<mode>hi3): Likewise.
1126 (*xor<mode>hi3): Likewise.
1127 (<optab>di3): Likewise.
1128 (*<optab>si3_extend): Likewise.
1129 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1130 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1131 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1133 2017-11-04 Andrew Waterman <andrew@sifive.com>
1135 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1136 TARGET_STRICT_ALIGN based upon -mtune argument.
1138 2017-11-04 Andrew Waterman <andrew@sifive.com>
1140 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1142 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1144 config/i386/i386.c (choose_basereg): Use optional scratch
1145 register and add assertion.
1146 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1147 needed, and don't allocate stack.
1148 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1149 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1150 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1152 2017-11-03 Jeff Law <law@redhat.com>
1154 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1155 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1156 to probe at the start of a noreturn function.
1158 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1160 PR tree-optimization/78821
1161 * gimple-ssa-store-merging.c: Update the file comment.
1162 (MAX_STORE_ALIAS_CHECKS): Define.
1163 (struct store_operand_info): New type.
1164 (store_operand_info::store_operand_info): New constructor.
1165 (struct store_immediate_info): Add rhs_code and ops data members.
1166 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1167 and op1r arguments to the ctor, initialize corresponding data members.
1168 (struct merged_store_group): Add load_align_base and load_align
1170 (merged_store_group::merged_store_group): Initialize them.
1171 (merged_store_group::do_merge): Update them.
1172 (merged_store_group::apply_stores): Pick the constant for
1173 encode_tree_to_bitpos from one of the two operands, or skip
1174 encode_tree_to_bitpos if neither operand is a constant.
1175 (class pass_store_merging): Add process_store method decl. Remove
1176 bool argument from terminate_all_aliasing_chains method decl.
1177 (pass_store_merging::terminate_all_aliasing_chains): Remove
1178 var_offset_p argument and corresponding handling.
1179 (stmts_may_clobber_ref_p): New function.
1180 (compatible_load_p): New function.
1181 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1182 if there is overlap and rhs_code is not INTEGER_CST. For
1183 non-overlapping stores terminate group if rhs is not mergeable.
1184 (get_alias_type_for_stmts): Change first argument from
1185 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1186 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1187 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1189 (get_location_for_stmts): Change first argument from
1190 auto_vec<gimple *> & to vec<gimple *> &.
1191 (struct split_store): Remove orig_stmts data member, add orig_stores.
1192 (split_store::split_store): Create orig_stores rather than orig_stmts.
1193 (find_constituent_stmts): Renamed to ...
1194 (find_constituent_stores): ... this. Change second argument from
1195 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1196 to info structures rather than the statements.
1197 (split_group): Rename ALLOW_UNALIGNED argument to
1198 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1199 it. Adjust find_constituent_stores caller.
1200 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1201 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1202 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1203 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1204 (mem_valid_for_store_merging): New function.
1205 (handled_load): New function.
1206 (pass_store_merging::process_store): New method.
1207 (pass_store_merging::execute): Use process_store method. Adjust
1208 terminate_all_aliasing_chains caller.
1210 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1212 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1213 Return true for more constants, symbols and label references.
1214 (aarch64_valid_floating_const): Remove unused function.
1216 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1219 * config/aarch64/aarch64.c (aarch64_layout_frame):
1220 Undo forcing of LR at bottom of frame.
1222 2017-11-03 Jeff Law <law@redhat.com>
1225 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1226 for int_registers_saved.
1228 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1229 extracted from tree-ssa-dom.c.
1230 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1231 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1232 (record_equivalences_from_incoming_edge): Add additional argument
1233 to single_pred_edge_ignoring_loop_edges call.
1234 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1235 (uncprop_dom_walker::before_dom_children): Add additional argument
1236 to single_pred_edge_ignoring_loop_edges call.
1237 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1238 single_pred_edge_ignoring_loop_edges rather than open coding.
1239 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1241 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1243 * match.pd (-(-A)): Rewrite.
1245 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1247 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1248 (rs6000_emit_int_cmove): New declaration.
1249 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1250 (rs6000_emit_sISEL): Delete.
1251 (rs6000_emit_int_cmove): Make non-static.
1252 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1253 instead of rs6000_emit_sISEL.
1255 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1257 * asan.c (create_cond_insert_point): Maintain profile.
1258 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1260 * basic-block.h (struct basic_block_def): Remove frequency.
1261 (EDGE_FREQUENCY): Use to_frequency
1262 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1264 (find_traces): Update to use to_frequency.
1265 (find_traces_1_round): Likewise; use only IPA counts.
1266 (bb_to_key): Likewise.
1267 (connect_traces): Use IPA counts only.
1268 (copy_bb_p): Update to use to_frequency.
1269 (fix_up_crossing_landing_pad): Likewise.
1270 (sanitize_hot_paths): Likewise.
1271 * bt-load.c (basic_block_freq): Likewise.
1272 * cfg.c (init_flow): Set count_max to uninitialized.
1273 (check_bb_profile): Remove frequencies; check counts.
1274 (dump_bb_info): Do not dump frequencies.
1275 (update_bb_profile_for_threading): Update counts only.
1276 (scale_bbs_frequencies_int): Likewise.
1277 (MAX_SAFE_MULTIPLIER): Remove.
1278 (scale_bbs_frequencies_gcov_type): Update counts only.
1279 (scale_bbs_frequencies_profile_count): Update counts only.
1280 (scale_bbs_frequencies): Update counts only.
1281 * cfg.h (struct control_flow_graph): Add count-max.
1282 (update_bb_profile_for_threading): Update prototype.
1283 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1284 (find_many_sub_basic_blocks): Likewise.
1285 * cfgcleanup.c (try_forward_edges): Likewise.
1286 (try_crossjump_to_edge): Likewise.
1287 * cfgexpand.c (expand_gimple_cond): Likewise.
1288 (expand_gimple_tailcall): Likewise.
1289 (construct_init_block): Likewise.
1290 (construct_exit_block): Likewise.
1291 * cfghooks.c (verify_flow_info): Check consistency of counts.
1292 (dump_bb_for_graph): Do not dump frequencies.
1293 (split_block_1): Do not update frequencies.
1294 (split_edge): Do not update frequencies.
1295 (make_forwarder_block): Do not update frequencies.
1296 (duplicate_block): Do not update frequencies.
1297 (account_profile_record): Do not update frequencies.
1298 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1299 for global heuristics.
1300 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1301 (expected_loop_iterations_unbounded): Use counts only.
1302 * cfgloopmanip.c (scale_loop_profile): Simplify.
1303 (create_empty_loop_on_edge): Simplify
1305 (duplicate_loop_to_header_edge): Simplify
1306 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1307 (update_br_prob_note): Take care of removing note when profile
1309 (relink_block_chain): Do not dump frequency.
1310 (rtl_account_profile_record): Use to_frequency.
1311 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1312 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1313 (cgraph_update_edges_for_call_stmt_node): Likewise.
1314 (cgraph_edge::verify_count_and_frequency): Update.
1315 (cgraph_node::verify_node): Temporarily disable frequency verification.
1316 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1317 to_cgraph_frequency.
1318 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1319 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1321 (cgraph_node::expand_thunk): Update profile.
1322 * except.c (dw2_build_landing_pads): Do not update frequency.
1323 * final.c (compute_alignments): Use to_frequency.
1324 (dump_basic_block_info): Do not dump frequency.
1325 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1326 (dump_gimple_bb_header): Do not dump frequency.
1327 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1329 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1330 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1331 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1332 (init_before_recovery): Do not update frequency.
1333 (sched_create_recovery_edges): Do not update frequency.
1334 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1335 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1336 (ipa_cp_c_finalize): Set max_count to uninitialized.
1337 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1338 (param_change_prob): Use counts.
1339 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1341 * ipa-split.c (consider_split): Use to_frequency.
1342 (split_function): Use to_frequency.
1343 * ira-build.c (loop_compare_func): Likewise.
1344 (mark_loops_for_removal): Likewise.
1345 (mark_all_loops_for_removal): Likewise.
1346 * loop-doloop.c (doloop_modify): Do not update frequency.
1347 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1349 * lto-streamer-in.c (input_function): Update count_max.
1350 * omp-expand.c (expand_omp_taskreg): Update count_max.
1351 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1352 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1353 (maybe_hot_count_p): Use ipa counts only.
1354 (maybe_hot_bb_p): Simplify.
1355 (maybe_hot_edge_p): Simplify.
1356 (probably_never_executed): Do not take frequency argument.
1357 (probably_never_executed_bb_p): Do not pass frequency.
1358 (probably_never_executed_edge_p): Likewise.
1359 (combine_predictions_for_bb): Check that profile is nonzero.
1360 (propagate_freq): Do not set frequency.
1361 (drop_profile): Simplify.
1362 (counts_to_freqs): Simplify.
1363 (expensive_function_p): Use to_frequency.
1364 (propagate_unlikely_bbs_forward): Simplify.
1365 (determine_unlikely_bbs): Simplify.
1366 (estimate_bb_frequencies): Add hack to silence graphite issues.
1367 (compute_function_frequency): Use ipa counts.
1368 (pass_profile::execute): Update.
1369 (rebuild_frequencies): Use counts only.
1370 (force_edge_cold): Use counts only.
1371 * profile-count.c (profile_count::dump): Dump new count types.
1372 (profile_count::differs_from_p): Check compatiblity.
1373 (profile_count::to_frequency): New function.
1374 (profile_count::to_cgraph_frequency): New function.
1375 * profile-count.h (struct function): Declare.
1376 (enum profile_quality): Add profile_guessed_local and
1377 profile_guessed_global0.
1378 (class profile_proability): Decrease number of bits to 29;
1379 update from_reg_br_prob_note and to_reg_br_prob_note.
1380 (class profile_count: Update comment; decrease number of bits
1381 to 61. Check compatibility.
1382 (profile_count::compatible_p): New private member function.
1383 (profile_count::ipa_p): New member function.
1384 (profile_count::operator<): Handle global zero correctly.
1385 (profile_count::operator>): Handle global zero correctly.
1386 (profile_count::operator<=): Handle global zero correctly.
1387 (profile_count::operator>=): Handle global zero correctly.
1388 (profile_count::nonzero_p): New member function.
1389 (profile_count::force_nonzero): New member function.
1390 (profile_count::max): New member function.
1391 (profile_count::apply_scale): Handle IPA scalling.
1392 (profile_count::guessed_local): New member function.
1393 (profile_count::global0): New member function.
1394 (profile_count::ipa): New member function.
1395 (profile_count::to_frequency): Declare.
1396 (profile_count::to_cgraph_frequency): Declare.
1397 * profile.c (OVERLAP_BASE): Delete.
1398 (compute_frequency_overlap): Delete.
1399 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1400 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1401 * sched-ebb.c (rank): Use counts only.
1402 * shrink-wrap.c (handle_simple_exit): Use counts only.
1403 (try_shrink_wrapping): Use counts only.
1404 (place_prologue_for_one_component): Use counts only.
1405 * tracer.c (find_best_predecessor): Use to_frequency.
1406 (find_trace): Use to_frequency.
1407 (tail_duplicate): Use to_frequency.
1408 * trans-mem.c (expand_transaction): Do not update frequency.
1409 * tree-call-cdce.c: Do not update frequency.
1410 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1411 (gimple_merge_blocks): Likewise.
1412 (gimple_split_edge): Likewise.
1413 (gimple_duplicate_sese_region): Likewise.
1414 (gimple_duplicate_sese_tail): Likewise.
1415 (move_sese_region_to_fn): Likewise.
1416 (gimple_account_profile_record): Likewise.
1417 (insert_cond_bb): Likewise.
1418 * tree-complex.c (expand_complex_div_wide): Likewise.
1419 * tree-eh.c (lower_resx): Update profile.
1420 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1422 (initialize_cfun): Do not initialize frequencies
1423 (freqs_to_counts): Delete.
1424 (copy_cfg_body): Ignore count parameter.
1425 (copy_body): Update.
1426 (expand_call_inline): Update count_max.
1427 (optimize_inline_calls): Update count_max.
1428 (tree_function_versioning): Update count_max.
1429 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1430 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1432 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1433 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1434 (try_peel_loop): Likewise.
1435 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1437 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1438 (tree_transform_and_unroll_loop): Do not use frequencies
1439 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1440 Use reliable prediction only.
1441 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1442 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1443 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1444 probability scaling.
1445 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1446 not update frequency
1447 (any_remaining_duplicated_blocks): Likewise.
1448 (update_profile): Likewise.
1449 (estimated_freqs_path): Delete.
1450 (freqs_to_counts_path): Delete.
1451 (clear_counts_path): Delete.
1452 (ssa_fix_duplicate_block_edges): Likewise.
1453 (duplicate_thread_path): Likewise.
1454 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1455 * tree-tailcall.c (decrease_profile): Do not update frequency.
1456 (eliminate_tail_call): Likewise.
1457 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1458 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1459 (optimize_mask_stores): Likewise.
1460 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1461 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1462 (ubsan_expand_ptr_ifn): Update profile.
1463 * value-prof.c (gimple_ic): Simplify.
1464 * value-prof.h (gimple_ic): Update prototype.
1465 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1466 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1468 (want_inline_self_recursive_call_p): Likewise.
1469 (resolve_noninline_speculation): Only cummulate defined counts.
1470 (inline_small_functions): Use nonzero_p.
1471 (ipa_inline): Do not access freed node.
1473 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1475 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1476 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1478 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1480 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1481 non-legitimate address.
1483 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1485 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1486 (*lt0_<mode>di, *lt0_<mode>si): New.
1488 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1490 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1491 TARGET_PAIRED_FLOAT.
1493 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1494 Jim Wilson <jim.wilson@linaro.org>
1496 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1497 * config/aarch64/aarch64-tune.md: Regenerated.
1498 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1499 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1501 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1503 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1504 (arc_expand_prologue): Restore blink for millicode.
1505 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1507 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1510 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1511 gen_vec_duplicate after forcing the scalar into a register.
1513 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1515 * combine (try_combine): Print the insns input to try_combine to the
1518 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1521 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1522 Remove second argument from aarch64_process_target_attr call.
1523 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1525 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1527 (aarch64_handle_attr_arch): Remove second argument.
1528 (aarch64_handle_attr_cpu): Ditto.
1529 (aarch64_handle_attr_tune): Ditto.
1530 (aarch64_handle_attr_isa_flags): Ditto.
1531 (aarch64_process_one_target_attr): Ditto.
1532 (aarch64_process_target_attr): Ditto.
1533 (aarch64_option_valid_attribute_p): Remove second argument.
1534 on aarch64_process_target_attr call.
1536 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1538 * diagnostic.c: Include "selftest-diagnostic.h".
1539 (selftest::assert_location_text): New function.
1540 (selftest::test_diagnostic_get_location_text): New function.
1541 (selftest::diagnostic_c_tests): Call it.
1543 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1545 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1546 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1547 (class selftest::test_diagnostic_context): Move to...
1548 * selftest-diagnostic.c: New file.
1549 * selftest-diagnostic.h: New file.
1551 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1553 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1554 offset range for FT32B.
1555 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1556 * config/ft32/ft32.md: Add TARGET_NOPM.
1557 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1558 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1560 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1562 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1564 2017-11-02 Jeff Law <law@redhat.com>
1566 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1567 virtual keyword on FINAL OVERRIDE members.
1569 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1570 virtuals together. Add virtual destructor.
1571 (substitute_and_fold_engine): Similarly.
1573 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1575 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1577 2017-11-02 Richard Biener <rguenther@suse.de>
1579 PR tree-optimization/82795
1580 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1582 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1584 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1586 (gcc_GAS_CHECK_FEATURE): Remove.
1587 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1588 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1589 * configure: Regenerate.
1591 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1593 * config/arc/arc.c (hwloop_optimize): Account for empty
1596 2017-11-02 Richard Biener <rguenther@suse.de>
1599 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1600 Truncate ARRAY_REF index and element size.
1602 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1604 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1606 2017-11-01 Jeff Law <law@redhat.com>
1608 * tree-ssa-ccp.c (ccp_folder): New class derived from
1609 substitute_and_fold_engine.
1610 (ccp_folder::get_value): New member function.
1611 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1612 (ccp_fold_stmt): Remove prototype.
1613 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1614 * tree-ssa-copy.c (copy_folder): New class derived from
1615 substitute_and_fold_engine.
1616 (copy_folder::get_value): Renamed from get_value.
1617 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1618 * tree-vrp.c (vrp_folder): New class derived from
1619 substitute_and_fold_engine.
1620 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1621 (vrp_folder::get_value): New member function.
1622 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1623 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1624 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1625 provide a class interface to folder/substitute routines.
1626 (ssa_prop_fold_stmt_fn): Remove typedef.
1627 (ssa_prop_get_value_fn): Likewise.
1628 (subsitute_and_fold): Remove prototype.
1629 (replace_uses_in): Likewise.
1630 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1631 Renamed from replace_uses_in. Call the virtual member function
1632 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1633 (substitute_and_fold_dom_walker): Remove initialization of
1634 data member entries for calbacks. Add substitute_and_fold_engine
1635 member and initialize it.
1636 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1637 member functions for get_value, replace_phi_args_in c
1638 replace_uses_in, and fold_stmt calls.
1639 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1640 substitute_and_fold. Remove assert. Update ctor call.
1642 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1643 (ssa_prop_visit_phi_fn): Likewise.
1644 (class ssa_propagation_engine): New class to provide an interface
1646 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1648 (ssa_prop_visit_phi): Likewise.
1649 (ssa_propagation_engine::simulate_stmt): Moved into class.
1650 Call visit_phi/visit_stmt from the class rather than via
1651 file scoped static variables.
1652 (ssa_propagation_engine::simulate_block): Moved into class.
1653 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1654 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1655 set file scoped statics for the visit_stmt/visit_phi callbacks.
1656 * tree-complex.c (complex_propagate): New class derived from
1657 ssa_propagation_engine.
1658 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1659 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1660 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1662 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1663 ssa_propagation_engine.
1664 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1665 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1666 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1667 * tree-ssa-copy.c (copy_prop): New class derived from
1668 ssa_propagation_engine.
1669 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1670 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1671 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1672 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1673 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1674 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1675 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1677 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1679 PR rtl-optimization/82778
1680 PR rtl-optimization/82597
1681 * compare-elim.c (struct comparison): Add in_a_setter field.
1682 (find_comparison_dom_walker::before_dom_children): Remove killed
1683 bitmap and df_simulate_find_defs call, instead walk the defs.
1684 Compute last_setter and initialize in_a_setter. Merge definitions
1685 with first initialization for a few variables.
1686 (try_validate_parallel): Use insn_invalid_p instead of
1687 recog_memoized. Return insn rather than just the pattern.
1688 (try_merge_compare): Fix up comment. Don't uselessly test if
1689 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1691 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1694 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1695 Alan Hayward <alan.hayward@arm.com>
1696 David Sherwood <david.sherwood@arm.com>
1698 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1699 aarch64_hard_regno_nregs to get the number of registers
1702 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1703 Alan Hayward <alan.hayward@arm.com>
1704 David Sherwood <david.sherwood@arm.com>
1706 * config/aarch64/constraints.md (Upl): Rename to...
1708 * config/aarch64/aarch64.md
1709 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1712 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1713 Alan Hayward <alan.hayward@arm.com>
1714 David Sherwood <david.sherwood@arm.com>
1716 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1717 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1720 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1721 Alan Hayward <alan.hayward@arm.com>
1722 David Sherwood <david.sherwood@arm.com>
1724 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1725 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1726 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1728 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1729 of a single element.
1730 * config/aarch64/iterators.md: Add a comment above the permute
1731 unspecs to say that they are generated directly by
1732 aarch64_expand_vec_perm_const.
1733 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1735 2017-11-01 Nathan Sidwell <nathan@acm.org>
1737 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1739 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1741 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1742 and medany code models, and describe what they do.
1744 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1746 Revert accidental duplicate:
1748 * combine.c (can_change_dest_mode): Reject changes in
1749 REGMODE_NATURAL_SIZE.
1751 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1753 PR rtl-optimization/64682
1754 PR rtl-optimization/69567
1755 PR rtl-optimization/69737
1756 PR rtl-optimization/82683
1757 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1758 register mentioned in the note, drop the note, unless it came from I3,
1759 in which case it should go to I3 again.
1761 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1763 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1764 and return false if not.
1765 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1767 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1769 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1770 range is known to be empty.
1772 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1773 Alan Hayward <alan.hayward@arm.com>
1774 David Sherwood <david.sherwood@arm.com>
1776 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1777 and CONST_VECTOR_NUNITS instead of computing the number of units from
1778 the byte sizes of the vector and element.
1779 (simplify_binary_operation_1): Likewise.
1780 (simplify_const_binary_operation): Likewise.
1781 (simplify_ternary_operation): Likewise.
1783 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1784 Alan Hayward <alan.hayward@arm.com>
1785 David Sherwood <david.sherwood@arm.com>
1787 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1788 (int_mem_offset): ...this new function.
1789 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1790 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1791 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1794 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1795 Alan Hayward <alan.hayward@arm.com>
1796 David Sherwood <david.sherwood@arm.com>
1798 * lower-subreg.c (interesting_mode_p): New function.
1799 (compute_costs, find_decomposable_subregs, decompose_register)
1800 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1801 (resolve_clobber, dump_choices): Use it.
1803 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1804 Alan Hayward <alan.hayward@arm.com>
1805 David Sherwood <david.sherwood@arm.com>
1807 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1809 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1810 Alan Hayward <alan.hayward@arm.com>
1811 David Sherwood <david.sherwood@arm.com>
1813 * alias.c (find_base_value, find_base_term): Only process integer
1814 truncations. Check the precision rather than the size.
1816 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1817 Alan Hayward <alan.hayward@arm.com>
1818 David Sherwood <david.sherwood@arm.com>
1820 * machmode.h (is_narrower_int_mode): New function
1821 * optabs.c (expand_float, expand_fix): Use it.
1822 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1824 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1825 Alan Hayward <alan.hayward@arm.com>
1826 David Sherwood <david.sherwood@arm.com>
1828 * rtl.h (narrower_subreg_mode): New function.
1829 * ira-color.c (update_costs_from_allocno): Use it.
1831 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1832 Alan Hayward <alan.hayward@arm.com>
1833 David Sherwood <david.sherwood@arm.com>
1835 * optabs-query.h (convert_optab_p): New function, split out from...
1836 (convert_optab_handler): ...here.
1837 (widening_optab_handler): Delete.
1838 (find_widening_optab_handler): Remove permit_non_widening parameter.
1839 (find_widening_optab_handler_and_mode): Likewise. Provide an
1840 override that operates on mode class wrappers.
1841 * optabs-query.c (widening_optab_handler): Delete.
1842 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1843 parameter. Assert that the two modes are the same class and that
1844 the "from" mode is narrower than the "to" mode. Use
1845 convert_optab_handler instead of widening_optab_handler.
1846 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1847 instead of widening_optab_handler.
1848 * expr.c (expand_expr_real_2): Update calls to
1849 find_widening_optab_handler.
1850 * optabs.c (expand_widen_pattern_expr): Likewise.
1851 (expand_binop_directly): Take the insn_code as a parameter.
1852 (expand_binop): Only call find_widening_optab_handler for
1853 conversion optabs; use optab_handler otherwise. Update calls
1854 to find_widening_optab_handler and expand_binop_directly.
1855 Use convert_optab_handler instead of widening_optab_handler.
1856 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1857 find_widening_optab_handler and use scalar_mode rather than
1859 (convert_plusminus_to_widen): Likewise.
1861 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1862 Alan Hayward <alan.hayward@arm.com>
1863 David Sherwood <david.sherwood@arm.com>
1865 * machmode.h (fixed_size_mode): New class.
1866 * rtl.h (get_pool_mode): Return fixed_size_mode.
1867 * gengtype.c (main): Add fixed_size_mode.
1868 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1869 (get_raw_arg_mode): Likewise.
1870 * doc/tm.texi: Regenerate.
1871 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1872 * targhooks.c (default_get_reg_raw_mode): Likewise.
1873 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1874 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1875 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1876 (msp430_get_raw_result_mode): Likewise.
1877 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1878 * dbxout.c (dbxout_parms): Require fixed-size modes.
1879 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1880 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1881 * omp-low.c (lower_oacc_reductions): Likewise.
1882 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1883 (simplify_subreg): Update accordingly.
1884 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1885 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1886 that aren't fixed-size.
1887 (get_pool_mode): Return a fixed_size_mode.
1888 (output_constant_pool_2): Take a fixed_size_mode.
1890 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1891 Alan Hayward <alan.hayward@arm.com>
1892 David Sherwood <david.sherwood@arm.com>
1894 * doc/rtl.texi (vec_series): Document.
1895 (const): Say that the operand can be a vec_series.
1896 * rtl.def (VEC_SERIES): New rtx code.
1897 * rtl.h (const_vec_series_p_1): Declare.
1898 (const_vec_series_p): New function.
1899 * emit-rtl.h (gen_const_vec_series): Declare.
1900 (gen_vec_series): Likewise.
1901 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1902 (gen_vec_series): Likewise.
1903 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1904 * simplify-rtx.c (simplify_unary_operation): Handle negations
1906 (simplify_binary_operation_series): New function.
1907 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1908 (test_vector_ops_series): New function.
1909 (test_vector_ops): Call it.
1910 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1911 gen_const_vec_series.
1912 (altivec_lvsr): Likewise.
1913 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1915 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1916 Alan Hayward <alan.hayward@arm.com>
1917 David Sherwood <david.sherwood@arm.com>
1919 * doc/rtl.texi (const): Update description of address constants.
1920 Say that vector constants are allowed too.
1921 * common.md (E, F): Use CONSTANT_P instead of checking for
1923 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1924 checking for CONST_VECTOR.
1925 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1927 * expr.c (expand_expr_real_2): Check for vector modes instead
1928 of checking for CONST_VECTOR.
1929 * rtl.h (const_vec_p): New function.
1930 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1931 (unwrap_const_vec_duplicate): Handle them here too.
1933 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1934 David Malcolm <dmalcolm@redhat.com>
1935 Alan Hayward <alan.hayward@arm.com>
1936 David Sherwood <david.sherwood@arm.com>
1938 * rtl.h (vec_duplicate_p): New function.
1939 * selftest-rtl.c (assert_rtx_eq_at): New function.
1940 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1941 (assert_rtx_eq_at): Declare.
1942 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1943 * selftest-run-tests.c (selftest::run_tests): Call it.
1944 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1945 (simplify_unary_operation_1): Recursively handle vector duplicates.
1946 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1948 (simplify_subreg): Handle subregs of vector duplicates.
1949 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1950 (selftest::simplify_rtx_c_tests): New functions.
1952 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1953 Alan Hayward <alan.hayward@arm.com>
1954 David Sherwood <david.sherwood@arm.com>
1956 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1957 (gen_vec_duplicate): Likewise.
1958 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1960 (gen_const_vector): ...here.
1961 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1962 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1963 whose elements are all equal.
1964 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1965 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1966 (simplify_relational_operation): Likewise.
1967 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1969 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1970 (aarch64_expand_vector_init): Likewise.
1971 * config/arm/arm.c (neon_vdup_constant): Likewise.
1972 (neon_expand_vector_init): Likewise.
1973 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1974 (arm_block_set_unaligned_vect): Likewise.
1975 (arm_block_set_aligned_vect): Likewise.
1976 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1977 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1978 (expand_vec_perm_even_odd_pack): Likewise.
1979 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1980 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1981 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1982 gen_const_vec_duplicate.
1983 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1984 * config/mips/mips.c (mips_gen_const_int_vector): Use
1985 gen_const_vec_duplicate.
1986 (mips_expand_vector_init): Use CONST0_RTX.
1987 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1988 (define_split): Use gen_const_vec_duplicate.
1989 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1990 (define_split): Use gen_const_vec_duplicate.
1991 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1992 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1993 * config/spu/spu.c (spu_const): Likewise.
1995 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1996 Alan Hayward <alan.hayward@arm.com>
1997 David Sherwood <david.sherwood@arm.com>
1999 * combine.c (can_change_dest_mode): Reject changes in
2000 REGMODE_NATURAL_SIZE.
2002 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2004 * configure.ac (--enable-libssp): New.
2005 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2006 trying to determine target-specific default. Adjust indentation.
2007 * configure: Regenerated.
2008 * doc/install.texi (Configuration): Expand --disable-libssp
2011 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2013 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2016 2017-10-31 Martin Jambor <mjambor@suse.cz>
2019 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2021 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2023 * auto-profile.c (autofdo_source_profile::read): Use
2024 UNKNOWN_LOCATION rather than 0.
2025 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2026 (warning_at): ...this overload.
2027 (warning_at_rich_loc_n): Rename to...
2028 (warning_n): ...this overload.
2029 (error_at_rich_loc): Rename to...
2030 (error_at): ...this overload.
2031 (pedwarn_at_rich_loc): Rename to...
2032 (pedwarn): ...this overload.
2033 (permerror_at_rich_loc): Rename to...
2034 (permerror): ...this overload.
2035 (inform_at_rich_loc): Rename to...
2036 (inform): ...this overload.
2037 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2038 (diagnostic_n_impl_richloc): Rename to...
2039 (diagnostic_n_impl): ...this rich_location *-based decl.
2040 (inform_at_rich_loc): Rename to...
2041 (inform): ...this, and add an assertion.
2042 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2043 (warning_at_rich_loc): Rename to...
2044 (warning_at): ...this, and add an assertion.
2045 (warning_at_rich_loc_n): Rename to...
2046 (warning_n): ...this, and add an assertion.
2047 (warning_n): Update location_t-based implementation for removal of
2048 location_t-based diagnostic_n_impl.
2049 (pedwarn_at_rich_loc): Rename to...
2050 (pedwarn): ...this, and add an assertion.
2051 (permerror_at_rich_loc): Rename to...
2052 (permerror): ...this, and add an assertion.
2053 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2054 (error_at_rich_loc): Rename to...
2055 (error_at): ...this, and add an assertion.
2056 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2057 (driver::do_spec_on_infiles): Likewise.
2058 * substring-locations.c (format_warning_va): Update for renaming
2059 of inform_at_rich_loc.
2061 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2063 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2064 _Float<N>X built-in functions so that the variant without the
2065 "__builtin_" prefix is only enabled for the GNU C and Objective C
2066 languages when they are in non-strict ANSI/ISO mode.
2067 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2068 * target.def (floatn_builtin_p): Add a target hook to control
2069 whether _Float<N> and _Float<N>X built-in functions without the
2070 "__builtin_" prefix are enabled, and return true for C and
2071 Objective C in the default hook. Include langhooks.h in
2073 * targhooks.h (default_floatn_builtin_p): Likewise.
2074 * targhooks.c (default_floatn_builtin_p): Likewise.
2075 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2076 floatn_builtin_p target hook.
2077 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2079 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2080 Eric Botcazou <ebotcazou@adacore.com>
2082 PR rtl-optimization/81803
2083 * lra-constraints.c (curr_insn_transform): Also reload the whole
2084 register for a strict subreg no wider than a word if this is for
2085 a WORD_REGISTER_OPERATIONS target.
2087 2017-10-31 Jason Merrill <jason@redhat.com>
2089 * gdbinit.in: Skip over inlines from timevar.h.
2091 2017-10-31 Martin Liska <mliska@suse.cz>
2093 * doc/gcov.texi: Document new option.
2094 * gcov.c (print_usage): Likewise print it.
2095 (process_args): Support the argument.
2096 (format_count): New function.
2097 (format_gcov): Use the function.
2099 2017-10-31 Martin Liska <mliska@suse.cz>
2101 * gcov.c (struct name_map): do not use typedef.
2102 Define operator== and operator<.
2103 (name_search): Remove.
2104 (name_sort): Remove.
2105 (main): Do not allocate names.
2106 (process_file): Add vertical space.
2107 (generate_results): Use std::find.
2108 (release_structures): Do not release memory.
2109 (find_source): Use std::find.
2111 2017-10-31 Martin Liska <mliska@suse.cz>
2113 * gcov.c (struct line_info): Remove it's typedef.
2114 (line_info::line_info): Add proper ctor.
2115 (line_info::has_block): Do not use a typedef.
2116 (struct source_info): Do not use typedef.
2117 (circuit): Likewise.
2118 (get_cycles_count): Likewise.
2119 (output_intermediate_file): Iterate via vector iterator.
2120 (add_line_counts): Use std::vector methods.
2121 (accumulate_line_counts): Likewise.
2122 (output_lines): Likewise.
2124 2017-10-31 Martin Liska <mliska@suse.cz>
2126 * gcov.c (struct source_info): Remove typedef.
2127 (source_info::source_info): Add proper ctor.
2128 (accumulate_line_counts): Use struct, not it's typedef.
2129 (output_gcov_file): Likewise.
2130 (output_lines): Likewise.
2131 (main): Do not allocate an array.
2132 (output_intermediate_file): Use size of vector container.
2133 (process_file): Resize the vector.
2134 (generate_results): Do not preallocate, use newly added vector
2136 (release_structures): Do not release sources.
2137 (find_source): Use vector methods.
2138 (add_line_counts): Do not use typedef.
2140 2017-10-31 Martin Liska <mliska@suse.cz>
2142 * doc/gcov.texi: Document that.
2143 * gcov.c (add_line_counts): Mark lines with a non-executed
2145 (output_line_beginning): Handle such lines.
2146 (output_lines): Pass new argument.
2147 (output_intermediate_file): Print it in intermediate format.
2149 2017-10-31 Martin Liska <mliska@suse.cz>
2151 * color-macros.h: New file.
2152 * diagnostic-color.c: Factor out color related to macros to
2154 * doc/gcov.texi: Document -k option.
2155 * gcov.c (INCLUDE_STRING): Include string.h.
2156 (print_usage): Add -k option.
2157 (process_args): Parse it.
2158 (pad_count_string): New function.
2159 (output_line_beginning): Likewise.
2160 (DEFAULT_LINE_START): New macro.
2161 (output_lines): Support color output.
2163 2017-10-31 Martin Liska <mliska@suse.cz>
2165 PR gcov-profile/82633
2166 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2167 their interaction with GCOV infrastructure.
2168 * configure.ac: Add -fkeep-{inline,static}-functions to
2170 * configure: Regenerate.
2172 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2175 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2177 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2180 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2181 into a register if it does not fit into an immediate offset field.
2183 2017-10-31 Olivier Hainque <hainque@adacore.com>
2185 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2187 2017-10-31 Julia Koval <julia.koval@intel.com>
2189 * config.gcc: Add gfniintrin.h.
2190 * config/i386/gfniintrin.h: New.
2191 * config/i386/i386-builtin-types.def
2192 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2193 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2194 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2195 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2196 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2197 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2198 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2199 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2200 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2201 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2202 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2203 * config/i386/immintrin.h: Include gfniintrin.h.
2204 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2206 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2208 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2210 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2212 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2213 (arm_ashldi3_1bit): Remove pattern.
2214 (ashrdi3): Remove shift by 1 expansion.
2215 (arm_ashrdi3_1bit): Remove pattern.
2216 (lshrdi3): Remove shift by 1 expansion.
2217 (arm_lshrdi3_1bit): Remove pattern.
2218 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2219 cost of ashldi3 by 1.
2220 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2221 (<shift>di3_neon): Likewise.
2223 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2225 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2226 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2227 and (*aarch64_simd_mov<VQ:mode>).
2228 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2229 pattern alternative.
2230 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2231 attributes to match pattern alternative.
2233 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2235 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2236 * config/rs6000/emmintrin.h: New file.
2237 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2239 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2241 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2242 * (movdi_vfp_cortexa8): Remove pattern.
2244 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2246 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2248 (Specific, alpha*-dec-osf5.1): Remove.
2249 (Specific, mips-sgi-irix5): Remove.
2250 (Specific, mips-sgi-irix6): Remove.
2252 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2255 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2256 arguments to clear_bit_region_be.
2258 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2260 * gimplify.c: Include memmodel.h.
2262 2017-10-30 Martin Jambor <mjambor@suse.cz>
2264 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2265 condition whether loop should be executed at all.
2267 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2269 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2270 gimple folding of vec_madd() intrinsics.
2271 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2272 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2273 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2275 2017-10-30 Richard Biener <rguenther@suse.de>
2277 PR tree-optimization/82762
2279 2017-10-23 Richard Biener <rguenther@suse.de>
2281 PR tree-optimization/82129
2283 2017-08-01 Richard Biener <rguenther@suse.de>
2285 PR tree-optimization/81181
2286 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2287 (compute_antic): ... end of iteration here.
2289 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2291 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2292 -std=iso9899:2017 and -std=gnu17.
2293 * doc/standards.texi (C Language): Document C17 support.
2294 * doc/cpp.texi (Overview): Mention -std=c17.
2295 (Standard Predefined Macros): Document C11 and C17 values of
2296 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2297 * doc/extend.texi (Inline): Do not list individual options for
2298 standards newer than C99.
2299 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2301 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2304 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2306 * asan.c (asan_finish_file): Align asan globals array by shadow
2309 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2312 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2313 (struct store_immediate_info): Add bitregion_start and bitregion_end
2315 (store_immediate_info::store_immediate_info): Add brs and bre
2316 arguments and initialize bitregion_{start,end} from those.
2317 (struct merged_store_group): Add bitregion_start, bitregion_end,
2318 align_base and mask fields. Drop unnecessary struct keyword from
2319 struct store_immediate_info. Add do_merge method.
2320 (clear_bit_region_be): Use memset instead of loop storing zeros.
2321 (merged_store_group::do_merge): New method.
2322 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2323 stores as long as the surrounding bitregions have no gaps.
2324 (merged_store_group::merge_overlapping): Use do_merge.
2325 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2326 is byte aligned, rather than requiring that start and width are
2327 byte aligned. Drop unnecessary struct keyword from
2328 struct store_immediate_info. Allocate and populate also mask array.
2329 Make start of the arrays relative to bitregion_start rather than
2330 start and size them according to bitregion_{end,start} difference.
2331 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2332 struct store_immediate_info.
2333 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2334 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2335 struct keyword from struct store_immediate_info.
2336 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2337 between stores as long as the surrounding bitregions have no gaps.
2339 (struct split_store): Add orig non-static data member.
2340 (split_store::split_store): Initialize orig to false.
2341 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2342 if there is exactly a single original stmt. Change stmts argument
2343 to pointer from reference, if NULL, don't push anything to it. Add
2344 first argument, use it to optimize skipping over orig stmts that
2345 are known to be before bitpos already. Simplify.
2346 (split_group): Return unsigned int count how many stores are or
2347 would be needed rather than a bool. Add allow_unaligned argument.
2348 Change split_stores argument from reference to pointer, if NULL,
2349 only do a dry run computing how many stores would be produced.
2350 Rewritten algorithm to use both alignment and misalign if
2351 !allow_unaligned and handle bitfield stores with gaps.
2352 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2353 from bitregion_start instead of start. Compute allow_unaligned
2354 here, if true, do 2 split_group dry runs to compute which one
2355 produces fewer stores and prefer aligned if equal. Punt if
2356 new count is bigger or equal than original before emitting any
2357 statements, rather than during that. Remove no longer needed
2358 new_ssa_names tracking. Replace num_stmts with
2359 split_stores.length (). Use 32-bit stack allocated entries
2360 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2361 if possible. Handle bitfields with gaps.
2362 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2363 Compute bitregion_{start,end} for the stores and construct
2364 store_immediate_info with that. Formatting fixes.
2366 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2369 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2370 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2372 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2374 * gimplify.c: Include tm_p.h.
2376 * common.opt (gcoff): Re-add as ignored option.
2377 (gcoff1, gcoff2, gcoff3): Likewise.
2379 * Makefile.in (OBJS): Delete sdbout.o.
2380 (GTFILES): Delete $(srcdir)/sdbout.c.
2381 * debug.h: Delete sdb_debug_hooks.
2382 * final.c: Delete sdbout.h include.
2383 (final_scan_insn): Delete SDB_DEBUG check.
2384 (rest_of_clean_state): Likewise.
2385 * output.h: Delete sdb_begin_function_line.
2388 * toplev.c: Delete sdbout.h include.
2389 (process_options): Delete SDB_DEBUG check.
2390 * tree-core.h (tree_type_common): Delete pointer field of
2392 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2393 TYPE_SYMTAB_POINTER.
2394 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2395 (TYPE_SYMTAB_IS_POINTER): Delete.
2396 (TYPE_SYMTAB_IS_DIE): Renumber.
2397 * xcoffout.c: Refer to former sdbout.c file.
2398 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2400 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2401 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2402 (Debugging Options): Delete -gcoff.
2403 (-gstabs): Delete SDB reference.
2405 (-gcoff@var{level}): Delete.
2406 * doc/passes.texi (Debugging information output): Delete SDB and
2407 sdbout.c references.
2408 * doc/tm.texi: Regenerate.
2409 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2410 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2412 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2413 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2414 (SDB_DEBUGGING_INFO): Delete.
2415 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2416 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2417 * target.def (output_source_filename): Delete COFF reference.
2419 * common.opt (gcoff): Delete.
2420 (gxcoff+): Update Negative chain.
2421 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2423 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2424 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2425 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2427 (expand_function_start): Change sdb reference to past tense.
2428 (expand_function_end): Change sdb reference to past tense.
2429 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2430 * opts.c (debug_type_names): Delete coff entry.
2431 (common_handle_option): Delete OPT_gcoff case.
2432 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2434 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2435 * config/cris/cris.h: Delete SDB reference in comment.
2436 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2437 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2438 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2439 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2441 (ix86_expand_prologue): Likewise.
2442 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2443 * config/ia64/ia64.h: Likewise.
2444 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2445 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2447 * config/mmix/mmix.h: Likewise.
2448 * config/nds32/nds32.c: Likewise.
2449 * config/stormy/storym16.h: Likewise.
2450 * config/visium/visium.h: Likewise.
2451 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2453 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2455 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2456 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2457 FRAME_POINTER_REGNUM point at high end of local var area.
2459 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2461 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2462 Move comment around. Do not reset best_edge for a copiable
2463 destination if the copy would cause a partition change.
2464 (better_edge_p): Remove redundant check.
2466 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2468 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2470 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2472 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2473 for math functions that have _Float<N> and _Float<N>X variants.
2474 (mathfn_built_in_2): Add support for math functions that have
2475 _Float<N> and _Float<N>X variants.
2476 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2477 (expand_builtin_mathfn_ternary): Add support for fma with
2478 _Float<N> and _Float<N>X variants.
2479 (expand_builtin): Likewise.
2480 (fold_builtin_3): Likewise.
2481 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2482 create math function _Float<N> and _Float<N>X variants as external
2484 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2485 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2486 the __builtin_ prefix and if not strict ansi, without the prefix.
2487 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2488 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2489 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2490 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2491 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2492 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2493 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2494 function signatures for fma _Float<N> and _Float<N>X variants.
2495 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2496 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2497 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2498 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2499 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2500 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2501 * gencfn-macros.c (print_case_cfn): Add support for math functions
2502 that have _Float<N> and _Float<N>X variants.
2503 (print_define_operator_list): Likewise.
2504 (fltfn_suffixes): Likewise.
2506 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2507 for math functions that have _Float<N> and _Float<N>X variants.
2508 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2509 and _Float<N>X variants.
2510 (COPYSIGN): Likewise.
2513 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2514 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2516 (integer_valued_read_call_p): Likewise.
2517 * fold-const-call.c (fold_const_call_ss): Likewise.
2518 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2519 _Float<N> and _Float<N>X variants.
2520 (fold_const_call_ssss): Add support for fma _Float<N> and
2521 _Float<N>X variants.
2522 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2523 support for copysign and fma _Float<N> and _Float<N>X variants.
2524 (backprop::process_builtin_call_use): Likewise.
2525 * tree-call-cdce.c (can_test_argument_range); Add support for
2526 sqrt _Float<N> and _Float<N>X variants.
2527 (edom_only_function): Likewise.
2528 (get_no_error_domain): Likewise.
2529 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2530 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2531 copysign _Float<N> and _Float<N>X variants.
2532 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2533 handled by machine independent code.
2534 (FMAF128): Likewise.
2535 * doc/cpp.texi (Common Predefined Macros): Document defining
2536 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2537 fma _Float<N> and _Float<N>X variants.
2539 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2542 * config/i386/i386-modes.def (CCFPU): Remove definition.
2543 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2544 (ix86_cc_modes_compatible): Ditto.
2545 (ix86_expand_carry_flag_compare): Ditto.
2546 (ix86_expand_int_movcc): Ditto.
2547 (ix86_expand_int_addcc): Ditto.
2548 (ix86_reverse_condition): Ditto.
2549 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2550 Return true/false for unordered/ordered fp comparisons.
2551 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2552 (ix86_prepare_fp_compare_args): Update for rename.
2553 (ix86_expand_fp_compare): Update for rename. Generate unordered
2554 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2555 (ix86_expand_sse_compare_and_jump): Ditto.
2556 * config/i386/predicates.md (fcmov_comparison_operator):
2557 Remove CCFPU mode handling.
2558 (ix86_comparison_operator): Ditto.
2559 (ix86_carry_flag_operator): Ditto.
2560 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2561 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2562 (*cmpu<mode>_cc_i387): Ditto.
2563 (FPCMP): Remove mode iterator.
2564 (unord): Remove mode attribute.
2565 (unord_subst): New define_subst transformation
2566 (unord): New define_subst attribute.
2568 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2569 (*cmpi<unord>xf_i387): Ditto.
2570 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2571 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2572 using unord_subst transformation.
2573 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2574 (round_saeonly): Also handle CCFP mode.
2575 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2576 Remove UNSPEC_SAHF unspec handling.
2578 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2580 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2582 2017-10-27 Jeff Law <law@redhat.com>
2584 * gimple-ssa-sprintf.c: Include domwalk.h.
2585 (class sprintf_dom_walker): New class, derived from dom_walker.
2586 (sprintf_dom_walker::before_dom_children): New function.
2587 (struct call_info): Moved into sprintf_dom_walker class
2588 (compute_formath_length, handle_gimple_call): Likewise.
2589 (sprintf_length::execute): Call the dominator walker rather
2590 than walking the statements.
2592 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2593 gimple statement locations.
2594 (check_array_bounds): Corresponding changes. Get the statement's
2595 location directly from wi->stmt.
2597 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2600 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2602 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2604 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2605 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2607 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2610 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2611 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2612 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2613 maybe_get_pool_constant.
2614 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2617 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2619 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2620 2.26 caveat. Update gas and gld versions.
2621 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2624 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2626 * cgraph.h (set_malloc_flag): Declare.
2627 * cgraph.c (set_malloc_flag_1): New function.
2628 (set_malloc_flag): Likewise.
2629 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2630 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2632 (read_ipa_call_summary): Add support for reading is_return_callee.
2633 (write_ipa_call_summary): Stream is_return_callee.
2634 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2635 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2636 ipa-prop.h, ipa-fnsummary.h.
2637 (pure_const_names): Change to static.
2638 (malloc_state_e): Define.
2639 (malloc_state_names): Define.
2640 (funct_state_d): Add field malloc_state.
2641 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2642 (check_retval_uses): New function.
2643 (malloc_candidate_p): Likewise.
2644 (analyze_function): Add support for malloc attribute.
2645 (pure_const_write_summary): Stream malloc_state.
2646 (pure_const_read_summary): Add support for reading malloc_state.
2647 (dump_malloc_lattice): New function.
2648 (propagate_malloc): New function.
2649 (warn_function_malloc): New function.
2650 (ipa_pure_const::execute): Call propagate_malloc and
2651 ipa_free_fn_summary.
2652 (pass_local_pure_const::execute): Add support for malloc attribute.
2653 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2654 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2656 2017-10-27 Martin Liska <mliska@suse.cz>
2658 PR gcov-profile/82457
2659 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2660 for fork-like functions to be properly instrumented.
2662 2017-10-27 Richard Biener <rguenther@suse.de>
2665 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2666 info when we redirected EH.
2668 2017-10-26 Michael Collison <michael.collison@arm.com>
2670 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2672 (<optab>_trunchf<GPI:mode>2: New pattern.
2673 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2674 * config/aarch64/iterators.md (wv): New mode attribute.
2675 (vf, VF): New mode attributes.
2676 (vgp, VGP): New mode attributes.
2677 (s): Update attribute with SImode and DImode prefixes.
2679 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2681 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2682 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2683 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2684 (nios2_option_overide): Initialize it. Don't allow R0-relative
2685 addressing with PIC.
2686 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2687 (nios2_symbolic_constant_p): Likewise.
2688 (nios2_legitimate_address_p): Likewise.
2689 (nios2_r0rel_section_name_p): New.
2690 (nios2_symbol_ref_in_r0rel_data_p): New.
2691 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2692 (r0rel_constant_p): New.
2693 (nios2_print_operand_address): Handle r0rel_constant_p.
2694 (nios2_cdx_narrow_form_p): Likewise.
2695 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2696 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2697 (Nios II Options): Document -mr0rel-sec.
2699 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2701 * config/nios2/nios2.c: Include xregex.h.
2702 (nios2_gprel_sec_regex): New.
2703 (nios2_option_overide): Initialize it. Don't allow GP-relative
2704 addressing with PIC.
2705 (nios2_small_section_name_p): Check for regex match.
2706 * config/nios2/nios2.opt (mgprel-sec=): New option.
2707 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2708 (Nios II Options): Document -mgprel-sec.
2710 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2712 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2714 2017-10-26 Tom de Vries <tom@codesourcery.com>
2716 PR tree-optimization/82707
2717 * gimple.c (gimple_copy): Fix unsharing of
2718 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2720 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2722 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2723 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2724 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2725 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2726 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2727 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2728 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2729 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2730 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2731 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2732 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2733 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2734 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2735 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2736 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2737 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2738 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2740 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2742 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2744 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2745 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2746 warning to rs6000.c. Remove the Undocumented flag, since it has
2748 (-mabi=ibmlongdouble): Likewise.
2749 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2750 already set, set the default format for long double.
2751 (rs6000_debug_reg_global): Print whether long double is IBM or
2753 (rs6000_option_override_internal): Rework setting long double
2754 format. Only warn if the user is changing the long double default
2755 and they did not use -Wno-psabi.
2756 * doc/invoke.texi (PowerPC options): Update the documentation for
2757 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2759 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2760 Alan Hayward <alan.hayward@arm.com>
2761 David Sherwood <david.sherwood@arm.com>
2763 * rtl.h (wider_subreg_mode): New function.
2764 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2765 rather than an unsigned int *.
2766 * ira-color.c (regno_max_ref_width): Replace with...
2767 (regno_max_ref_mode): ...this new variable.
2768 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2769 Use wider_subreg_mode.
2770 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2771 rather than an unsigned int *.
2772 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2773 (process_alt_operands): Likewise.
2774 (invariant_p): Likewise.
2775 * lra-spills.c (assign_mem_slot): Likewise.
2776 (add_pseudo_to_slot): Likewise.
2777 * lra.c (collect_non_operand_hard_regs): Likewise.
2778 (add_regs_to_insn_regno_info): Likewise.
2779 * reload1.c (regno_max_ref_width): Replace with...
2780 (regno_max_ref_mode): ...this new variable.
2781 (reload): Update accordingly. Update call to
2782 ira_sort_regnos_for_alter_reg.
2783 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2784 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2785 (scan_paradoxical_subregs): Likewise.
2787 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2789 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2790 (aarch64_frame): Add emit_frame_chain boolean.
2791 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2792 Move eh_return case to aarch64_layout_frame.
2793 (aarch64_layout_frame): Initialize emit_frame_chain.
2794 (aarch64_expand_prologue): Use emit_frame_chain.
2796 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2798 * config/aarch64/aarch64.c (aarch64_layout_frame):
2799 Ensure LR is always stored at the bottom of the callee-saves.
2800 Remove rarely used frame layout which saves callee-saves at top of
2801 frame, so the store of LR can be used as a valid probe in all cases.
2803 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2805 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2806 Improve unaligned TImode/TFmode base/offset split.
2808 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2809 Alan Hayward <alan.hayward@arm.com>
2810 David Sherwood <david.sherwood@arm.com>
2812 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2813 * combine.c (find_single_use_1): Likewise.
2814 (expand_field_assignment): Likewise.
2815 (move_deaths): Likewise.
2816 * lra-constraints.c (simplify_operand_subreg): Likewise.
2817 (curr_insn_transform): Likewise.
2818 * lra.c (collect_non_operand_hard_regs): Likewise.
2819 (add_regs_to_insn_regno_info): Likewise.
2820 * rtlanal.c (reg_referenced_p): Likewise.
2821 (covers_regno_no_parallel_p): Likewise.
2823 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2825 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2826 Don't print any bits outside the precision of the value.
2827 * wide-int.cc (test_printing): Add some new tests.
2829 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2831 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2832 supports -xbrace_comment option.
2833 * configure: Regenerate.
2834 * config.in: Regenerate.
2835 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2836 (ASM_CPU_SPEC): Use it.
2838 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2840 * target.def (static_rtx_alignment): New hook.
2841 * targhooks.h (default_static_rtx_alignment): Declare.
2842 * targhooks.c (default_static_rtx_alignment): New function.
2843 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2844 * doc/tm.texi: Regenerate.
2845 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2846 instead of targetm.constant_alignment. Remove call to
2848 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2849 (cris_preferred_mininum_alignment): New function, split out from...
2850 (cris_constant_alignment): ...here.
2851 (cris_static_rtx_alignment): New function.
2852 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2854 (ix86_constant_alignment): ...here.
2855 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2856 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2857 (mmix_static_rtx_alignment): New function.
2858 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2859 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2861 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2864 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2865 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2867 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2869 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2870 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2871 mark as a sign-extending load.
2872 (local_pic_load_u): Define.
2874 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2877 * fold-const.c (operand_equal_for_comparison_p): Also return true
2878 if ARG0 is a simple variant of ARG1 with narrower precision.
2879 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2881 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2883 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2885 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2886 scatter_static, scatter_per_elt.
2887 * x86-tune-costs.h: Add new cost entries.
2889 2017-10-25 Richard Biener <rguenther@suse.de>
2891 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2892 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2893 class pass_fre): Move to ...
2894 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2895 class pass_fre): ... here and adjust for statistics.
2897 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2900 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2901 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2902 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2903 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2906 2017-10-25 Richard Biener <rguenther@suse.de>
2908 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2909 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2910 eliminate_push_avail, eliminate_insert): Move inside...
2911 (class eliminate_dom_walker): ... this class in preparation
2913 (fini_eliminate): Remove by merging with ...
2914 (eliminate): ... this function. Adjust for class changes.
2915 (pass_pre::execute): Remove fini_eliminate call.
2916 (pass_fre::execute): Likewise.
2918 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2921 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2922 (VPERMI2, VPERMI2I): New mode iterators.
2923 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2924 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2926 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2927 mode iterator. Remove 3 old define_insn patterns.
2928 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2929 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2930 VPERMI2 mode iterator, remove the other two expanders.
2931 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2932 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2933 instructions, remove the other two patterns.
2934 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2935 mode iterator, remove the other two patterns.
2936 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2937 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2938 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2939 and adjust argument order accordingly.
2940 (ix86_expand_vec_perm): Adjust caller.
2941 (expand_vec_perm_1): Likewise.
2942 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2943 (expand_vec_perm_vpermt2_vpshub2): ... this.
2944 (ix86_expand_vec_perm_const_1): Adjust caller.
2945 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2948 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2949 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2950 (vec_shl_<mode>): Remove unused expander.
2951 (avx512bw_<shift_insn><mode>3): New define_insn.
2952 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2953 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2955 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2958 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2961 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2963 PR rtl-optimization/82396
2964 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2965 (autopref_multipass_init): Simplify initialization.
2966 (autopref_rank_data): Simplify sort order.
2967 * gcc/sched-int.h (autopref_multipass_data_): Remove
2968 multi_mem_insn_p, min_offset and max_offset.
2970 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2973 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2974 Check special value of flag_omit_frame_pointer.
2975 (aarch64_can_eliminate): Likewise.
2976 (aarch64_override_options_after_change_1): Simplify handling of
2977 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2979 2017-10-24 Richard Biener <rguenther@suse.de>
2981 PR tree-optimization/82697
2982 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2983 zero for conditional load and unconditional store.
2985 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2987 * doc/install.texi: Document bootstrap-cet.
2989 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2992 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2993 ENDBR instruction at function entrance if function is only
2996 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2999 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3000 patterns to better describe from which operation the CF is computed.
3001 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3002 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3003 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3004 is 0, use _0 suffixed expanders instead of emitting a comparison
3007 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3009 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3010 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3012 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3015 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3016 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3017 * loop-iv.c (iv_get_reaching_def): Likewise.
3018 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3019 variable is promoted and the partition contains undefined values.
3021 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3023 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3025 (nios2_address_cost): Define.
3026 (nios2_legitimize_address): Recognize (exp + constant) directly.
3027 (TARGET_ADDRESS_COST): Define.
3029 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3031 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3032 (nios2_symbolic_memory_operand_p): Declare.
3033 (nios2_split_large_constant): Declare.
3034 (nios2_split_symbolic_memory_operand): Declare.
3035 * config/nios2/nios2.c: Adjust includes.
3036 (nios2_symbolic_constant_allowed): New.
3037 (nios2_symbolic_constant_p): New.
3038 (nios2_plus_symbolic_constant_p): New.
3039 (nios2_valid_addr_expr_p): Recognize addresses involving
3041 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3042 (nios2_symbolic_memory_operand_p): New.
3043 (nios2_large_constant_p): New.
3044 (nios2_split_large_constant): New.
3045 (nios2_split_plus_large_constant): New.
3046 (nios2_split_symbolic_memory_operand): New.
3047 (nios2_legitimize_address): Code refactoring. Handle addresses
3048 involving symbolic constants.
3049 (nios2_emit_move_sequence): Likewise.
3050 (nios2_print_operand): Improve error output.
3051 (nios2_print_operand_address): Handle LO_SUM.
3052 (nios2_cdx_narrow_form_p): Likewise.
3053 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3054 operands involving symbolic constants.
3055 (movhi_internal, movsi_internal): Likewise.
3056 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3057 (extendhisi2, extendqi<mode>2): Likewise.
3059 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3061 * tree-pass.h (PROP_rtl_split_insns): Define.
3062 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3064 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3066 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3068 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3071 * target.def (const_not_ok_for_debug_p): Default to
3072 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3073 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3074 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3075 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3076 which targetm.const_not_ok_for_debug_p returned true.
3077 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3079 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3081 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3082 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3083 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3085 (ix86_const_not_ok_for_debug_p): New function.
3086 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3087 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3089 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3092 * system.h: Conditionally include "unique-ptr.h" if
3093 INCLUDE_UNIQUE_PTR is defined.
3094 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3095 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3097 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3099 * config/rl78/rl78.md: New define_expand "subdi3".
3101 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3104 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3105 DF_REF_INSN if DF_REF_INSN_INFO is false.
3107 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3109 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3110 xmm_move instead of sse_move.
3111 (sse_store_index): New function.
3112 (ix86_register_move_cost): Be more sensible about mismatch stall;
3113 model AVX moves correctly; make difference between sse->integer and
3115 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3116 moves; make difference between SSE and AVX.
3117 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3118 and zmm_move. Increase size of sse load and store tables;
3119 add unaligned load and store tables; add ssemmx_to_integer.
3120 * x86-tune-costs.h: Update all entries according to real
3121 move latencies from Agner Fog's manual and chip documentation.
3123 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3126 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3127 * config/i386/constraints.md (Wf): New constraint.
3128 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3129 (cmp<dwi>_doubleword): Removed.
3130 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3131 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3132 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3133 expand with cmp<dwi>_doubleword. For LTU and GEU use
3134 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3136 * common.opt (gcolumn-info): Enable by default.
3137 * doc/invoke.texi (gcolumn-info): Document new default.
3139 2017-10-23 Richard Biener <rguenther@suse.de>
3141 PR tree-optimization/82672
3142 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3143 Fold the stmt if we propagated into it.
3145 2017-10-23 Richard Biener <rguenther@suse.de>
3147 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3148 (bitmap_remove_expr_from_set): ... this. All callers call this
3149 for non-constant values.
3150 (bitmap_set_subtract): Rename to...
3151 (bitmap_set_subtract_expressions): ... this. Adjust and
3153 (bitmap_set_contains_value): Remove superfluous check.
3154 (bitmap_set_replace_value): Inline into single caller ...
3155 (bitmap_value_replace_in_set): ... here and simplify.
3156 (dependent_clean): Merge into ...
3157 (clean): ... this using an overload. Adjust.
3158 (prune_clobbered_mems): Adjust.
3159 (compute_antic_aux): Likewise.
3160 (compute_partial_antic_aux): Likewise.
3162 2017-10-23 Richard Biener <rguenther@suse.de>
3164 PR tree-optimization/82129
3166 2017-08-01 Richard Biener <rguenther@suse.de>
3168 PR tree-optimization/81181
3169 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3170 (compute_antic): ... end of iteration here.
3172 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3174 * target.def (starting_frame_offset): New hook.
3175 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3176 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3177 * doc/tm.texi.in: Regenerate.
3178 * hooks.h (hook_hwi_void_0): Declare.
3179 * hooks.c (hook_hwi_void_0): New function.
3180 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3181 STARTING_FRAME_OFFSET.
3182 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3183 * reload1.c (reload): Likewise.
3184 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3185 instead of STARTING_FRAME_OFFSET.
3186 * function.c (try_fit_stack_local): Likewise.
3187 (assign_stack_local_1): Likewise
3188 (instantiate_virtual_regs): Likewise.
3189 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3190 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3191 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3192 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3193 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3194 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3195 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3196 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3197 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3198 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3199 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3200 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3201 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3202 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3203 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3204 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3205 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3206 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3207 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3208 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3209 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3210 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3211 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3212 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3213 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3214 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3215 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3216 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3217 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3218 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3219 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3220 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3221 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3222 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3223 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3224 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3225 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3226 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3227 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3228 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3229 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3230 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3231 return a HOST_WIDE_INT.
3232 (avr_builtin_setjmp_frame_value): Use it instead of
3233 STARTING_FRAME_OFFSET.
3234 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3235 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3236 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3238 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3239 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3240 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3241 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3242 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3243 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3244 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3245 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3246 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3247 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3248 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3249 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3251 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3252 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3253 * config/mips/mips.c (mips_compute_frame_info): Refer to
3254 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3255 (mips_starting_frame_offset): New function.
3256 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3257 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3258 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3259 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3260 and return a HOST_WIDE_INT.
3261 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3262 (mmix_initial_elimination_offset): Refer to
3263 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3264 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3265 * config/pa/pa.c (pa_starting_frame_offset): New function.
3266 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3267 (pa_expand_prologue): Likewise.
3268 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3269 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3270 !FRAME_GROWS_DOWNWARD handling to...
3271 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3272 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3273 !FRAME_GROWS_DOWNWARD handling to...
3274 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3275 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3276 !FRAME_GROWS_DOWNWARD handling to...
3277 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3278 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3280 (rs6000_starting_frame_offset): New function.
3281 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3282 !FRAME_GROWS_DOWNWARD handling to...
3283 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3284 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3285 !FRAME_GROWS_DOWNWARD handling to...
3286 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3287 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3288 !FRAME_GROWS_DOWNWARD handling to...
3289 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3290 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3291 (rs6000_starting_frame_offset): New function.
3292 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3293 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3294 * config/vax/vax.c (vax_starting_frame_offset): New function.
3295 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3296 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3297 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3298 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3299 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3300 * system.h (STARTING_FRAME_OFFSET): Poison.
3302 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3304 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3305 SCALAR_TYPE_MODE instead of TYPE_MODE.
3307 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3308 Alan Hayward <alan.hayward@arm.com>
3309 David Sherwood <david.sherwood@arm.com>
3311 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3313 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3314 Alan Hayward <alan.hayward@arm.com>
3315 David Sherwood <david.sherwood@arm.com>
3317 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3319 2017-10-23 Richard Biener <rguenther@suse.de>
3321 PR tree-optimization/82129
3322 * tree-ssa-pre.c (bitmap_set_and): Remove.
3323 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3324 canonicalizing expressions in the set to those with lowest
3325 ID rather than taking that from the first edge.
3327 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3329 * combine.c (rtx_equal_for_field_assignment_p): Use
3330 byte_lowpart_offset.
3332 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3333 Alan Hayward <alan.hayward@arm.com>
3334 David Sherwood <david.sherwood@arm.com>
3336 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3337 to a SUBREG_PROMOTED_VAR.
3339 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3340 Alan Hayward <alan.hayward@arm.com>
3341 David Sherwood <david.sherwood@arm.com>
3343 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3344 (expand_debug_source_expr): Likewise.
3345 * combine.c (combine_simplify_rtx): Likewise.
3346 * cse.c (fold_rtx): Likewise.
3347 * optabs.c (expand_float): Likewise.
3348 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3349 (simplify_binary_operation_1): Likewise.
3351 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3352 Alan Hayward <alan.hayward@arm.com>
3353 David Sherwood <david.sherwood@arm.com>
3355 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3356 (record_promoted_value): Likewise.
3357 * expr.c (expand_expr_real_2): Likewise.
3358 * ree.c (update_reg_equal_equiv_notes): Likewise.
3359 (combine_set_extension): Likewise.
3360 * rtlanal.c (low_bitmask_len): Likewise.
3361 * simplify-rtx.c (neg_const_int): Likewise.
3362 (simplify_binary_operation_1): Likewise.
3364 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3365 Alan Hayward <alan.hayward@arm.com>
3366 David Sherwood <david.sherwood@arm.com>
3368 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3369 * regcprop.c (maybe_mode_change): Likewise.
3370 * reload1.c (alter_reg): Likewise.
3372 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3374 * inchash.h (inchash::hash::add_wide_int): New function.
3375 * lto-streamer-out.c (hash_tree): Use it.
3377 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3379 * inchash.h (inchash::hash::add_wide_int): Rename to...
3380 (inchash::hash::add_hwi): ...this.
3381 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3382 (polymorphic_call_target_hasher::hash): Likewise.
3383 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3384 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3385 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3386 * lto-streamer-out.c (hash_tree): Likewise.
3387 * optc-save-gen.awk: Likewise.
3388 * tree.c (add_expr): Likewise.
3390 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3393 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3394 for ordered inequality comparisons even with TARGET_IEEE_FP.
3396 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3399 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3400 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3401 Expand with cmp<dwi>_doubleword.
3403 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3405 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3406 List CET intrinsics.
3407 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3408 specific to -fcf-protection option.
3410 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3412 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3413 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3414 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3415 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3416 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3417 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3418 (extra_objs): Add cet.o for Linux/x86 targets.
3419 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3420 * config/i386/cet.c: New file.
3421 * config/i386/cetintrin.h: Likewise.
3422 * config/i386/t-cet: Likewise.
3423 * config/i386/cpuid.h (bit_SHSTK): New.
3424 (bit_IBT): Likewise.
3425 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3426 pass IBT and SHSTK bits.
3427 * config/i386/i386-builtin-types.def
3428 (VOID_FTYPE_UNSIGNED_PVOID): New.
3429 (VOID_FTYPE_UINT64_PVOID): Likewise.
3430 * config/i386/i386-builtin.def: Add CET intrinsics.
3431 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3432 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3433 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3434 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3436 * config/i386/i386.c (rest_of_insert_endbranch): New.
3437 (pass_data_insert_endbranch): Likewise.
3438 (pass_insert_endbranch): Likewise.
3439 (make_pass_insert_endbranch): Likewise.
3440 (ix86_notrack_prefixed_insn_p): Likewise.
3441 (ix86_target_string): Add -mibt, -mshstk flags.
3442 (ix86_option_override_internal): Add flag_cf_protection
3444 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3445 (ix86_print_operand): Add 'notrack' prefix output.
3446 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3447 (ix86_expand_builtin): Expand CET intrinsics.
3448 (x86_output_mi_thunk): Add 'endbranch' instruction.
3449 * config/i386/i386.h (TARGET_IBT): New.
3450 (TARGET_IBT_P): Likewise.
3451 (TARGET_SHSTK): Likewise.
3452 (TARGET_SHSTK_P): Likewise.
3453 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3454 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3455 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3456 (builtin_setjmp_setup): New pattern.
3457 (builtin_longjmp): Likewise.
3458 (rdssp<mode>): Likewise.
3459 (incssp<mode>): Likewise.
3460 (saveprevssp): Likewise.
3461 (rstorssp): Likewise.
3462 (wrss<mode>): Likewise.
3463 (wruss<mode>): Likewise.
3464 (setssbsy): Likewise.
3465 (clrssbsy): Likewise.
3466 (nop_endbr): Likewise.
3467 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3469 * config/i386/immintrin.h: Include <cetintrin.h>.
3470 * config/i386/linux-common.h
3471 (file_end_indicate_exec_stack_and_cet): New prototype.
3472 (TARGET_ASM_FILE_END): New.
3474 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3476 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3477 latencies instead of having separate table; make difference between
3478 integer and float costs.
3479 * i386.h (processor_costs): Remove scalar_stmt_cost,
3480 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3481 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3483 * x86-tune-costs.h: Remove entries which has been removed in
3484 procesor_costs from all tables; make cond_taken_branch_cost
3485 and cond_not_taken_branch_cost COST_N_INSNS based.
3487 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3489 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3491 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3493 * config/i386/i386.md (isa): Remove fma_avx512f.
3494 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3495 <avx512>_fmadd_<mode>_mask3<round_name>,
3496 <avx512>_fmsub_<mode>_mask<round_name>,
3497 <avx512>_fmsub_<mode>_mask3<round_name>,
3498 <avx512>_fnmadd_<mode>_mask<round_name>,
3499 <avx512>_fnmadd_<mode>_mask3<round_name>,
3500 <avx512>_fnmsub_<mode>_mask<round_name>,
3501 <avx512>_fnmsub_<mode>_mask3<round_name>,
3502 <avx512>_fmaddsub_<mode>_mask<round_name>,
3503 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3504 <avx512>_fmsubadd_<mode>_mask<round_name>,
3505 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3506 (*vec_widen_umult_even_v16si<mask_name>,
3507 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3508 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3510 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3512 * extend.texi: Add 'nocf_check' documentation.
3513 * gimple.texi: Add second parameter to
3514 gimple_build_call_from_tree.
3515 * invoke.texi: Add -fcf-protection documentation.
3516 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3518 2017-10-20 Richard Biener <rguenther@suse.de>
3520 PR tree-optimization/82473
3521 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3522 the largest input type.
3524 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3526 * c-attribs.c (handle_nocf_check_attribute): New function.
3527 (c_common_attribute_table): Add 'nocf_check' handling.
3528 * gimple-parser.c: Add second argument NULL to
3529 gimple_build_call_from_tree.
3530 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3531 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3533 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3534 * common.opt: Add fcf-protection flag.
3535 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3536 * flag-types.h: Add enum cf_protection_level.
3537 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3538 Add 'nocf_check' attribute propagation to gimple call.
3539 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3540 (gimple_build_call_from_tree): Update prototype.
3541 (gimple_call_nocf_check_p): New function.
3542 (gimple_call_set_nocf_check): Likewise.
3543 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3544 * ipa-icf.c: Add nocf_check attribute in statement hash.
3545 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3546 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3547 * toplev.c (process_options): Add flag_cf_protection handling.
3549 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3551 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3553 2017-10-20 Richard Biener <rguenther@suse.de>
3555 PR tree-optimization/82603
3556 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3557 remove false predicated stores.
3559 2017-10-20 Richard Biener <rguenther@suse.de>
3561 * graphite-isl-ast-to-gimple.c
3562 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3563 Remove return value and simplify, dump copied stmt after lhs
3565 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3566 Reduce dump verbosity.
3567 (gsi_insert_earliest): Likewise.
3568 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3569 * graphite.c (print_global_statistics): Adjust dumping.
3570 (print_graphite_scop_statistics): Likewise.
3571 (print_graphite_statistics): Do not dump loops here.
3572 (graphite_transform_loops): But here.
3574 2017-10-20 Nicolas Roche <roche@adacore.com>
3576 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3577 * configure: Regenerate.
3579 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3582 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3583 functions when optimizing replace GIMPLE_RETURN stmts with
3584 calls to __builtin_unreachable ().
3587 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3588 for -fsanitize=thread link of executables.
3589 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3590 link of executables.
3593 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3595 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3596 define_insns for logical vector shifts to use VI248_AVX512BW
3597 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3598 condition, useless isa and prefix attributes. Change the first
3599 2 of these define_insns to ...
3600 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3601 define_insn for avx512vl.
3602 (<shift_insn><mode>3): ... and this, new define_insn without
3603 masking for non-avx512vl.
3606 * config/i386/sse.md (*andnot<mode>3,
3607 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3608 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3609 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3610 not applied use empty suffix even for TARGET_AVX512VL.
3611 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3612 is applied, supply evex,evex or evex,evex,evex instead of just
3615 2017-10-20 Julia Koval <julia.koval@intel.com>
3617 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3618 (OPTION_MASK_ISA_GFNI_UNSET): New.
3619 (ix86_handle_option): Handle OPT_mgfni.
3620 * config/i386/cpuid.h (bit_GFNI): New.
3621 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3622 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3623 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3624 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3625 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3626 * config/i386/i386.opt: Add mgfni.
3628 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3630 * config/msp430/msp430.c (msp430_option_override): Disable
3631 -fdelete-null-pointer-checks.
3632 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3634 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3636 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3637 of x87 and SSE instructions.
3639 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3641 * asan.c (create_cond_insert_point): Do not update edge count.
3642 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3643 (afdo_propagate_circuit): Likewise.
3644 (afdo_calculate_branch_prob): Likewise.
3645 (afdo_annotate_cfg): Likewise.
3646 * basic-block.h (struct edge_def): Remove count.
3647 (edge_def::count): New accessor.
3648 * bb-reorder.c (rotate_loop): Update.
3649 (find_traces_1_round): Update.
3650 (connect_traces): Update.
3651 (sanitize_hot_paths): Update.
3652 * cfg.c (unchecked_make_edge): Update.
3653 (make_single_succ_edge): Update.
3654 (check_bb_profile): Update.
3655 (dump_edge_info): Update.
3656 (update_bb_profile_for_threading): Update.
3657 (scale_bbs_frequencies_int): Update.
3658 (scale_bbs_frequencies_gcov_type): Update.
3659 (scale_bbs_frequencies_profile_count): Update.
3660 (scale_bbs_frequencies): Update.
3661 * cfganal.c (connect_infinite_loops_to_exit): Update.
3662 * cfgbuild.c (compute_outgoing_frequencies): Update.
3663 (find_many_sub_basic_blocks): Update.
3664 * cfgcleanup.c (try_forward_edges): Update.
3665 (try_crossjump_to_edge): Update
3666 * cfgexpand.c (expand_gimple_cond): Update
3667 (expand_gimple_tailcall): Update
3668 (construct_exit_block): Update
3669 * cfghooks.c (verify_flow_info): Update
3670 (redirect_edge_succ_nodup): Update
3671 (split_edge): Update
3672 (make_forwarder_block): Update
3673 (duplicate_block): Update
3674 (account_profile_record): Update
3675 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3676 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3677 * cfgloopmanip.c (scale_loop_profile): Update.
3679 (lv_adjust_loop_entry_edge): Update.
3680 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3681 (force_nonfallthru_and_redirect): Update.
3682 (purge_dead_edges): Update.
3683 (rtl_flow_call_edges_add): Update.
3684 * cgraphunit.c (init_lowered_empty_function): Update.
3685 (cgraph_node::expand_thunk): Update.
3686 * gimple-pretty-print.c (dump_probability): Update.
3687 (dump_edge_probability): Update.
3688 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3689 * haifa-sched.c (sched_create_recovery_edges): Update.
3690 * hsa-gen.c (convert_switch_statements): Update.
3691 * ifcvt.c (dead_or_predicable): Update.
3692 * ipa-inline-transform.c (inline_transform): Update.
3693 * ipa-split.c (split_function): Update.
3694 * ipa-utils.c (ipa_merge_profiles): Update.
3695 * loop-doloop.c (add_test): Update.
3696 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3697 * lto-streamer-in.c (input_cfg): Update.
3698 (input_function): Update.
3699 * lto-streamer-out.c (output_cfg): Update.
3700 * modulo-sched.c (sms_schedule): Update.
3701 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3702 * predict.c (maybe_hot_edge_p): Update.
3703 (unlikely_executed_edge_p): Update.
3704 (probably_never_executed_edge_p): Update.
3705 (dump_prediction): Update.
3706 (drop_profile): Update.
3707 (propagate_unlikely_bbs_forward): Update.
3708 (determine_unlikely_bbs): Update.
3709 (force_edge_cold): Update.
3710 * profile.c (compute_branch_probabilities): Update.
3711 * reg-stack.c (better_edge): Update.
3712 * shrink-wrap.c (handle_simple_exit): Update.
3713 * tracer.c (better_p): Update.
3714 * trans-mem.c (expand_transaction): Update.
3715 (split_bb_make_tm_edge): Update.
3716 * tree-call-cdce.c: Update.
3717 * tree-cfg.c (gimple_find_sub_bbs): Update.
3718 (gimple_split_edge): Update.
3719 (gimple_duplicate_sese_region): Update.
3720 (gimple_duplicate_sese_tail): Update.
3721 (gimple_flow_call_edges_add): Update.
3722 (insert_cond_bb): Update.
3723 (execute_fixup_cfg): Update.
3724 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3725 * tree-complex.c (expand_complex_div_wide): Update.
3726 * tree-eh.c (lower_resx): Update.
3727 (unsplit_eh): Update.
3728 (cleanup_empty_eh_move_lp): Update.
3729 * tree-inline.c (copy_edges_for_bb): Update.
3730 (freqs_to_counts): Update.
3731 (copy_cfg_body): Update.
3732 * tree-ssa-dce.c (remove_dead_stmt): Update.
3733 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3734 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3735 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3736 (unloop_loops): Update.
3737 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3738 * tree-ssa-loop-split.c (connect_loops): Update.
3739 (split_loop): Update.
3740 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3741 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3742 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3743 * tree-ssa-reassoc.c (branch_fixup): Update.
3744 * tree-ssa-tail-merge.c (replace_block_by): Update.
3745 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3746 (compute_path_counts): Update.
3747 (update_profile): Update.
3748 (recompute_probabilities): Update.
3749 (update_joiner_offpath_counts): Update.
3750 (estimated_freqs_path): Update.
3751 (freqs_to_counts_path): Update.
3752 (clear_counts_path): Update.
3753 (ssa_fix_duplicate_block_edges): Update.
3754 (duplicate_thread_path): Update.
3755 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3756 (case_bit_test_cmp): Update.
3757 (collect_switch_conv_info): Update.
3758 (gen_inbound_check): Update.
3759 (do_jump_if_equal): Update.
3760 (emit_cmp_and_jump_insns): Update.
3761 * tree-tailcall.c (decrease_profile): Update.
3762 (eliminate_tail_call): Update.
3763 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3764 (vect_do_peeling): Update.
3765 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3766 * ubsan.c (ubsan_expand_null_ifn): Update.
3767 (ubsan_expand_ptr_ifn): Update.
3768 * value-prof.c (gimple_divmod_fixed_value): Update.
3769 (gimple_mod_pow2): Update.
3770 (gimple_mod_subtract): Update.
3771 (gimple_ic): Update.
3772 (gimple_stringop_fixed_value): Update.
3774 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3777 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3779 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3781 PR rtl-optimization/82395
3782 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3783 based on non_spilled_static_chain_regno_p.
3785 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3787 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3788 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3789 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3791 2017-10-19 Martin Sebor <msebor@redhat.com>
3793 PR tree-optimization/82596
3794 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3796 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3798 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3799 (get_mem_refs_of_builtin_call): Likewise.
3800 * builtins.c (expand_builtin_apply): Adjust call to
3801 allocate_dynamic_stack_space.
3802 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3803 the third argument to allocate_dynamic_stack_space, otherwise -1.
3804 (expand_builtin): Deal with all alloca variants.
3805 (is_inexpensive_builtin): Likewise.
3806 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3807 * calls.c (special_function_p): Deal with all alloca variants.
3808 (initialize_argument_information): Adjust call to
3809 allocate_dynamic_stack_space.
3810 (expand_call): Likewise.
3811 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3812 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3813 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3814 use it for the stack usage computation.
3815 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3816 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3817 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3818 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3819 (in_loop_p): Remove first argument and useless check.
3820 (pass_walloca::execute): Remove useless test and adjust call to above.
3821 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3822 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3823 (gimplify_call_expr): Deal with all alloca variants.
3824 * hsa-gen.c (gen_hsa_alloca): Likewise.
3825 (gen_hsa_insns_for_call): Likewise.
3826 * ipa-pure-const.c (special_builtin_state): Likewise.
3827 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3828 * tree-object-size.c (alloc_object_size): Likewise.
3829 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3830 (call_may_clobber_ref_p_1): Likewise.
3831 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3832 (ccp_fold_stmt): Likewise.
3833 (optimize_stack_restore): Likewise.
3834 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3835 (mark_all_reaching_defs_necessary_1): Likewise.
3836 (propagate_necessity): Likewise.
3837 (eliminate_unnecessary_stmts): Likewise.
3838 * tree.c (build_common_builtin_nodes): Build
3839 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3840 (build_alloca_call_expr): New function.
3841 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3842 (CASE_BUILT_IN_ALLOCA): Likewise.
3843 (build_alloca_call_expr): Declare.
3844 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3846 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3849 * dwarf2out.c (new_die_raw): New static inline function.
3850 (new_die): Use it to create the DIE.
3851 (add_AT_external_die_ref): Likewise.
3852 (clone_die): Likewise.
3853 (clone_as_declaration): Likewise.
3854 (dwarf2out_vms_debug_main_pointer): Likewise.
3855 (base_type_die): Likewise. Remove early return for corner cases.
3856 Do not call add_pubtype on the DIE here.
3857 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3858 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3859 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3860 native order exists for base types, attach the DIE manually and call
3861 add_pubtype on it. Do not equate a reverse order DIE to the type.
3863 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3865 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3866 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3868 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3870 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3872 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3873 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3874 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3875 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3876 * opts.c (sanitizer_opts): Add builtin.
3877 * ubsan.c (instrument_builtin): New function.
3878 (pass_ubsan::execute): Call it.
3879 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3880 * doc/invoke.texi: Document -fsanitize=builtin.
3882 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3883 builtins, store max (log2 (align), 0) into uchar field instead of
3884 align into uptr field.
3885 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3886 store uchar 0 field instead of uptr 0 field.
3887 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3888 instead of passing one address of struct with 2 locations pass
3889 two addresses of structs with 1 location each.
3890 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3891 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3892 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3893 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3894 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3895 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3896 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3897 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3899 2017-10-19 Martin Liska <mliska@suse.cz>
3902 * file-find.c (remove_prefix): Remove.
3903 * file-find.h (remove_prefix): Likewise.
3904 * gcc-ar.c: Remove smartness of lookup.
3906 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3908 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3909 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3910 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3912 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3915 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3916 (setcc + and to xor + setcc): New peephole2.
3918 2017-10-19 Tom de Vries <tom@codesourcery.com>
3920 * doc/sourcebuild.texi (Test Directives, Variants of
3921 dg-require-support): Add dg-require-stack-size.
3923 2017-10-19 Martin Liska <mliska@suse.cz>
3926 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3927 that have a large alignment.
3928 (gimplify_target_expr): Likewise.
3930 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3932 PR rtl-optimization/82602
3933 * ira.c (rtx_moveable_p): Return false for volatile asm.
3935 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3938 * config/i386/i386-modes.def (CCGZ): New CC mode.
3939 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3940 * config/i386/predicates.md (ix86_comparison_operator):
3942 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3943 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3944 with double-word subtraction.
3945 (put_condition_code): Handle CCGZmode.
3947 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3949 * wide-int.cc (debug (const wide_int &)): New.
3950 (debug (const wide_int *)): New.
3951 (debug (const widest_int &)): New.
3952 (debug (const widest_int *)): New.
3954 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3957 * lra-constraints.c (curr_insn_transform): Use non-input operand
3958 instead of output one for matched reload.
3960 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3962 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3963 (tree-ssa-loop-ivopts.h): New header file.
3964 (struct builtin_info): New fields.
3965 (classify_builtin_1): Compute and record base and offset parts for
3966 memset builtin partition by calling strip_offset.
3967 (offset_cmp, fuse_memset_builtins): New functions.
3968 (finalize_partitions): Fuse adjacent memset partitions by calling
3970 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3971 Expose the interface.
3972 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3974 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3976 PR tree-optimization/82574
3977 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3978 that data reference must be executed exactly once per iteration
3979 against the outermost loop in nest.
3980 (classify_partition): Update call to above function.
3982 2017-10-18 Richard Biener <rguenther@suse.de>
3984 PR tree-optimization/82591
3985 * graphite.c (graphite_transform_loops): Move code gen message
3987 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3988 Here. Handle scop_to_isl_ast failing.
3989 (scop_to_isl_ast): Limit the number of ISL operations.
3991 2017-10-18 Richard Biener <rguenther@suse.de>
3993 * graphite-isl-ast-to-gimple.c
3994 (translate_isl_ast_to_gimple::set_rename): Simplify.
3995 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3996 (graphite_copy_stmts_from_block): ... here.
3997 (copy_bb_and_scalar_dependences): Simplify.
3998 (add_parameters_to_ivs_params): Canonicalize.
3999 (generate_entry_out_of_ssa_copies): Simplify.
4000 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4001 by passing in ISL dimension.
4002 (parameter_index_in_region_1): Rename to ...
4003 (parameter_index_in_region): ... this.
4004 (extract_affine): Adjust assert, pass down parameter index.
4005 (add_param_constraints): Use range-info when available.
4006 (build_scop_context): Adjust.
4007 * sese.c (new_sese_info): Adjust.
4008 (free_sese_info): Likewise.
4009 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4010 Remove unused typedefs.
4011 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4013 2017-10-18 Martin Liska <mliska@suse.cz>
4015 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4017 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4019 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4020 (s390_sched_init): Do not reset s390_sched_state if we entered the
4021 current basic block via a fallthru edge and all others are unlikely.
4023 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4025 * config/s390/s390.c (NUM_SIDES): New variable.
4026 (LONGRUNNING_THRESHOLD): New variable.
4027 (LATENCY_FACTOR): New variable.
4028 (s390_sched_score): Decrease score for long-running instructions on
4030 (s390_sched_variable_issue): Perform bookkeeping for long-running
4033 2017-10-18 Richard Biener <rguenther@suse.de>
4035 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4036 Simplify with removal of the parameter rename map.
4037 (set_rename): Likewise.
4038 (should_copy_to_new_region): Likewise.
4039 (graphite_copy_stmts_from_block): Likewise.
4040 (copy_bb_and_scalar_dependences): Remove initialization of
4041 unused copied_bb_map.
4043 (copy_internal_parameters): Likewise.
4044 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4045 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4046 Use INTEGRAL_TYPE_P.
4047 (parameter_index_in_region_1): Rename to ...
4048 (assign_parameter_index_in_region): ... this. Assert we have
4049 a parameter we handle.
4050 (scan_tree_for_params): Adjust.
4051 * sese.h (parameter_rename_map_t): Remove.
4052 (struct sese_info_t): Remove unused parameter_rename_map and
4053 copied_bb_map members.
4054 * sese.c (new_sese_info): Adjust.
4055 (free_sese_info): Likewise.
4057 2017-10-18 Martin Liska <mliska@suse.cz>
4060 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4061 on an abnormal edge.
4063 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4065 * doc/invoke.texi (ffunction-sections and fdata-sections):
4068 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4070 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4071 the use statement can throw internally.
4073 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4075 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4076 any RTX present on the RHS of a SET.
4077 * compare-elim.c (try_eliminate_compare): Restore comment.
4079 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4081 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4082 may be also called on tcc_type nodes.
4083 * langhooks.c (lhd_tree_size): Likewise.
4085 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4087 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4088 format_warning_at_substring.
4089 (maybe_warn): Convert source_range * param to a location_t. Pass
4090 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4091 (format_directive): Remove code to extract source_ranges and
4092 source_range * in favor of just a location_t.
4093 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4095 * substring-locations.c (format_warning_va): Convert
4096 source_range * param to a location_t.
4097 (format_warning_at_substring): Likewise.
4098 * substring-locations.h (format_warning_va): Likewise.
4099 (format_warning_at_substring): Likewise.
4101 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4103 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4105 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4106 and scatter/gather ops.
4108 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4109 vec_gather_load and vec_scatter_store.
4110 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4111 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4112 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4113 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4114 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4115 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4117 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4119 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4120 Detect FTST instruction and handle its register pops. Only pop
4121 second operand if can_pop_second_op is true.
4122 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4123 set can_pop_second_op to false in the compare_for_stack_reg call.
4125 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4126 output_fp_compare for stack register operands.
4127 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4128 instructions here. Do not emit stack register pops here. Assert
4129 that FCOMPP pops next to top stack register. Rewrite function.
4131 2017-10-17 Nathan Sidwell <nathan@acm.org>
4134 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4135 use DECL_ASSEMBLER_NAME_RAW.
4138 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4141 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4142 Wilco Dijkstra <wilco.dijkstra@arm.com>
4144 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4145 converstion to Pmode from the buf_addr.
4147 2017-10-17 Richard Biener <rguenther@suse.de>
4149 * graphite-dependences.c (scop_get_reads_and_writes): Change
4150 output parameters to references.
4152 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4154 PR 71026/tree-optimization
4155 * fold-const.c (distribute_real_division): Removed.
4156 (fold_binary_loc): Remove calls to distribute_real_divison.
4158 2017-10-17 Richard Biener <rguenther@suse.de>
4160 * graphite-scop-detection.c
4161 (scop_detection::stmt_has_simple_data_refs_p): Always use
4162 the full nest as region.
4163 (try_generate_gimple_bb): Likewise.
4164 * sese.c (scalar_evolution_in_region): Simplify now that
4165 SCEV can handle instantiation in regions.
4166 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4167 in the non-loop part of a function if requested.
4169 2017-10-17 Richard Biener <rguenther@suse.de>
4171 PR tree-optimization/82563
4172 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4174 (graphite_regenerate_ast_isl): Call it.
4175 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4177 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4179 PR tree-optimization/82549
4180 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4181 Formatting fixes. Instead of calling make_bit_field_ref with negative
4184 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4186 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4187 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4188 _mm_maskz_reduce_ss): New.
4189 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4190 __builtin_ia32_reducess_mask): Ditto..
4191 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4192 * config/i386/sse.md (reduces<mode>): Renamed to ...
4193 (reduces<mode><mask_scalar_name>): ... this.
4194 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4196 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4197 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4199 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4201 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4202 * selftest-run-tests.c (selftest::run_tests): Call
4203 selftest::unique_ptr_tests_cc_tests.
4204 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4205 * unique-ptr-tests.cc: New file.
4207 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4210 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4212 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4213 (make_hard_regno_born, make_hard_regno_dead): Update
4214 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4216 2017-10-16 Jeff Law <law@redhat.com>
4218 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4220 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4222 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4224 2017-10-16 Olivier Hainque <hainque@adacore.com>
4226 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4227 with_cpu if we were configured for an e500v2 target cpu name.
4229 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4231 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4232 * doc/invoke.texi: Document +nodsp as a valid extension for
4235 2017-10-16 Martin Liska <mliska@suse.cz>
4237 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4238 (test_set_range): Likewise.
4239 (test_range_functions): Rename to ...
4240 (test_bit_in_range): ... this.
4241 (sbitmap_c_tests): Add new test.
4243 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4245 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4247 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4248 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4250 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4252 * config/aarch64/aarch64-builtins.c
4253 (aarch64_types_quadopu_lane_qualifiers): New.
4254 (TYPES_QUADOPU_LANE): New.
4255 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4256 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4257 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4258 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4259 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4260 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4261 (Vdottype, DOTPROD): New.
4262 (sur): Add SDOT and UDOT.
4264 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4266 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4267 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4268 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4270 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4271 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4272 Enable TARGET_DOTPROD.
4273 (cortex-a75.cortex-a55): Likewise.
4274 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4276 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4278 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4279 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4281 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4283 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4284 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4285 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4286 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4287 * config/arm/types.md (neon_dot, neon_dot_q): New.
4288 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4290 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4292 * config/arm/arm.h (TARGET_DOTPROD): New.
4293 * config/arm/arm.c (arm_arch_dotprod): New.
4294 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4295 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4296 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4297 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4298 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4299 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4300 * doc/invoke.texi (armv8.2-a): Document dotprod
4302 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4304 * i386.c (ix86_vec_cost): New function.
4305 (ix86_rtx_costs): Handle vector operations better.
4306 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4307 * x86-tune-costs.h: Add new costs to all tables.
4309 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4311 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4313 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4314 divsd, sqrtss and sqrtsd
4315 * x86-tune-costs.h: Add new entries to all costs.
4316 (znver1_cost): Fix to match real instruction latencies.
4318 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4319 Michael Collison <michael.collison@arm.com>
4321 * compare-elim.c: Include emit-rtl.h.
4322 (can_merge_compare_into_arith): New function.
4323 (try_validate_parallel): Likewise.
4324 (try_merge_compare): Likewise.
4325 (try_eliminate_compare): Call the above when no previous clobber
4327 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4330 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4334 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4335 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4339 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4340 to be any operand_equal_p operands. For & (B - 1) require
4341 B to be power of 2. Recognize
4342 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4344 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4347 * optabs.c (expand_memory_blockage): Fix call of
4348 targetm.have_memory_blockage.
4350 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4353 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4354 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4355 objects to extra_objs instead of overwriting it.
4357 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4359 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4360 Use any_fp_register_operand as operand[3] predicate. Simplify
4361 equality test for operands[2] and operands[4] memory location.
4362 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4363 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4364 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4365 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4366 any_fp_register_operand as operand[1] predicate. Simplify
4367 equality test for operands[0] and operands[3] memory location.
4368 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4369 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4370 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4372 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4374 * target-insns.def: Add memory_blockage.
4375 * optabs.c (expand_memory_blockage): New function.
4376 (expand_asm_memory_barrier): Rename ...
4377 (expand_asm_memory_blockage): ... to this.
4378 (expand_mem_thread_fence): Call expand_memory_blockage
4379 instead of expand_asm_memory_barrier.
4380 (expand_mem_singnal_fence): Ditto.
4381 (expand_atomic_load): Ditto.
4382 (expand_atomic_store): Ditto.
4383 * doc/md.texi (Standard Pattern Names For Generation):
4384 Document memory_blockage instruction pattern.
4386 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4388 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4389 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4390 * config/rl78/rl78.md: New define_expand "adddi3".
4392 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4394 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4397 2017-10-13 Jeff Law <law@redhat.com>
4399 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4401 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4404 * internal-fn.c (expand_mul_overflow): If both operands have
4405 the same highpart of -1 or 0 and the topmost bit of lowpart
4406 is different, overflow is if res <= 0 rather than res < 0.
4408 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4410 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4411 TARGET_P9_VECTOR code for unaligned_load case.
4413 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4415 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4417 2017-10-13 Nathan Sidwell <nathan@acm.org>
4419 * tree-core.h (tree_contains_struct): Make bool.
4420 * tree.c (tree_contains_struct): Likewise.
4421 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4422 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4423 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4424 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4425 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4427 2017-10-13 Richard Biener <rguenther@suse.de>
4429 * graphite-isl-ast-to-gimple.c
4430 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4431 parameters and dominance check.
4432 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4433 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4434 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4435 Do not update SSA form here or do intermediate IL verification.
4436 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4437 (graphite_initialize): Remove check on the number of loops in
4438 the function and inline into graphite_transform_loops.
4439 (graphite_finalize): Inline into graphite_transform_loops.
4440 (graphite_transform_loops): Perform SSA update and IL verification
4442 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4444 2017-10-13 Richard Biener <rguenther@suse.de>
4446 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4447 graphite_expression_type_precision): Avoid global constructor
4449 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4450 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4451 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4452 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4453 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4454 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4456 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4459 * config/i386/i386.h (ix86_red_zone_size): New.
4460 * config/i386/i386.md (push peephole2s): Replace
4461 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4463 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4464 Alan Hayward <alan.hayward@arm.com>
4465 David Sherwood <david.sherwood@arm.com>
4467 * combine.c (can_change_dest_mode): Reject changes in
4468 REGMODE_NATURAL_SIZE.
4470 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4471 Alan Hayward <alan.hayward@arm.com>
4472 David Sherwood <david.sherwood@arm.com>
4474 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4475 (expand_debug_source_expr): Likewise.
4476 * combine.c (combine_simplify_rtx): Likewise.
4477 * cse.c (fold_rtx): Likewise.
4478 * fwprop.c (canonicalize_address): Likewise.
4479 * targhooks.c (default_shift_truncation_mask): Likewise.
4481 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4482 Alan Hayward <alan.hayward@arm.com>
4483 David Sherwood <david.sherwood@arm.com>
4485 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4486 (widened_mode): Likewise.
4487 (expand_unop): Likewise.
4488 * ree.c (transform_ifelse): Likewise.
4489 (merge_def_and_ext): Likewise.
4490 (combine_reaching_defs): Likewise.
4491 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4493 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4494 Alan Hayward <alan.hayward@arm.com>
4495 David Sherwood <david.sherwood@arm.com>
4497 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4498 * combine.c (gen_lowpart_for_combine): Likewise.
4499 * dwarf2out.c (rtl_for_decl_location): Likewise.
4500 * final.c (alter_subreg): Likewise.
4501 * rtlhooks.c (gen_lowpart_general): Likewise.
4502 (gen_lowpart_if_possible): Likewise.
4504 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4505 Alan Hayward <alan.hayward@arm.com>
4506 David Sherwood <david.sherwood@arm.com>
4508 * calls.c (expand_call): Use subreg_lowpart_offset.
4509 * cse.c (cse_insn): Likewise.
4510 * regcprop.c (copy_value): Likewise.
4511 (copyprop_hardreg_forward_1): Likewise.
4513 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4516 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4517 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4518 =Q constraints to +Q and into insn condition add check
4519 that operands[0] and operands[1] are equal.
4520 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4521 =Q constraints to +Q and into insn condition add check
4522 that operands[0] is equal to either operands[1] or operands[2].
4525 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4526 instead of handling MINUS_EXPR twice (once for each argument),
4527 canonicalize operand order and handle just once, use rtype where
4528 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4531 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4532 any values of __C while still being pattern recognizable as a simple
4535 2017-10-13 Richard Biener <rguenther@suse.de>
4537 PR tree-optimization/82451
4539 2017-10-02 Richard Biener <rguenther@suse.de>
4541 PR tree-optimization/82355
4542 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4543 a mapping for the enclosing loop but avoid generating one for
4545 (copy_bb_and_scalar_dependences): Remove premature codegen
4546 error on PHIs in blocks duplicated into multiple places.
4547 * graphite-scop-detection.c
4548 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4549 in the region use it as loop and nest to analyze the DR in.
4550 (try_generate_gimple_bb): Likewise.
4551 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4552 (add_loop_constraints): For blocks in a loop not in the region
4553 create a dimension with a single iteration.
4554 * sese.h (gbb_loop_at_index): Remove assert.
4556 * cfgloop.c (loop_preheader_edge): For the loop tree root
4557 return the single successor of the entry block.
4558 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4559 Reset the SCEV hashtable and niters.
4560 * graphite-scop-detection.c
4561 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4562 assert that we only have POLYNOMIAL_CHREC that vary in loops
4563 contained in the region.
4564 (scop_detection::graphite_can_represent_expr): Adjust.
4565 (scop_detection::stmt_has_simple_data_refs_p): For loops
4566 not in the region set loop to NULL. The nest is now the
4567 entry edge to the region.
4568 (try_generate_gimple_bb): Likewise.
4569 * sese.c (scalar_evolution_in_region): Adjust for
4570 instantiate_scev change.
4571 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4572 Make nest parameter the edge into the region.
4573 (create_data_ref): Likewise.
4574 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4575 entry edge into a region and adjust instantiate_scev calls.
4576 (create_data_ref): Likewise.
4577 (graphite_find_data_references_in_stmt): Likewise.
4578 (find_data_references_in_stmt): Pass the loop preheader edge
4579 from the nest argument.
4580 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4581 parameter the edge into the region.
4582 (instantiate_parameters): Use the loop preheader edge as entry.
4583 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4585 (get_instantiated_value_entry): Make instantiate_below parameter
4586 the edge into the region.
4587 (instantiate_scev_name): Likewise. Adjust dominance checks,
4588 when we cannot use loop-based instantiation instantiate by
4589 walking use-def chains.
4590 (instantiate_scev_poly): Adjust.
4591 (instantiate_scev_binary): Likewise.
4592 (instantiate_scev_convert): Likewise.
4593 (instantiate_scev_not): Likewise.
4594 (instantiate_array_ref): Remove.
4595 (instantiate_scev_3): Likewise.
4596 (instantiate_scev_2): Likewise.
4597 (instantiate_scev_1): Likewise.
4598 (instantiate_scev_r): Do not blindly handle N-operand trees.
4599 Do not instantiate array-refs. Handle all constants and invariants.
4600 (instantiate_scev): Make instantiate_below parameter
4601 the edge into the region.
4602 (resolve_mixers): Use the loop preheader edge for the region
4603 parameter to instantiate_scev_r.
4604 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4606 2017-10-13 Richard Biener <rguenther@suse.de>
4608 PR tree-optimization/82525
4609 * graphite-isl-ast-to-gimple.c
4610 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4612 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4613 Fail code generation when we cannot represent the isl integer.
4614 (binary_op_to_tree): Elide modulo operations that are no-ops
4615 in the type we code generate. Remove now superfluous code
4618 2017-10-13 Richard Biener <rguenther@suse.de>
4620 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4621 (scop_detection::harmful_loop_in_region): Remove premature
4622 IV type restriction.
4623 (scop_detection::graphite_can_represent_scev): We can handle
4624 pointer IVs just fine.
4626 2017-10-13 Alan Modra <amodra@gmail.com>
4628 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4629 "Clobbers and Scratch Registers". Add paragraph on
4630 alternative to clobbers for scratch registers and OpenBLAS
4633 2017-10-13 Alan Modra <amodra@gmail.com>
4635 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4636 example of a memory input for a string of known length. Move
4637 commentary out of table. Add a number of new examples
4638 covering array memory inputs.
4640 2017-10-12 Martin Liska <mliska@suse.cz>
4642 PR tree-optimization/82493
4643 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4644 (test_range_functions): New function.
4645 (sbitmap_c_tests): Likewise.
4646 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4647 * selftest.h (sbitmap_c_tests): New function.
4649 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4651 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4653 * config/rs6000/amo.h: Fix spacing issue.
4655 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4658 * config/i386/i386.md (*ashl<mode>3_mask_1,
4659 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4660 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4663 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4665 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4666 (profile_probability): Set max_probability
4667 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4669 * profile-count.c (profile_probability::differs_from_p): Do not
4670 rely on max_probaiblity == 10000
4672 2017-10-12 Jeff Law <law@redhat.com>
4674 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4677 2017-10-12 Martin Sebor <msebor@redhat.com>
4681 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4682 (handle_alias_pairs): Call it.
4683 * common.opt (-Wattribute-alias): New option.
4684 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4685 * doc/invoke.texi (-Wattribute-alias): Document.
4687 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4690 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4692 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4694 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4695 (make_hard_regno_born, make_hard_regno_dead): Update
4696 bb_killed_pseudos and bb_gen_pseudos.
4698 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4700 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4702 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4704 * config/alpha/alpha.c (alpha_split_conditional_move):
4705 Use std::swap instead of manually swapping.
4706 (alpha_stdarg_optimize_hook): Ditto.
4707 (alpha_canonicalize_comparison): Ditto.
4709 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4711 * tree-loop-distribution.c (struct builtin_info): New struct.
4712 (struct partition): Refactor fields into struct builtin_info.
4713 (partition_free): Free struct builtin_info.
4714 (build_size_arg_loc, build_addr_arg_loc): Delete.
4715 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4716 information from struct builtin_info.
4717 (find_single_drs): New function refactored from classify_partition.
4718 Also moved builtin validity checks to this function.
4719 (compute_access_range, alloc_builtin): New functions.
4720 (classify_builtin_st, classify_builtin_ldst): New functions.
4721 (classify_partition): Refactor code into functions find_single_drs,
4722 classify_builtin_st and classify_builtin_ldst.
4723 (distribute_loop): Don't do runtime alias check when distributing
4725 (find_seed_stmts_for_distribution): New function.
4726 (pass_loop_distribution::execute): Refactor code finding seed
4727 stmts into above function. Support distribution for the innermost
4728 two-level loop nest. Adjust dump information.
4730 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4732 * tree-loop-distribution.c: Adjust the general comment.
4733 (NUM_PARTITION_THRESHOLD): New macro.
4734 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4735 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4736 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4737 in call to build_partition_graph.
4738 (finalize_partitions): New parameter. Make loop distribution more
4739 conservative by fusing more partitions.
4740 (distribute_loop): Don't do runtime alias check in case of loop nest
4742 (find_seed_stmts_for_distribution): New function.
4743 (prepare_perfect_loop_nest): New function.
4744 (pass_loop_distribution::execute): Refactor code finding seed stmts
4745 and loop nest into above functions. Support loop nest distribution.
4746 Adjust dump information accordingly.
4748 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4750 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4751 and set PTYPE_SEQUENTIAL for merged partition.
4753 2017-10-12 Richard Biener <rguenther@suse.de>
4755 PR tree-optimization/69728
4757 2017-09-19 Richard Biener <rguenther@suse.de>
4759 PR tree-optimization/69728
4760 * graphite-sese-to-poly.c (schedule_error): New global.
4761 (add_loop_schedule): Handle empty domain by failing the
4763 (build_original_schedule): Handle schedule_error.
4765 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4766 domain by returning an unchanged schedule.
4768 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4770 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4771 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4773 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4775 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4778 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4781 * expr.c (store_field): Don't optimize away bitsize == 0 store
4782 from CALL_EXPR with addressable return type.
4784 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4786 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4787 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4788 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4789 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4790 TARGET_ISEL instead of TARGET_ISEL<sel>.
4792 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4794 * config/rs6000/rs6000.c
4795 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4797 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4799 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4801 (reg_or_cint_operand): Fix comment.
4802 (reg_or_zero_operand): New predicate.
4803 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4804 * config/rs6000/rs6000.c (output_isel): Delete.
4805 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4806 instead of reg_or_cint_operand. Output instruction directly (not via
4808 (isel_unsigned_<mode>): Ditto.
4809 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4810 gpc_reg_operand. Add an instruction alternative for this. Output
4811 instruction directly.
4812 (*isel_reversed_unsigned_<mode>): Ditto.
4814 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4816 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4817 (TARGET_CANONICALIZE_COMPARISON): Define.
4819 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4822 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4823 Check whether the dest is REG before adding REG_EQUIV note.
4825 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4828 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4830 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4831 (make_hard_regno_born, make_hard_regno_dead): Update
4832 bb_killed_pseudos and bb_gen_pseudos.
4834 2017-10-11 Nathan Sidwell <nathan@acm.org>
4836 * incpath.h (enum incpath_kind): Name enum, prefix values.
4837 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4838 * incpath.c (heads, tails): Use INC_MAX.
4839 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4840 (merge_include_chains, split_quote_chain,
4841 register_include_chains): Update incpath_kind names.
4842 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4843 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4845 (add_framework_path, darwin_register_objc_includes): Likewise.
4846 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4848 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4850 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4851 Do not use float_operator operator predicate.
4852 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4853 * config/i386/predicates.md (float_operator): Remove predicate.
4855 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4857 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4858 (*jccxf_i387): Ditto.
4859 (*jcc<mode>_i387): Ditto.
4860 (*jccu<mode>_i387): Ditto.
4861 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4862 (*jcc_*_i387 splitters): Remove.
4863 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4864 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4865 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4868 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4870 * profile-count.h (slow_safe_scale_64bit): New function.
4871 (safe_scale_64bit): New inline.
4872 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4873 * profile-count.c: Include wide-int.h
4874 (slow_safe_scale_64bit): New.
4876 2017-10-11 Nathan Sidwell <nathan@acm.org>
4878 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4879 HAS_DECL_ASSEMBLER_NAME_P.
4880 * gimple-expr.c (gimple_decl_printable_name: Check
4881 HAS_DECL_ASSEMBLER_NAME_P too.
4882 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4883 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4884 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4885 * passes.c (rest_of_decl_compilation): Check
4886 HAS_DECL_ASSEMBLER_NAME_P too.
4887 * recog.c (verify_changes): Likewise.
4888 * tree-pretty-print.c (dump_decl_name): Likewise.
4889 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4891 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4892 (SET_DECL_ASSEMBLER_NAME): Use it.
4893 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4894 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4895 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4897 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4899 * config.gcc (i386, x86_64): Add extra objects.
4900 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4901 (ix86_min_insn_size): Declare.
4902 (ix86_issue_rate): Declare.
4903 (ix86_adjust_cost): Declare.
4904 (ia32_multipass_dfa_lookahead): Declare.
4905 (ix86_macro_fusion_p): Declare.
4906 (ix86_macro_fusion_pair_p): Declare.
4907 (ix86_bd_has_dispatch): Declare.
4908 (ix86_bd_do_dispatch): Declare.
4909 (ix86_core2i7_init_hooks): Declare.
4910 (ix86_atom_sched_reorder): Declare.
4911 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4912 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4913 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4914 (rip_relative_addr_p): Rename to ...
4915 (ix86_rip_relative_addr_p): ... this one; export.
4916 (memory_address_length): Update.
4917 (ix86_issue_rate): Move to x86-tune-sched.c.
4918 (ix86_flags_dependent): Move to x86-tune-sched.c.
4919 (ix86_agi_dependent): Move to x86-tune-sched.c.
4920 (exact_dependency_1): Move to x86-tune-sched.c.
4921 (exact_store_load_dependency): Move to x86-tune-sched.c.
4922 (ix86_adjust_cost): Move to x86-tune-sched.c.
4923 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4924 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4925 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4926 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4927 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4928 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4929 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4930 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4931 (min_insn_size): Rename to ...
4932 (ix86_min_insn_size): ... this one; export.
4933 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4934 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4935 (core2i7_first_cycle_multipass_backtrack): Move to
4936 x86-tune-sched-core.c.
4937 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4938 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4939 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4940 (ix86_avoid_jump_mispredicts): Update.
4941 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4942 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4943 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4944 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4945 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4946 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4947 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4948 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4949 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4950 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4951 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4952 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4953 (BIG): Move to ix86-tune-sched-bd.c.
4954 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4955 (enum insn_path): Move to ix86-tune-sched-bd.c.
4956 (get_mem_group): Move to ix86-tune-sched-bd.c.
4957 (is_cmp): Move to ix86-tune-sched-bd.c.
4958 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4959 (is_branch): Move to ix86-tune-sched-bd.c.
4960 (is_prefetch): Move to ix86-tune-sched-bd.c.
4961 (init_window): Move to ix86-tune-sched-bd.c.
4962 (allocate_window): Move to ix86-tune-sched-bd.c.
4963 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4964 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4965 (process_end_window): Move to ix86-tune-sched-bd.c.
4966 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4967 (find_constant): Move to ix86-tune-sched-bd.c.
4968 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4969 (has_immediate): Move to ix86-tune-sched-bd.c.
4970 (get_insn_path): Move to ix86-tune-sched-bd.c.
4971 (get_insn_group): Move to ix86-tune-sched-bd.c.
4972 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4973 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4974 (add_insn_window): Move to ix86-tune-sched-bd.c.
4975 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4976 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4977 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4978 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4979 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4980 (do_dispatch): Move to ix86-tune-sched-bd.c.
4981 (has_dispatch): Move to ix86-tune-sched-bd.c.
4982 * config/i386/t-i386: Add new object files.
4983 * config/i386/x86-tune-costs.h: New file.
4984 * config/i386/x86-tune-sched-atom.c: New file.
4985 * config/i386/x86-tune-sched-bd.c: New file.
4986 * config/i386/x86-tune-sched-core.c: New file.
4987 * config/i386/x86-tune-sched.c: New file.
4989 2017-10-11 Liu Hao <lh_mouse@126.com>
4991 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4992 the generic version below instead.
4993 (should_colorize): Recognize Windows consoles as terminals
4995 * pretty-print.c [__MINGW32__] (write_all): New function.
4996 [__MINGW32__] (find_esc_head): Likewise.
4997 [__MINGW32__] (find_esc_terminator): Likewise.
4998 [__MINGW32__] (eat_esc_sequence): Likewise.
4999 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5001 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5004 2017-10-11 Richard Biener <rguenther@suse.de>
5006 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5007 Properly call analyze_scalar_evolution with the loop of the stmt.
5009 2017-10-11 Richard Biener <rguenther@suse.de>
5011 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5012 * tree-core.h (tree_base): Add chrec_var union member.
5013 * tree.h (CHREC_VAR): Remove.
5014 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5015 * tree-chrec.h (build_polynomial_chrec): Adjust.
5016 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5017 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5019 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5021 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5022 * match.pd: ... here.
5023 ((T) X == (T) Y): Relax condition.
5025 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5027 PR tree-optimization/82472
5028 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5030 (break_alias_scc_partitions): Update postorder number.
5032 2017-10-11 Martin Liska <mliska@suse.cz>
5035 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5037 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5039 2017-10-11 Martin Liska <mliska@suse.cz>
5041 * print-rtl.c (print_insn): Move declaration of idbuf
5042 to same scope as name.
5044 2017-10-11 Martin Liska <mliska@suse.cz>
5049 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5052 2017-10-11 Martin Liska <mliska@suse.cz>
5055 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5058 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5060 * config.gcc (default_gnu_indirect_function): Default to yes for
5061 arm*-*-linux* with glibc.
5063 2017-10-11 Richard Biener <rguenther@suse.de>
5065 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5066 default-defs and types we do not want to analyze.
5067 (interpret_loop_phi): Replace unreachable code with an assert.
5068 (compute_scalar_evolution_in_loop): Remove and inline ...
5069 (analyze_scalar_evolution_1): ... here, replacing condition with
5070 what makes the intent clearer. Remove handling of cases
5071 get_scalar_evolution now handles.
5073 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5075 PR rtl-optimization/81434
5076 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5077 comment for main loop. In sched_group_found if, also add checks for
5078 pass and min_cost_group.
5080 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5082 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5083 (rs6000_insn_cost): New function.
5084 * config/rs6000/rs6000.md (cost): New attribute.
5086 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5087 H.J. Lu <hongjiu.lu@intel.com>
5091 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5092 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5093 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5094 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5095 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5096 OPTION_MASK_ISA_MMX and some other bit set require both
5097 mmx and the ISAs without the mmx bit.
5098 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5099 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5100 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5101 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5102 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5103 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5104 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5105 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5106 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5107 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5108 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5109 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5110 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5111 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5112 Add OPTION_MASK_ISA_MMX.
5114 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5116 * config.gcc (armv7*-*-freebsd*): New target.
5117 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5119 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5121 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5122 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5125 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5127 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5128 a template parameter.
5129 (WIDE_INT_REF_FOR): Update accordingly.
5130 * tree.h (wi::int_traits <const_tree>): Delete.
5131 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5132 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5133 (wi::tree_to_wide_ref): New typedef.
5134 (wi::to_wide): New function.
5135 * calls.c (get_size_range): Use wi::to_wide when operating on
5137 * cgraph.c (cgraph_node::create_thunk): Likewise.
5138 * config/i386/i386.c (ix86_data_alignment): Likewise.
5139 (ix86_local_alignment): Likewise.
5140 * dbxout.c (stabstr_O): Likewise.
5141 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5142 * expr.c (const_vector_from_tree): Likewise.
5143 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5144 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5145 (fold_negate_expr_1, int_const_binop_1, const_binop)
5146 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5147 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5148 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5149 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5150 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5151 (fold_not_const, round_up_loc): Likewise.
5152 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5153 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5154 (alloca_call_type): Likewise.
5155 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5156 * godump.c (go_output_typedef): Likewise.
5157 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5158 * internal-fn.c (get_min_precision): Likewise.
5159 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5160 * ipa-polymorphic-call.c
5161 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5162 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5163 (ipa_modify_call_arguments): Likewise.
5164 * match.pd: Likewise.
5165 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5166 * print-tree.c (print_node_brief, print_node): Likewise.
5167 * stmt.c (expand_case): Likewise.
5168 * stor-layout.c (layout_type): Likewise.
5169 * tree-affine.c (tree_to_aff_combination): Likewise.
5170 * tree-cfg.c (group_case_labels_stmt): Likewise.
5171 * tree-data-ref.c (dr_analyze_indices): Likewise.
5172 (prune_runtime_alias_test_list): Likewise.
5173 * tree-dump.c (dequeue_and_dump): Likewise.
5174 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5175 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5176 * tree-pretty-print.c (dump_generic_node): Likewise.
5177 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5178 (simple_iv_with_niters): Likewise.
5179 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5180 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5181 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5182 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5183 (refine_value_range_using_guard, number_of_iterations_ne_max)
5184 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5185 (get_cst_init_from_scev, record_nonwrapping_iv)
5186 (scev_var_range_cant_overflow): Likewise.
5187 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5188 * tree-ssa-pre.c (compute_avail): Likewise.
5189 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5190 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5191 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5192 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5193 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5194 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5195 (dump_case_nodes, try_switch_expansion): Likewise.
5196 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5197 (vect_do_peeling): Likewise.
5198 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5199 * tree-vect-stmts.c (vectorizable_load): Likewise.
5200 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5201 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5202 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5203 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5204 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5205 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5206 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5207 (evrp_dom_walker::before_dom_children): Likewise.
5208 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5209 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5210 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5211 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5212 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5213 (get_range_pos_neg): Likewise.
5214 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5215 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5216 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5217 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5218 * config/avr/avr.c (avr_fold_builtin): Likewise.
5219 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5220 * config/msp430/msp430.c (msp430_attr): Likewise.
5221 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5222 * config/powerpcspe/powerpcspe-c.c
5223 (altivec_resolve_overloaded_builtin): Likewise.
5224 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5225 (rs6000_expand_ternop_builtin): Likewise.
5226 * config/rs6000/rs6000-c.c
5227 (altivec_resolve_overloaded_builtin): Likewise.
5228 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5229 (rs6000_expand_ternop_builtin): Likewise.
5230 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5232 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5234 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5235 when copying loop nest with only one inner loop.
5237 2017-10-10 Richard Biener <rguenther@suse.de>
5239 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5240 blocks if SCEV is active.
5241 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5243 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5244 (scev_initialize): Assert we are not yet initialized.
5246 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5248 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5249 inner loop's exit stmt by making it always exit the loop, otherwise
5250 we would generate an infinite empty loop.
5252 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5254 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5255 renaming variables in new preheader if it's deleted.
5257 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5259 * tree-loop-distribution.c (struct partition): Remove unused field
5260 loops of the structure.
5261 (partition_alloc, partition_free): Ditto.
5262 (build_rdg_partition_for_vertex): Ditto.
5264 2017-10-09 Jeff Law <law@redhat.com>
5266 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5267 return type to match prototype and documentation.
5269 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5271 * config/rs6000/rs6000.c (processor_costs): Move to ...
5272 * config/rs6000/rs6000.h: ... here.
5273 (rs6000_cost): Declare.
5275 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5277 * except.c (setjmp_fn): New global variable.
5278 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5279 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5280 if DONT_USE_BUILTIN_SETJMP is defined.
5282 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5284 * target.def (insn_cost): New hook.
5285 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5286 * doc/tm.texi: Regenerate.
5287 * rtlanal.c (insn_cost): Use the new hook.
5289 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5291 * combine.c (combine_validate_cost): Compute the new insn_cost,
5292 not just pattern_cost.
5293 (try_combine): Adjust comment.
5295 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5297 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5299 * combine.c (uid_insn_cost): Adjust comment.
5300 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5302 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5303 * dse.c (find_shift_sequence): Ditto.
5304 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5305 (bb_valid_for_noce_process_p): Use pattern_cost.
5306 * rtl.h (insn_rtx_cost): Delete.
5307 (pattern_cost): New prototype.
5308 (insn_cost): New prototype.
5309 * rtlanal.c (insn_rtx_cost): Rename to...
5310 (pattern_cost): ... this.
5313 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5315 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5316 (*jcc<mode>_0_r_i387): Ditto.
5317 (*jccxf_r_i387): Ditto.
5318 (*jcc<mode>_r_i387): Ditto.
5319 (*jccu<mode>_r_i387): Ditto.
5320 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5321 (*jcc): Rename from *jcc_1.
5323 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5325 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5326 deferred rescans after the lvx/stvx recombination pre-pass.
5328 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5330 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5331 memory operation instruction support.
5332 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5333 (rs6000-ibm-aix[789]*): Likewise.
5334 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5335 Document new functions.
5337 2017-10-09 Richard Biener <rguenther@suse.de>
5339 PR tree-optimization/82397
5340 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5341 equality only for semantically equal trees.
5343 2017-10-09 Richard Biener <rguenther@suse.de>
5345 PR tree-optimization/82449
5346 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5347 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5348 allow constant addresses.
5349 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5352 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5354 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5357 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5360 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5363 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5366 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5368 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5371 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5372 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5374 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5376 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5377 (WI_BINARY_PREDICATE_RESULT): Likewise.
5378 (wi::binary_traits::operator_result): New type.
5379 (wi::binary_traits::predicate_result): Likewise.
5380 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5381 (generic_wide_int::operator==, generic_wide_int::operator!=)
5382 (generic_wide_int::operator&, generic_wide_int::and_not)
5383 (generic_wide_int::operator|, generic_wide_int::or_not)
5384 (generic_wide_int::operator^, generic_wide_int::operator+
5385 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5387 (operator~, unary operator-, operator==, operator!=, operator&)
5388 (operator|, operator^, operator+, binary operator-, operator*): New
5390 * expr.c (get_inner_reference): Use wi::bit_and_not.
5391 * fold-const.c (fold_binary_loc): Likewise.
5392 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5393 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5394 (bit_value_binop): Likewise.
5395 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5396 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5397 (extract_range_from_binary_expr_1): Likewise.
5398 (masked_increment): Likewise.
5399 (simplify_bit_ops_using_ranges): Likewise.
5401 2017-10-09 Martin Jambor <mjambor@suse.cz>
5404 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5405 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5406 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5408 (hsa_fixup_mov_insn_type): New function.
5409 (hsa_op_with_type::get_in_type): Use it.
5410 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5411 immediates in an assert.
5412 (hsa_op_with_type::extend_int_to_32bit): New method.
5413 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5414 types. Convert to dest type if necessary.
5415 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5416 (reg_for_gimple_ssa): Pass false as min32int to
5417 hsa_type_for_scalar_tree_type.
5418 (gen_hsa_addr): Fixup type when creating addresable temporary.
5419 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5420 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5421 necessary. Call hsa_fixup_mov_insn_type.
5422 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5423 extend operands and convert to dest type if necessary.
5424 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5425 to dest type if necessary.
5426 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5427 if conversion nt necessary and size matches.
5428 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5429 to dest type if necessary.
5430 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5431 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5433 (gen_hsa_clrsb): Likewise.
5434 (gen_hsa_ffs): Likewise.
5435 (gen_hsa_divmod): Extend operands and convert to dest type if
5437 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5439 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5441 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5442 Remove empty default arguments. Use a brace block as output
5444 (conditional return): Ditto.
5446 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5447 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5449 (group_ending_nop): Ditto.
5450 (doloop_end): Ditto.
5451 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5452 (splitters for those): Ditto.
5454 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5456 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5457 a conditional jump (and the compare for it) so that pc_rtx is the
5459 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5460 for the deleted and renamed ctr<mode>_internal[234] patterns.
5461 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5462 Delete second conditional return pattern.
5463 (ctr<mode>_internal2): Delete this second bdnz pattern.
5464 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5465 (ctr<mode>_internal4): Delete this second bdz pattern.
5467 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5469 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5470 (always_initialized_rtx_for_ssa_name_p): New predicate.
5471 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5472 (finish_out_of_ssa): Free new field of SA.
5473 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5474 * tree-ssa-coalesce.c: Include tree-ssa.h.
5475 (get_parm_default_def_partitions): Remove extern keyword.
5476 (get_undefined_value_partitions): New function.
5477 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5478 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5480 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5482 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5484 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5486 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5488 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5489 for targets that preffer 128bit.
5491 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5493 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5495 2017-10-08 Olivier Hainque <hainque@adacore.com>
5497 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5498 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5500 (thumb_set_return_address): Likewise.
5502 2017-10-08 Olivier Hainque <hainque@adacore.com>
5504 * common/config/arm/arm-common.c (arm_except_unwind_info):
5505 Handle DWARF2_UNWIND_INFO.
5507 2017-10-07 Michael Collison <michael.collison@arm.com>
5509 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5512 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5514 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5515 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5516 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5517 defined, force the creation of a new block for a dispatch label.
5519 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5521 * invoke.texi (Wsuggest-attribute=cold): Document.
5522 * common.opt (Wsuggest-attribute=cold): New
5523 * ipa-pure-const.c (warn_function_cold): New function.
5524 * predict.c (compute_function_frequency): Use it.
5525 * predict.h (warn_function_cold): Declare.
5527 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5529 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5532 2017-10-06 Martin Liska <mliska@suse.cz>
5534 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5535 keyword for member functions.
5536 (struct sanopt_tree_couple): New struct.
5537 (struct sanopt_tree_couple_hash): New function.
5538 (struct sanopt_ctx): Add new hash_map.
5539 (has_dominating_ubsan_ptr_check): New function.
5540 (record_ubsan_ptr_check_stmt): Likewise.
5541 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5542 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5543 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5545 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5548 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5549 aarch64_simd_valid_immediate on CONST_VECTORs.
5550 (aarch64_reg_or_bic_imm): Likewise.
5552 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5554 PR rtl-optimization/82396
5555 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5557 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5559 * graphite-dependences.c (scop_get_reads): Move code to...
5560 (scop_get_must_writes): Move code to...
5561 (scop_get_may_writes): Move code to...
5562 (scop_get_reads_and_writes): ... here.
5563 (scop_get_dependences): Call scop_get_reads_and_writes.
5565 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5567 PR tree-optimization/82434
5568 * fold-const.h (can_native_encode_type_p,
5569 can_native_encode_string_p): Remove.
5570 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5571 don't encode anything, just return what would be otherwise returned.
5572 (native_encode_fixed, native_encode_complex, native_encode_vector):
5574 (native_encode_string): Likewise. Inline by hand
5575 can_native_encode_string_p.
5576 (can_native_encode_type_p): Remove.
5577 (can_native_encode_string_p): Remove.
5578 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5579 STRING_CSTs using can_native_encode_string_p, test all
5580 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5581 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5582 argument from native_encode_expr.
5583 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5584 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5587 2017-10-06 Richard Biener <rguenther@suse.de>
5589 PR tree-optimization/82397
5590 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5591 operand_equal_p but rely on data_ref_compare_tree for detecting
5593 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5594 to match up with dr_group_sort_cmp.
5596 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5599 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5601 * config/s390/s390-builtin-types.def: Regenerate.
5603 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5606 * config/s390/s390-builtin-types.def: Regenerate.
5607 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5608 Change flag from B_VXE to B_VX.
5609 (s390_vec_min_dbl): Remove B_VXE flag.
5611 2017-10-06 Richard Biener <rguenther@suse.de>
5613 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5614 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5615 translate_isl_ast_to_gimple::is_valid_rename,
5616 translate_isl_ast_to_gimple::get_rename,
5617 translate_isl_ast_to_gimple::get_def_bb_for_const,
5618 translate_isl_ast_to_gimple::get_new_name,
5619 translate_isl_ast_to_gimple::collect_all_ssa_names,
5620 translate_isl_ast_to_gimple::copy_loop_phi_args,
5621 translate_isl_ast_to_gimple::collect_all_ssa_names,
5622 translate_isl_ast_to_gimple::copy_loop_phi_args,
5623 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5624 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5625 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5626 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5627 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5628 translate_isl_ast_to_gimple::copy_cond_phi_args,
5629 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5630 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5631 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5632 translate_isl_ast_to_gimple::rename_uses,
5633 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5634 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5635 (set_rename_for_each_def): Likewise.
5636 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5637 here. Handle rewriting SCEV analyzable uses here.
5638 (copy_bb_and_scalar_dependences): Generate code for PHI
5640 (graphite_regenerate_ast_isl): Adjust.
5641 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5642 (add_write, add_read): New functions.
5643 (build_cross_bb_scalars_def): Use it and simplify.
5644 (build_cross_bb_scalars_use): Likewise.
5645 (graphite_find_cross_bb_scalar_vars): Inline into...
5646 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5647 simulating out-of-SSA. Compute liveout and add dependencies.
5648 (build_scops): Force an empty entry block.
5649 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5651 (sese_build_liveouts): Declare.
5652 (sese_trivially_empty_bb_p): Likewise.
5653 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5654 compute liveout and debug_liveout.
5655 (sese_bad_liveouts_use): Remove.
5656 (sese_reset_debug_liveouts_bb): Likewise.
5657 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5658 (sese_build_liveouts): Build liveout and debug_liveout and store
5660 (new_sese_info): Adjust.
5661 (free_sese_info): Likewise.
5662 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5663 do not build liveout here.
5664 (move_sese_in_condition): Adjust region entry.
5665 (scev_analyzable_p): Match up with chrec_apply requirements.
5666 (sese_trivially_empty_bb_p): New.
5667 * tree-into-ssa.c (get_reaching_def): Properly support generating
5668 default-defs for incremental rewrite of anonymous names.
5670 2017-10-06 Richard Biener <rguenther@suse.de>
5672 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5673 precision do not perform modulo reduction.
5675 2017-10-06 Richard Biener <rguenther@suse.de>
5677 PR tree-optimization/82436
5678 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5679 conservatively choose the vectorization factor when checking
5680 whether we can perform the required load permutation.
5681 (vect_transform_slp_perm_load): Assert when we may not fail.
5683 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5685 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5686 message for incompatible -msdata=* and -mcall-* options.
5688 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5690 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5691 rate for post-reload scheduling.
5693 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5695 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5697 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5699 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5700 to improve monte carlo in scimark.
5702 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5704 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5705 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5706 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5707 pentium4_cost, nocona_cost): Set reassociation width to 1.
5708 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5709 width to 2 for fp operations and 1 otherwise.
5710 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5712 (atom_cost): Set reassociation width to 2.
5713 (slm_cost, generic_cost): Set fp reassociation width
5714 to 2 and 1 otherwise.
5715 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5716 (core_cost): Set fp reassociation width to 4 and vector to 2.
5717 (ix86_reassociation_width): Rewrite using cost table; special case
5718 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5719 and TARGET_AVX128_OPTIMAL.
5720 * config/i386/i386.h (processor_costs): Add
5721 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5722 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5723 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5724 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5725 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5726 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5728 2017-10-05 Nathan Sidwell <nathan@acm.org>
5730 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5732 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5734 * config/arm/arm.c (arm_test_fpu_data): New.
5735 (arm_run_selftests): Call arm_test_fpu_data.
5737 2017-10-04 Nathan Sidwell <nathan@acm.org>
5739 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5741 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5743 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5745 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5746 check type for aarch64_simd_valid_immediate.
5747 (aarch64_output_simd_mov_immediate): Update prototype.
5748 (aarch64_simd_valid_immediate): Update prototype.
5749 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5750 support for ORR-immediate.
5751 (and<mode>3): modified pattern to add support for BIC-immediate.
5752 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5753 now checks for valid immediate for BIC and ORR based on new enum
5755 (aarch64_output_simd_mov_immediate): Function now used to output
5756 BIC/ORR imm as well based on new enum argument.
5757 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5759 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5760 (aarch64_reg_or_bic_imm): Likewise.
5762 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5764 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5765 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5767 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5771 PR rtl-optimization/82396
5772 * haifa-sched.c (autopref_multipass_init): Simplify
5774 (autopref_rank_data): Simplify sort order.
5775 * sched-int.h (autopref_multipass_data_): Remove
5776 multi_mem_insn_p, min_offset and max_offset.
5778 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5780 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5782 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5784 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5787 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5789 * doc/sourcebuild.texi: Document vect_long_mult.
5791 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5793 PR tree-optimization/82413
5794 * fold-const.c (build_range_check): Use widest_int when comparing
5795 the maximum ETYPE value with HIGH.
5797 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5799 PR rtl-optimization/82396
5800 * haifa-sched.c (autopref_multipass_init): Simplify
5802 (autopref_rank_data): Simplify sort order.
5803 * sched-int.h (autopref_multipass_data_): Remove
5804 multi_mem_insn_p, min_offset and max_offset.
5806 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5808 PR tree-optimization/82381
5809 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5810 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5813 PR tree-optimization/82374
5814 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5815 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5816 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5817 current_function_decl to the new decl.
5819 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5821 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5822 helper macro for IEEE float128 hardware built-in functions.
5823 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5825 (TRUNCF128_ODD): Likewise.
5826 (ADDF128_ODD): Likewise.
5827 (SUBF128_ODD): Likewise.
5828 (MULF128_ODD): Likewise.
5829 (DIVF128_ODD): Likewise.
5830 (FMAF128_ODD): Likewise.
5831 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5832 UNSPEC_TRUNC_ROUND_TO_ODD.
5833 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5834 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5835 floating point round to odd instructions.
5836 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5837 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5838 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5839 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5840 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5841 (trunc<mode>sf2_hw): Change the truncate with round to odd
5842 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5843 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5844 to odd hardware instructions.
5845 (sub<mode>3_odd): Likewise.
5846 (mul<mode>3_odd): Likewise.
5847 (div<mode>3_odd): Likewise.
5848 (sqrt<mode>2_odd): Likewise.
5849 (fma<mode>4_odd): Likewise.
5850 (fms<mode>4_odd): Likewise.
5851 (nfma<mode>4_odd): Likewise.
5852 (nfms<mode>4_odd): Likewise.
5853 (trunc<mode>df2_odd): Change the truncate with round to odd
5854 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5856 * doc/extend.texi (PowerPC built-in functions): Update documentation
5857 for existing IEEE float128-bit built-in functions. Add built-in
5858 functions that generate the IEEE 128-bit floating point round to
5861 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5863 PR rtl-optimization/77729
5864 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5865 to (X&(C1&~C2))|C2 transformations.
5867 2017-10-03 Martin Jambor <mjambor@suse.cz>
5869 PR tree-optimization/82363
5870 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5871 mismatch, mark lacc written regardless of racc.
5873 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5875 PR tree-optimization/82381
5876 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5877 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5878 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5879 fallthrough into reassoc_stmt_dominates_stmt_p.
5882 * combine.c (combine_instructions): Don't combine in unreachable
5885 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5888 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5889 function to not use the have_cpu variable. Do not set cpu_index,
5890 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5892 (rs6000_valid_attribute_p): Remove duplicate initializations of
5893 old_optimize and func_optimize.
5894 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5895 (rs6000_activate_target_options): Make global.
5896 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5899 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5901 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5902 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5903 Set *poffset to 0 and *psize and *pmax_size to -1 if
5904 *poffset + *psize overflows in HOST_WIDE_INT.
5906 PR tree-optimization/82387
5907 PR tree-optimization/82388
5908 PR tree-optimization/82389
5909 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5910 instead of live_bytes non-NULL.
5912 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5915 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5918 2017-10-02 Richard Biener <rguenther@suse.de>
5920 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5921 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5922 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5924 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5926 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5927 requested precision matches the type's.
5928 * calls.c (alloc_max_size): Calculate the new candidate size as
5929 a widest_int and use wi::to_widest when comparing it with the
5930 current candidate size.
5931 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5932 zero rather than integer_zero_node.
5933 * match.pd: Check for a no-op conversion before using wi::add
5934 rather than after. Use tree_to_uhwi when summing small shift
5935 counts into an unsigned int.
5937 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5938 Alan Hayward <alan.hayward@arm.com>
5939 David Sherwood <david.sherwood@arm.com>
5942 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5943 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5944 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5945 POINTER_AND_FP_REGS.
5947 2017-10-02 Richard Biener <rguenther@suse.de>
5949 PR tree-optimization/82355
5950 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5951 a mapping for the enclosing loop but avoid generating one for
5953 (copy_bb_and_scalar_dependences): Remove premature codegen
5954 error on PHIs in blocks duplicated into multiple places.
5955 * graphite-scop-detection.c
5956 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5957 in the region use it as loop and nest to analyze the DR in.
5958 (try_generate_gimple_bb): Likewise.
5959 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5960 (add_loop_constraints): For blocks in a loop not in the region
5961 create a dimension with a single iteration.
5962 * sese.h (gbb_loop_at_index): Remove assert.
5964 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5966 * omp-expand.c (adjust_context_scope): New function.
5967 (expand_parallel_call): Call adjust_context_scope.
5969 2017-10-01 Jeff Law <law@redhat.com>
5971 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5972 dom_opt_dom_walker class with direct access to private members.
5973 Add comments. Call test_for_singularity.
5974 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5975 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5976 m_dummy_cond anymore.
5977 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5979 (pass_dominator:execute): Build the dummy_cond here and pass it
5980 to the dom_opt_dom_walker ctor.
5981 (test_for_singularity): New function.
5983 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5984 Maya Rashish <coypu@sdf.org>
5986 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5987 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5988 (alpha*-*-netbsd*) Use nbsd_tm_file.
5989 (arm*-*-netbsdelf*) Likewise.
5990 (i[34567]86-*-netbsdelf*) Likewise.
5991 (x86_64-*-netbsd*) Likewise.
5992 (mips*-*-netbsd*) Likewise.
5993 (powerpc-*-netbsd*) Likewise.
5994 (sh*-*-netbsd*) Likewise.
5995 (sparc-*-netbsdelf*) Likewise.
5996 (sparc64-*-netbsd*) Likewise.
5997 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5999 (vax-*-netbsdelf*) Likewise.
6000 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6001 (UINT_FAST8_TYPE) Likewise.
6002 (INT_FAST16_TYPE) Check CHAR_FAST16.
6003 (UINT_FAST16_TYPE) Likewise.
6005 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6008 * config/i386/i386.md
6009 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6010 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6011 *divmodsi4_zext_2): New define_insn_and_split.
6012 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6013 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6014 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6015 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6016 New define_insn_and_split.
6017 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6018 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6019 operands[1] having DImode when mode is SImode.
6021 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6022 always SImode for DIV and MOD in REG_EQUAL notes.
6024 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6027 * match.pd: Fix handling of NaNs in pattern.
6029 2017-09-29 Jeff Law <law@redhat.com>
6031 * sbitmap.c (bitmap_bit_in_range_p): New function.
6032 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6033 * tree-ssa-dse.c (live_bytes_read): New function.
6034 (dse_classify_store): Ignore reads of dead bytes.
6036 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6037 typos and whitespace errors.
6038 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6039 * config/s390/s390.c (s390_emit_prologue): Likewise.
6041 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6044 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6045 with a symbol for LRA.
6047 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6049 PR rtl-optimization/82338
6050 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6052 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6054 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6055 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6056 (qsort_chk): Declare.
6057 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6058 (qsort_chk): New function.
6060 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6062 PR tree-optimization/82337
6063 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6064 phi definition if the PHI result appears in an abnormal PHI.
6065 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6066 basis appears in an abnormal PHI.
6068 2017-09-29 Richard Biener <rguenther@suse.de>
6070 * graphite-isl-ast-to-gimple.c
6071 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6072 (binary_op_to_tree): Use it.
6073 (get_rename_from_scev): Likewise.
6074 (copy_loop_phi_nodes): Likewise.
6075 (copy_bb_and_scalar_dependences): Likewise.
6076 (translate_pending_phi_nodes): Likewise.
6078 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6081 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6082 for movabsq $(i32 << shift), r64.
6084 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6086 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6087 index when encoding %esp as %rsp to avoid 0x67 prefix.
6089 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6091 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6092 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6094 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6096 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6097 Extensions with more than 16 double VFP registers.
6098 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6099 to_clear_mask and all code related to it. Replace the remaining
6100 entry by a sbitmap and adapt code accordingly.
6102 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6104 * brig-builtins.def: Change pure attributes to const.
6106 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6108 * config.gcc (default_gnu_indirect_function): Default to yes for
6109 sparc*-*-linux* with glibc.
6111 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6113 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6114 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6115 when creating .init_array and .fini_array sections with priority
6118 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6121 * config/aarch64/aarch64.c
6122 (aarch64_builtin_support_vector_misalignment): Always return false
6123 when misalignment is unknown.
6125 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6127 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6128 this function to return false if the definition used by the swap
6129 instruction is artificial, or if the memory address from which the
6130 constant value is loaded is not represented by a base address held
6131 in a register or if the base address register is a frame or stack
6132 pointer. Additionally, return false if the base address of the
6133 loaded constant is a SYMBOL_REF but is not considered to be a
6135 (replace_swapped_load_constant): New function.
6136 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6137 loaded constant vector with a load of a swapped constant vector.
6139 2017-09-27 Carl Love <cel@us.ibm.com>
6141 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6142 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6143 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6146 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6148 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6149 first, always call autopref_rank_data otherwise.
6151 2017-09-27 Richard Biener <rguenther@suse.de>
6153 * graphite-scop-detection.c (find_scop_parameters): Move
6154 loop bound handling ...
6155 (gather_bbs::before_dom_children): ... here, avoiding the need
6156 to build scop_info->loop_nest.
6157 (record_loop_in_sese): Remove.
6158 * sese.h (sese_info_t::loop_nest): Remove.
6159 * sese.c (new_sese_info): Do not allocate loop_nest.
6160 (free_sese_info): Do not free loop_nest.
6162 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6165 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6166 lhs from calls if the lhs has addressable type.
6168 2017-09-27 Richard Biener <rguenther@suse.de>
6170 * graphite.h (scop::max_alias_set): New member.
6171 * graphite-scop-detection.c: Remove references to non-existing
6172 --param in comments.
6173 (build_alias_sets): Record the maximum alias set used for drs.
6174 (build_scops): Support zero as unlimited for
6175 --param graphite-max-arrays-per-scop.
6176 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6178 (build_poly_sr_1): ... here. Compute alias set based on the
6179 maximum alias set used for drs rather than
6180 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6182 2017-09-27 Richard Biener <rguenther@suse.de>
6184 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6185 --param loop-block-tile-size=0 to disable tiling.
6187 2017-09-27 Richard Biener <rguenther@suse.de>
6189 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6190 (graphite-max-nb-scop-params): Document special value zero.
6191 * domwalk.h (dom_walker::STOP): New symbolical constant.
6192 (dom_walker::dom_walker): Add optional parameter for bb to
6194 (dom_walker::~dom_walker): Declare.
6195 (dom_walker::before_dom_children): Document STOP return value.
6196 (dom_walker::m_user_bb_to_rpo): New member.
6197 (dom_walker::m_bb_to_rpo): Likewise.
6198 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6199 mapping here if not provided by the user.
6200 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6201 provided by the user.
6202 (dom_walker::STOP): Define.
6203 (dom_walker::walk): Do not compute bb to RPO mapping here.
6204 Support STOP return value from before_dom_children to stop
6206 * graphite-optimize-isl.c (optimize_isl): If the schedule
6207 is the same still generate code if -fgraphite-identity
6208 or -floop-parallelize-all are given.
6209 * graphite-scop-detection.c: Include cfganal.h.
6210 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6212 (gather_bbs::before_dom_children): Return STOP for BBs
6214 (build_scops): Compute bb to RPO mapping and pass it to
6215 the domwalk. Treat --param graphite-max-nb-scop-params=0
6216 as not limiting the number of params.
6217 * graphite.c (graphite_initialize): Remove limit on the
6218 number of basic-blocks in a function.
6219 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6220 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6221 default value of 10.
6223 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6225 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6226 Adjust code to eliminate needing to do the shift right 32-bits
6227 operation after XSCVDPSPN.
6229 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6231 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6232 ((X / Y) != 0 -> X >= Y): Likewise.
6234 2017-09-26 Carl Love <cel@us.ibm.com>
6236 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6237 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6238 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6239 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6240 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6241 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6242 definitions and overloading.
6243 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6244 statement for P9V_BUILTIN_XST_LEN_R.
6245 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6246 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6247 define_expand and define_insn for the instructions and builtins.
6248 * doc/extend.texi: Update the built-in documentation file for the new
6250 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6251 define_insn for the instructions
6253 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6256 * gcc/config/netbsd-protos.h: New file.
6257 * gcc/config/netbsd.c: New file.
6258 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6259 * gcc/config/t-netbsd: New file.
6260 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6261 (tmake_file) Add t-netbsd.
6262 (extra_objs) Add netbsd.o.
6264 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6268 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6270 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6272 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6273 sign extension from a vector register to a GPR by doing a 32-bit
6274 direct move and then an EXTSW.
6275 (extendsi<mode>2 splitter): Likewise.
6276 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6277 right or vector extract after doing XSCVDPSPN. Use
6278 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6280 (movdi_from_sf_zero_ext): Likewise.
6281 (reload_gpr_from_vsxsf): Likewise.
6282 (p8_mfvsrd_4_disf): Delete, no longer used.
6283 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6284 then needing to move the SFmode to a GPR to use the XSCVDPSP
6285 instruction instead of FRSP and XSCVDPSPN.
6286 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6287 it is adjacent to the other XSCVSPDP insns.
6288 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6289 SFmode to be in traditional Altivec registers.
6290 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6291 (vsx_xscvspdpn): Likewise.
6292 (vsx_xscvspdpn_scalar): Likewise.
6294 2017-09-26 Martin Jambor <mjambor@suse.cz>
6296 * tree-sra.c (compare_access_positions): Put integral types first,
6297 stabilize sorting of integral types, remove conditions putting
6298 non-full-precision integers last.
6299 (sort_and_splice_var_accesses): Disable scalarization if a
6300 non-integert would be represented by a non-full-precision integer.
6302 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6304 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6305 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6306 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6307 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6308 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6309 conditionals inside the function instead of around it. Call
6310 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6311 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6313 2017-09-26 Richard Biener <rguenther@suse.de>
6315 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6317 (scop_detection::build_scop_breadth): ... this. Removed.
6318 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6319 (scop_detection::harmful_stmt_in_bb): Likewise.
6320 (scop_detection::graphite_can_represent_stmt): Likewise.
6321 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6322 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6323 (scop_detection::can_represent_loop_1): ... this. Removed.
6324 (scop_detection::harmful_loop_in_region): Simplify after inlining
6325 the above and remove more quadraticness.
6326 (build_scops): Adjust.
6327 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6330 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6333 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6334 REGNO (base) == SP_REG if base is a REG.
6337 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6338 if it is different SSA_NAME.
6339 (optimize_range_tests_cmp_bitwise): New function.
6340 (optimize_range_tests): Call it.
6342 2017-09-26 Richard Biener <rguenther@suse.de>
6344 PR tree-optimization/82321
6345 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6346 for the def being inside the loop.
6348 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6350 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6352 * config/s390/s390-builtins.def: Fix constraint on op4.
6354 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6356 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6357 independent expanders.
6358 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6359 ("vec_ordered", "vec_unordered"): New expanders.
6361 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6363 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6366 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6368 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6369 vec_unpacks_lo_v16qi.
6370 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6372 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6374 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6375 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6376 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6378 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6380 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6382 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6384 ("*vec_slb<mode>"): New insn pattern.
6385 ("vec_shr_<mode>"): New expander.
6386 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6387 and force the shift count operand to V16QImode.
6388 ("vec_srb<mode>"): Set shift count mode to V16QI.
6390 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6392 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6393 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6394 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6396 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6399 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6401 2017-09-26 Richard Biener <rguenther@suse.de>
6403 PR tree-optimization/82320
6404 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6407 2017-09-25 Jeff Law <law@redhat.com>
6409 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6410 prototype for new argument.
6411 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6412 mostly extracted from rs6000_emit_allocate_stack.
6413 (rs6000_emit_probe_stack_range_stack_clash): New function.
6414 (rs6000_emit_allocate_stack): Call
6415 rs6000_emit_probe_stack_range_stack_clash as needed.
6416 (rs6000_emit_probe_stack_range): Add additional argument
6417 to call to gen_probe_stack_range{si,di}.
6418 (output_probe_stack_range): New.
6419 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6420 (output_probe_stack_range_stack_clash): New.
6421 (rs6000_emit_prologue): Emit notes into dump file as requested.
6422 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6423 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6424 Add additional operand and pass it to output_probe_stack_range.
6426 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6428 PR tree-optimization/82163
6429 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6430 (checking_verify_loop_closed_ssa): New parameter.
6431 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6432 (check_loop_closed_ssa_stmt): Delete.
6433 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6434 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6435 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6438 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6440 * brig-builtins.def: Treat HSAIL barrier builtins as
6441 setjmp/longjump style functions.
6443 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6445 * target.def (constant_alignment): New hook.
6446 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6447 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6448 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6449 * doc/tm.texi: Regenerate.
6450 * targhooks.h (default_constant_alignment): Declare.
6451 (constant_alignment_word_strings): Likewise.
6452 * targhooks.c (default_constant_alignment): New function.
6453 (constant_alignment_word_strings): Likewise.
6454 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6455 instead of CONSTANT_ALIGNMENT.
6456 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6457 (force_const_mem): Likewise.
6458 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6459 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6460 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6461 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6462 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6464 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6465 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6466 constant_alignment_word_strings.
6467 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6468 (CONSTANT_ALIGNMENT): Likewise.
6469 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6470 (arm_constant_alignment): New function.
6471 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6472 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6473 constant_alignment_word_strings.
6474 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6475 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6476 constant_alignment_word_strings.
6477 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6478 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6479 (cris_constant_alignment): New function.
6480 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6481 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6482 (epiphany_constant_alignment): New function.
6483 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6484 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6485 constant_alignment_word_strings.
6486 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6487 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6488 constant_alignment_word_strings.
6489 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6490 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6491 constant_alignment_word_strings.
6492 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6493 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6494 * config/i386/i386.c (ix86_constant_alignment): Make static.
6495 Use the same interface as the target hook.
6496 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6497 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6498 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6499 constant_alignment_word_strings.
6500 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6501 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6502 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6503 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6504 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6505 constant_alignment_word_strings.
6506 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6507 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6508 constant_alignment_word_strings.
6509 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6510 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6511 constant_alignment_word_strings.
6512 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6513 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6515 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6516 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6517 * config/mips/mips.c (mips_constant_alignment): New function.
6518 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6519 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6520 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6521 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6522 (mmix_constant_alignment): Make static. Use the same interface
6524 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6525 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6526 constant_alignment_word_strings.
6527 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6528 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6529 constant_alignment_word_strings.
6530 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6531 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6532 constant_alignment_word_strings.
6533 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6534 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6535 (rs6000_constant_alignment): New function.
6536 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6537 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6538 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6539 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6540 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6541 (rs6000_constant_alignment): New function.
6542 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6543 * config/s390/s390.c (s390_constant_alignment): New function.
6544 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6545 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6546 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6547 constant_alignment_word_strings.
6548 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6549 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6550 (sparc_constant_alignment): New function.
6551 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6552 * config/spu/spu.c (spu_constant_alignment): New function.
6553 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6554 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6555 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6556 constant_alignment_word_strings.
6557 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6558 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6559 constant_alignment_word_strings.
6560 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6561 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6562 constant_alignment_word_strings.
6563 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6564 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6565 (visium_constant_alignment): New function.
6566 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6567 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6568 (xtensa_constant_alignment): New function.
6569 * system.h (CONSTANT_ALIGNMENT): Poison.
6571 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6573 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6574 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6575 (rs6000_builtin_valid_without_lhs): New helper function.
6576 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6577 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6579 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6581 * target.h (vec_perm_indices): Use unsigned short rather than
6583 (auto_vec_perm_indices): Likewise.
6584 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6585 Use unsigned int rather than unsigned char.
6586 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6588 2017-09-25 Richard Biener <rguenther@suse.de>
6590 * cfgloop.h (sort_sibling_loops): Declare.
6591 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6592 (sort_sibling_loops): New function sorting the sibling loop list
6594 * graphite.c (graphite_transform_loops): Sort sibling loops.
6596 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6598 * target.def (vec_perm_const_ok): Change sel parameter to
6600 * optabs-query.c (can_vec_perm_p): Update accordingly.
6601 * doc/tm.texi: Regenerate.
6602 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6603 auto_vec_perm_indices and remove separate nelt field.
6604 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6605 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6606 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6607 (aarch64_expand_vec_perm_const): Update accordingly.
6608 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6609 to vec_perm_indices.
6610 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6611 auto_vec_perm_indices and remove separate nelt field.
6612 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6613 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6614 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6616 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6617 to vec_perm_indices.
6618 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6619 sel to vec_perm_indices.
6620 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6621 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6622 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6624 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6627 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6630 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6631 on the FUNCTION_DECL function context if it has a DIE that is a
6634 2017-09-25 Richard Biener <rguenther@suse.de>
6636 PR tree-optimization/82285
6637 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6640 2017-09-25 Tom de Vries <tom@codesourcery.com>
6644 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6647 2017-09-25 Richard Biener <rguenther@suse.de>
6649 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6650 ISL errors other than isl_error_quota happen. Dump if the
6651 schedule is the same.
6652 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6653 errors instead of aborting inside ISL.
6655 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6658 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6659 of libgcc_eh for m64.
6660 * config/i386/darwin64.h: Likewise.
6662 2017-09-25 Richard Biener <rguenther@suse.de>
6665 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6666 attribute for incomplete types nor twice for complete ones.
6668 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6671 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6672 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6675 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6678 * config/i386/i386.opt (mprefer-avx256): Use
6679 ix86_target_flags variable.
6680 * config/i386/i386.c (ix86_target_string): Move
6681 -mprefer-avx256 to flag2_opts.
6683 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6686 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6687 and x != -1 | y != -1 into (x & y) != -1.
6689 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6691 * config.gcc: Add new case statement to set
6692 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6693 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6694 s390x-*-linux* case statements. Added aarch64 to the list of
6695 supported architectures.
6697 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6699 PR tree-optimization/82289
6700 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6701 STMT_VINFO_RELEVANT_P.
6703 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6704 Alan Hayward <alan.hayward@arm.com>
6705 David Sherwood <david.sherwood@arm.com>
6707 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6708 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6709 (extract_range_from_binary_expr_1): Don't call
6710 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6712 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6713 Alan Hayward <alan.hayward@arm.com>
6714 David Sherwood <david.sherwood@arm.com>
6716 * target.def (preferred_vector_alignment): New hook.
6717 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6719 * doc/tm.texi: Regenerate.
6720 * targhooks.h (default_preferred_vector_alignment): Declare.
6721 * targhooks.c (default_preferred_vector_alignment): New function.
6722 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6724 (DR_TARGET_ALIGNMENT): New macro.
6725 (aligned_access_p): Update commentary.
6726 (vect_known_alignment_in_bytes): New function.
6727 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6729 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6730 Calculate the misalignment based on the target alignment rather than
6732 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6733 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6734 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6735 the target alignment, rather than masking the element misalignment
6736 with the number of elements in a vector. Also use the target
6737 alignment when calculating the maximum number of peels.
6738 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6739 instead of TYPE_ALIGN_UNIT.
6740 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6741 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6742 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6743 (vect_create_data_ref_ptr): Likewise.
6744 (vect_setup_realignment): Realign by ANDing with
6745 -DR_TARGET_MISALIGNMENT.
6746 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6747 the number of peels based on DR_TARGET_ALIGNMENT.
6748 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6749 with the guaranteed alignment boundary when deciding whether
6751 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6752 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6753 (ensure_base_align): Remove stmt_info parameter. Get the
6754 target base alignment from DR_TARGET_ALIGNMENT.
6755 (vectorizable_store): Update call accordingly. Interpret
6756 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6758 (vectorizable_load): Likewise.
6760 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6761 Alan Hayward <alan.hayward@arm.com>
6762 David Sherwood <david.sherwood@arm.com>
6764 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6765 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6766 (vect_enhance_data_refs_alignment): Likewise.
6768 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6770 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6771 error. Only quit immediately if parsing is complete.
6772 (BEGIN): Initialize fatal_err and parse_done.
6773 (begin fpu, end fpu): Check number of arguments.
6774 (begin arch, end arch): Likewise.
6775 (begin cpu, end cpu): Likewise.
6776 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6777 (optalias): Likewise.
6779 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6781 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6782 * config/arm/arm-isa.h: Delete. Move definitions to ...
6783 * arm-cpus.in: ... here. Use new feature and fgroup values.
6784 * config/arm/arm.c (arm_option_override): Use lower case for feature
6786 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6787 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6788 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6790 (print_isa_bits_for): New function.
6791 (gen_isa): New function.
6792 (gen_comm_data): Use print_isa_bits_for.
6793 (define feature): New keyword.
6794 (define fgroup): New keyword.
6795 * config/arm/t-arm (TM_H): Remove.
6796 (GTM_H): Add arm-isa.h.
6797 (arm-isa.h): Add rule to generate file.
6798 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6799 case for feature bit names.
6801 2017-09-22 Richard Biener <rguenther@suse.de>
6803 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6805 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6806 print of no dependency loops ...
6807 * graphite.c (graphite_transform_loops): ... here.
6808 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6810 (same_close_phi_node, remove_duplicate_close_phi,
6811 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6812 (canonicalize_loop_closed_ssa): ... here and simplify.
6813 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6814 (optimize_isl): Use dump_printf_loc to tell when we stopped
6815 optimizing because of an ISL timeout.
6817 2017-09-22 Richard Biener <rguenther@suse.de>
6819 PR tree-optimization/82291
6820 * tree-if-conv.c (predicate_mem_writes): Make sure to
6821 remove writes in blocks predicated with false.
6823 2017-09-22 Richard Biener <rguenther@suse.de>
6825 * sese.c: Include cfganal.h.
6826 (if_region_set_false_region): Remove.
6827 (create_if_region_on_edge): Likewise.
6828 (move_sese_in_condition): Re-implement without destroying
6831 2017-09-22 Richard Biener <rguenther@suse.de>
6833 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6834 Verify both BBs contain loop PHI nodes before dispatching to
6836 (graphite_regenerate_ast_isl): Do not recompute dominators,
6837 do not verify three times. Restructure for clarity.
6838 * graphite-scop-detection.c (same_close_phi_node,
6839 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6840 defined_in_loop_p, canonicalize_loop_closed_ssa,
6841 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6842 checking and SSA rewrite, move to ...
6843 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6844 (graphite_initialize): Do not pass in ctx, do not reset the
6845 SCEV cache, compute only dominators.
6846 (graphite_transform_loops): Allocate ISL ctx after
6847 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6848 Maintain post-dominators only around build_scops.
6849 * sese.c (if_region_set_false_region): Make static. Free
6850 and recompute dominators.
6851 (move_sese_in_condition): Assert we don't get called with
6852 post-dominators computed.
6853 * sese.h (if_region_set_false_region): Remove.
6855 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6857 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6858 mode attribute for TARGET_AVX512VL.
6860 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6862 * config/i386/i386.opt (mprefer-avx256): New option.
6863 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6865 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6866 for TARGET_PREFER_AVX256.
6867 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6869 2017-09-21 Jeff Law <law@redhat.com>
6871 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6872 Fix dump output if the only stack space is for pushed registers.
6874 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6876 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6879 2017-09-21 Martin Sebor <msebor@redhat.com>
6882 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6883 code (in C++) or code that triggers warnings.
6885 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6887 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6889 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6891 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6892 * sched-rgn.c: Ditto.
6893 * sel-sched-ir.c: Ditto.
6895 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6897 * toplev.h (set_random_seed): Adjust return type.
6898 * toplev.c (init_local_tick): Move eager initialization of random_seed
6899 to get_random_seed. Adjust comment.
6900 (init_random_seed): Inline to get_random_seed, delete.
6901 (get_random_seed): Initialize random_seed lazily.
6902 (set_random_seed): Do not return previous value.
6903 (print_switch_value): Do not call get_random_seed.
6905 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6907 * cgraph.c (delete_function_version): New, broken out from...
6908 (cgraph_node::delete_function_version): ...here. Rename to
6909 cgraph_node::delete_function_version_by_decl. Update all uses.
6910 (cgraph_node::remove): Call delete_function_version.
6912 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6915 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6916 VAR_DECLs to which addressable non-volatile parameters are mapped
6917 and for id->retvar after the return value assignment. Clear
6918 id->retval and id->retbnd after inlining.
6920 2017-09-21 Richard Biener <rguenther@suse.de>
6922 PR tree-optimization/82276
6923 PR tree-optimization/82244
6924 * tree-vrp.c (build_assert_expr_for): Set
6925 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6927 (remove_range_assertions): Revert earlier change.
6929 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6932 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6934 2017-09-21 Richard Biener <rguenther@suse.de>
6936 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6937 Restore valid IL after code generation errors.
6938 * graphite.c (graphite_transform_loops): Diagnose code
6939 generation issues as MSG_MISSED_OPTIMIZATION and continue
6940 with processing SCOPs.
6942 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6943 Alan Hayward <alan.hayward@arm.com>
6944 David Sherwood <david.sherwood@arm.com>
6946 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6947 rather than choosing between plus_constant and gen_rtx_<CODE>.
6948 * expr.c (emit_push_insn): Likewise.
6949 (expand_expr_real_2): Likewise.
6951 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6952 Alan Hayward <alan.hayward@arm.com>
6953 David Sherwood <david.sherwood@arm.com>
6955 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6957 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6958 Alan Hayward <alan.hayward@arm.com>
6959 David Sherwood <david.sherwood@arm.com>
6961 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6962 calling tree_to_uhwi.
6964 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6965 Alan Hayward <alan.hayward@arm.com>
6966 David Sherwood <david.sherwood@arm.com>
6968 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6969 INTEGER_CST rather than a negative test for ADDR_EXPR.
6971 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6972 Alan Hayward <alan.hayward@arm.com>
6973 David Sherwood <david.sherwood@arm.com>
6975 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6976 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6978 2017-09-21 Richard Biener <rguenther@suse.de>
6980 PR tree-optimization/71351
6981 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6982 graphite_create_new_loop_guard): Remove, fold remaining parts
6984 (translate_isl_ast_node_for): ... here and simplify.
6986 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6989 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6990 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6991 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6992 alternative always use QI mode, for -Os imov (=R,R) alternative
6993 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6996 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6997 Jeff Law <law@redhat.com>
6999 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7000 (allocate_stack_space): New function, partially extracted from
7002 (s390_emit_prologue): Track offset to most recent stack probe.
7003 Code to allocate space moved into allocate_stack_space.
7004 Dump actions when no stack is allocated.
7005 (s390_prologue_plus_offset): New function.
7006 (s390_emit_stack_probe): Likewise.
7008 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7010 * common.opt (Wa, Wl, Wp, g, gz=): Add
7012 (gno-column-info): Remove.
7013 (gcolumn-info): Drop RejectNegative.
7015 (gno-record-gcc-switches): Remove.
7016 (grecord-gcc-switches): Drop RejectNegative.
7017 (gno-split-dwarf): Remove.
7018 (gsplit-dwarf): Drop RejectNegative.
7019 (gno-strict-dwarf): Remove.
7020 (gstrict-dwarf): Drop RejectNegative.
7021 * config/darwin.opt (gfull, gused): Add RejectNegative.
7022 * dwarf2out.c (gen_producer_string): Drop
7023 gno-record-gcc-switches handler.
7024 * optc-gen.awk: Add g to prefixes with negative forms.
7025 * opts-common.c (remapping_prefix_p): New.
7026 (find_opt): Check it.
7027 (generate_canonical_option): Test g prefix.
7028 (option_map): Add -gno- mapping.
7029 (add_misspelling_candidates): Check remapping_prefix_p.
7031 2017-09-20 Jeff Law <law@redhat.com>
7033 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7034 thinko in stack clash protection support.
7036 * explow.c (compute_stack_clash_protection_loop_data): Use
7037 CONST_INT_P instead of explicit test. Verify object is a
7038 CONST_INT_P before looking at INTVAL.
7039 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7040 instead of explicit test.
7042 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7045 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7046 address instead of to r1 and r11.
7048 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7050 * config.gcc: Support "knm".
7051 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7052 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7054 * config/i386/i386.c (m_KNM): Define.
7055 (processor_target_table): Add "knm".
7057 (ix86_option_override_internal): Add "knm".
7058 (ix86_issue_rate): Add PROCESSOR_KNM.
7059 (ix86_adjust_cost): Ditto.
7060 (ia32_multipass_dfa_lookahead): Ditto.
7061 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7062 (fold_builtin_cpu): Add M_INTEL_KNM.
7063 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7064 (processor_type): Add PROCESSOR_KNM.
7065 * config/i386/x86-tune.def: Add m_KNM.
7066 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7068 2017-09-20 Richard Biener <rguenther@suse.de>
7070 PR tree-optimization/80213
7071 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7072 are allowed in empty BBs as well.
7073 (canonicalize_loop_closed_ssa): Also look for other complex
7075 (scop_detection::get_sese): Include the loop-closed PHI block
7077 (scop_detection::merge_sese): Remove code adding extra blocks.
7078 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7079 (build_scops): Assert the final returned scop is invalid.
7081 2017-09-20 Richard Biener <rguenther@suse.de>
7083 PR tree-optimization/82264
7084 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7086 (vn_phi_lookup): Likewise.
7087 (vn_phi_insert): Likewise.
7089 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7091 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7092 that fits into uhwi or shwi, add DW_AT_const_value regardless
7093 of early_dwarf without going through RTL, using add_AT_unsigned
7096 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7097 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7098 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7099 (DEBUG_MACRO_SECTION): Likewise.
7100 (DEBUG_LTO_MACRO_SECTION): Likewise.
7101 (DEBUG_STR_DWO_SECTION): Likewise.
7102 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7103 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7104 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7105 (DEBUG_LTO_LINE_STR_SECTION): Define.
7106 (init_sections_and_labels): Initialize debug_line_str_section
7107 variable. Initialize debug_loc_section for -gdwarf-5 to
7108 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7110 2017-09-20 Richard Biener <rguenther@suse.de>
7112 * graphite-sese-to-poly.c (extract_affine): Properly handle
7113 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7115 2017-09-20 Richard Biener <rguenther@suse.de>
7117 PR tree-optimization/81373
7118 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7119 Force SESE live-out defs to be handled even if they are
7122 2017-09-19 Jeff Law <law@redhat.com>
7124 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7125 nothing for stack adjustments with REG_STACK_CHECK.
7126 * sched-deps.c (parse_add_or_inc): Reject insns with
7127 REG_STACK_CHECK from dependency breaking.
7128 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7129 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7130 * reg-notes.def (STACK_CHECK): New note.
7132 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7133 (ix86_expand_prologue): Dump stack clash info as needed.
7134 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7136 * function.c (dump_stack_clash_frame_info): New function.
7137 * function.h (dump_stack_clash_frame_info): Prototype.
7138 (enum stack_clash_probes): New enum.
7140 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7141 flag_stack_clash_protection.
7142 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7143 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7144 (arm_frame_pointer_required): Likewise.
7145 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7146 (ia64_expand_prologue): Likewise.
7147 * config/mips/mips.c (mips_expand_prologue): Likewise.
7148 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7149 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7150 (sparc_flat_expand_prologue): Likewise.
7151 * config/spu/spu.c (spu_expand_prologue): Likewise.
7153 * explow.c: Include "params.h".
7154 (anti_adjust_stack_and_probe_stack_clash): New function.
7155 (get_stack_check_protect): Likewise.
7156 (compute_stack_clash_protection_loop_data): Likewise.
7157 (emit_stack_clash_protection_loop_start): Likewise.
7158 (emit_stack_clash_protection_loop_end): Likewise.
7159 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7160 Use anti_adjust_stack_and_probe_stack_clash.
7161 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7162 (emit_stack_clash_protection_loop_start): Likewise.
7163 (emit_stack_clash_protection_loop_end): Likewise.
7164 * rtl.h (get_stack_check_protect): Prototype.
7165 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7166 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7167 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7169 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7171 * doc/tm.texi: Rebuilt.
7172 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7173 get_stack_check_protect.
7174 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7175 * config/arm/arm.c (arm_expand_prologue): Likewise.
7176 (arm_frame_pointer_required): Likewise.
7177 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7178 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7179 * config/mips/mips.c (mips_expand_prologue): Likewise.
7180 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7181 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7182 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7183 (sparc_flat_expand_prologue): Likewise.
7185 * common.opt (-fstack-clash-protection): New option.
7186 * flag-types.h (enum stack_check_type): Note difference between
7187 -fstack-check= and -fstack-clash-protection.
7188 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7189 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7190 * toplev.c (process_options): Issue warnings/errors for cases
7191 not handled with -fstack-clash-protection.
7192 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7193 (-fstack-check): Note additional problem with -fstack-check=generic.
7194 Note that -fstack-check is primarily for Ada and refer users
7195 to -fstack-clash-protection for stack-clash-protection.
7196 Document new params for stack clash protection.
7198 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7200 * config/i386/i386.c (ix86_split_long_move): Do not handle
7201 address used for LEA in a special way.
7203 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7205 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7207 2017-09-19 Martin Sebor <msebor@redhat.com>
7210 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7211 of incompatible types.
7213 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7215 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7216 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7217 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7218 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7220 2017-09-19 Richard Biener <rguenther@suse.de>
7222 PR tree-optimization/82244
7223 * tree-vrp.c (remove_range_assertions): Do not propagate
7224 a constant to abnormals but replace the assert with a copy.
7226 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7228 PR rtl-optimization/57878
7229 PR rtl-optimization/68988
7230 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7231 avoidance test involving non_reload_pseudos. Move frequency test
7232 below the general fragmentation avoidance test.
7234 2017-09-19 Richard Biener <rguenther@suse.de>
7236 PR tree-optimization/69728
7237 * graphite-sese-to-poly.c (schedule_error): New global.
7238 (add_loop_schedule): Handle empty domain by failing the
7240 (build_original_schedule): Handle schedule_error.
7242 2017-09-19 Richard Biener <rguenther@suse.de>
7244 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7245 Do not iterate to sibling loops but only to siblings of inner
7248 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7251 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7252 that the registers are different.
7254 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7256 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7257 to processor_model and "amdfam17h" to arch_names_table.
7258 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7260 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7263 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7265 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7266 Alan Hayward <alan.hayward@arm.com>
7267 David Sherwood <david.sherwood@arm.com>
7269 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7271 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7273 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7274 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7275 vect_schedule_slp_instance.
7276 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7277 Update call to vect_slp_analyze_node_operations. Simplify return
7279 (vect_slp_analyze_bb_1): Update call accordingly.
7280 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7281 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7282 (vect_schedule_slp): Update call accordingly.
7284 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7285 Alan Hayward <alan.hayward@arm.com>
7286 David Sherwood <david.sherwood@arm.com>
7288 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7289 with types that aren't in fact scalar.
7291 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7293 * tree-vect-slp.c (vect_record_max_nunits): New function,
7295 (vect_build_slp_tree_1): ...here.
7296 (vect_build_slp_tree_2): Call it for phis too.
7298 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7300 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7301 to vect_get_vec_def_for_operand when getting the mask operand.
7303 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7304 Alan Hayward <alan.hayward@arm.com>
7305 David Sherwood <david.sherwood@arm.com>
7307 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7310 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7311 Alan Hayward <alan.hayward@arm.com>
7312 David Sherwood <david.sherwood@arm.com>
7314 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7315 calculation for vector booleans.
7317 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7318 Alan Hayward <alan.hayward@arm.com>
7319 David Sherwood <david.sherwood@arm.com>
7321 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7323 (vect_transform_stmt): ...here.
7324 (vect_analyze_stmt): Use it instead of calling
7325 vectorizable_live_operation directly.
7327 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7329 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7330 non-SIMT targets in acc vector loops.
7332 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7334 * configure.ac: Add arc and check if assembler supports gdwarf2.
7335 * configure: Regenerate.
7337 2017-09-18 Richard Biener <rguenther@suse.de>
7339 PR tree-optimization/82220
7340 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7341 epilogue niters from the min_profitable_iters compute.
7343 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7346 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7347 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7348 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7350 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7353 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7354 switching to a new text section.
7356 2017-09-18 Richard Biener <rguenther@suse.de>
7358 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7360 (build_alias_set): Reject aliases with no access function.
7362 2017-09-18 Richard Biener <rguenther@suse.de>
7364 PR tree-optimization/79622
7365 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7367 (build_cross_bb_scalars_use): Likewise.
7369 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7371 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7373 2017-09-18 Alan Modra <amodra@gmail.com>
7376 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7377 stack_pointer_rtx for count 0. Update comments. Break up
7378 large rtl expression.
7380 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7382 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7383 Increase to 20 bytes.
7384 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7385 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7386 or avx version of the stub.
7388 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7391 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7392 compute the minimum stack alignment. Also update preferred stack
7393 boundary for leaf functions.
7395 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7397 PR tree-optimization/82228
7398 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7401 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7403 * common/config/nds32/nds32-common.c
7404 (nds32_option_optimization_table): Refine formatting.
7405 (nds32_option_optimization_table): Use -fsched-pressure and
7406 -fomit-frame-pointer for specific optimization level.
7408 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7410 * config/nds32/nds32.c: Refine formatting and comments.
7411 * config/nds32/nds32.h: Likewise.
7412 * config/nds32/nds32.md: Likewise.
7413 * config/nds32/nds32-cost.c: Likewise.
7414 * config/nds32/nds32-isr.c: Likewise.
7415 * config/nds32/nds32-md-auxiliary.c: Likewise.
7416 * config/nds32/nds32-multiple.md: Likewise.
7417 * config/nds32/nds32-predicates.c: Likewise.
7419 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7420 Jakub Jelinek <jakub@redhat.com>
7422 Add support for -std=c++2a.
7423 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7425 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7427 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7430 * doc/extend.texi (Common Function Attributes): Add
7431 references to ARM, AArch64, and S/390 specific attributes.
7432 (Function Specific Option Pragmas): Add AArch64 and S/390
7433 to list of back ends that support the target pragma.
7435 2017-09-15 Nathan Sidwell <nathan@acm.org>
7437 * doc/standards.texi: Fix C++17 description. Update URLs for
7440 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7442 * common.opt (Wcast-align=strict): New warning option.
7443 * doc/invoke.texi: Document -Wcast-align=strict.
7445 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7447 * cgraph.h (cgraph_thunk_info): Add comments.
7448 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7449 assert for VIRTUAL_* arguments stricter.
7451 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7453 PR tree-optimization/71026
7454 * match.pd: Move RDIV patterns from fold-const.c
7455 * fold-const.c (distribute_real_division): Removed.
7456 (fold_binary_loc): Remove calls to distribute_real_divison.
7458 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7460 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7461 c++1z and gnu++1z as deprecated. Change other references to
7462 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7463 Change -Wc++1z-compat to -Wc++17-compat.
7464 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7465 * dwarf2out.c (highest_c_language): Handle C++17.
7466 (gen_compile_unit_die): Likewise.
7468 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7470 PR rtl-optimization/82192
7471 * combine.c (make_extraction): Don't look through non-paradoxical
7472 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7475 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7476 Alan Hayward <alan.hayward@arm.com>
7477 David Sherwood <david.sherwood@arm.com>
7479 * target.def (function_arg_offset): New hook.
7480 * targhooks.h (default_function_arg_offset): Declare.
7481 * targhooks.c (default_function_arg_offset): New function.
7482 * function.c (locate_and_pad_parm): Use
7483 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7484 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7485 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7486 * doc/tm.texi: Regenerate.
7487 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7488 * config/spu/spu.c (spu_function_arg_offset): New function.
7489 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7490 * system.h (FUNCTION_ARG_OFFSET): Poison.
7492 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7493 Alan Hayard <alan.hayward@arm.com>
7494 David Sherwood <david.sherwood@arm.com>
7496 * target.def (truly_noop_truncation): New hook.
7497 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7498 than TRULY_NOOP_TRUNCATION.
7499 * hooks.h (hook_bool_uint_uint_true): Declare.
7500 * hooks.c (hook_bool_uint_uint_true): New function.
7501 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7502 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7503 * doc/tm.texi: Regenerate.
7504 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7505 rather than TRULY_NOOP_TRUNCATION in comments.
7506 (simplify_comparison): Likewise.
7507 (record_truncated_value): Likewise.
7508 * expmed.c (extract_bit_field_1): Likewise.
7509 (extract_split_bit_field): Likewise.
7510 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7511 instead of TRULY_NOOP_TRUNCATION.
7512 * function.c (assign_parm_setup_block): Likewise.
7513 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7514 * rtlhooks.c: Include target.h.
7515 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7516 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7517 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7518 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7519 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7520 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7521 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7522 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7523 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7524 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7525 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7526 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7527 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7528 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7529 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7530 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7531 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7532 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7533 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7534 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7535 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7536 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7537 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7538 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7539 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7540 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7541 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7542 rather than TRULY_NOOP_TRUNCATION in comments.
7543 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7544 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7545 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7546 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7547 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7548 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7549 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7550 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7551 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7552 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7553 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7554 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7555 rather than TRULY_NOOP_TRUNCATION in comments.
7556 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7557 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7558 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7559 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7560 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7561 TRULY_NOOP_TRUNCATION condition.
7562 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7563 (TRULY_NOOP_TRUNCATION): Delete.
7564 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7565 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7566 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7567 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7568 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7569 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7570 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7571 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7572 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7573 rather than TRULY_NOOP_TRUNCATION in comments.
7574 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7575 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7576 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7577 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7578 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7579 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7581 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7584 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7585 (*cmp_ior): Likewise.
7586 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7587 (*ior_scc_scc_cmp): Likewise.
7588 (*and_scc_scc): Likewise.
7589 (*and_scc_scc_cmp): Likewise.
7591 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7592 Alan Hayard <alan.hayward@arm.com>
7593 David Sherwood <david.sherwood@arm.com>
7595 * target.def (can_change_mode_class): New hook.
7596 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7597 (hard_regno_nregs): Likewise.
7598 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7599 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7600 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7601 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7602 (LOAD_EXTEND_OP): Update accordingly.
7603 * doc/tm.texi: Regenerate.
7604 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7605 CANNOT_CHANGE_MODE_CLASS.
7606 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7607 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7608 * combine.c (simplify_set): Update accordingly.
7609 * emit-rtl.c (validate_subreg): Likewise.
7610 * recog.c (general_operand): Likewise.
7611 * regcprop.c (mode_change_ok): Likewise.
7612 * reload1.c (choose_reload_regs): Likewise.
7613 (inherit_piecemeal_p): Likewise.
7614 * rtlanal.c (simplify_subreg_regno): Likewise.
7615 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7616 instead of CANNOT_CHANGE_MODE_CLASS.
7617 (reload_cse_simplify_operands): Likewise.
7618 * reload.c (push_reload): Use targetm.can_change_mode_class
7619 instead of CANNOT_CHANGE_MODE_CLASS.
7620 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7621 REG_CANNOT_CHANGE_MODE_P.
7622 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7623 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7624 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7625 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7626 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7627 (arm_can_change_mode_class): New function.
7628 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7629 than CANNOT_CHANGE_MODE_CLASS in comments.
7630 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7631 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7632 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7633 (ix86_can_change_mode_class): ...this new function, inverting the
7634 sense of the return value.
7635 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7636 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7637 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7638 (ia64_can_change_mode_class): New function.
7639 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7640 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7641 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7642 (m32c_can_change_mode_class): ...this new function, inverting the
7643 sense of the return value.
7644 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7645 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7646 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7647 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7648 (mips_can_change_mode_class): ...this new function, inverting the
7649 sense of the return value.
7650 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7651 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7652 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7653 (msp430_can_change_mode_class): New function.
7654 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7655 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7656 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7657 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7658 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7659 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7660 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7661 (pa_cannot_change_mode_class): Replace with...
7662 (pa_can_change_mode_class): ...this new function, inverting the
7663 sense of the return value.
7664 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7665 than CANNOT_CHANGE_MODE_CLASS in comments.
7666 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7667 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7668 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7669 (pdp11_cannot_change_mode_class): Replace with...
7670 (pdp11_can_change_mode_class): ...this new function, inverting the
7671 sense of the return value.
7672 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7673 * config/powerpcspe/powerpcspe-protos.h
7674 (rs6000_cannot_change_mode_class_ptr): Delete.
7675 * config/powerpcspe/powerpcspe.c
7676 (rs6000_cannot_change_mode_class_ptr): Delete.
7677 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7678 (rs6000_option_override_internal): Assign to
7679 targetm.can_change_mode_class instead of
7680 rs6000_cannot_change_mode_class_ptr.
7681 (rs6000_cannot_change_mode_class): Replace with...
7682 (rs6000_can_change_mode_class): ...this new function, inverting the
7683 sense of the return value.
7684 (rs6000_debug_cannot_change_mode_class): Replace with...
7685 (rs6000_debug_can_change_mode_class): ...this new function.
7686 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7687 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7688 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7689 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7690 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7692 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7693 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7694 (rs6000_option_override_internal): Assign to
7695 targetm.can_change_mode_class instead of
7696 rs6000_cannot_change_mode_class_ptr.
7697 (rs6000_cannot_change_mode_class): Replace with...
7698 (rs6000_can_change_mode_class): ...this new function, inverting the
7699 sense of the return value.
7700 (rs6000_debug_cannot_change_mode_class): Replace with...
7701 (rs6000_debug_can_change_mode_class): ...this new function.
7702 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7703 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7704 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7705 (s390_can_change_mode_class): ...this new function, inverting the
7706 sense of the return value.
7707 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7708 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7709 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7710 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7711 (sh_cannot_change_mode_class): Replace with...
7712 (sh_can_change_mode_class): ...this new function, inverting the
7713 sense of the return value.
7714 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7715 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7716 (sparc_can_change_mode_class): New function.
7717 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7718 * config/spu/spu.c (spu_can_change_mode_class): New function.
7719 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7720 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7721 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7722 (visium_can_change_mode_class): New function.
7723 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7725 2017-09-15 Richard Biener <rguenther@suse.de>
7727 PR tree-optimization/82217
7728 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7729 but not undefined case.
7731 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7734 * postreload.c (reload_cse_simplify_operands): Skip
7735 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7737 2017-09-15 Richard Biener <rguenther@suse.de>
7739 PR tree-optimization/68823
7740 * graphite-scop-detection.c (build_alias_set): If we have a
7741 possible dependence check whether we can handle them by just
7742 looking at the DRs DR_ACCESS_FNs.
7743 (build_scops): If build_alias_set fails, fail the SCOP.
7745 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7747 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7748 to support float128 built-in functions that require the ISA 3.0
7750 (BU_FLOAT128_3_HW): Likewise.
7751 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7753 (FMAF128): Likewise.
7755 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7756 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7757 floating point instructions.
7758 (rs6000_invalid_builtin): Likewise.
7759 (rs6000_builtin_mask_names): Likewise.
7760 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7761 (RS6000_BTM_FLOAT128_HW): Likewise.
7762 (RS6000_BTM_COMMON): Likewise.
7763 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7765 * doc/extend.texi (RS/6000 built-in functions): Document the
7766 IEEE 128-bit floating point square root and fused multiply-add
7769 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7771 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7772 reg (r2) isn't in the set of registers defined in the prologue.
7774 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7775 Alan Hayward <alan.hayward@arm.com>
7776 David Sherwood <david.sherwood@arm.com>
7778 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7779 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7780 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7781 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7782 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7784 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7785 max_vectorization_factor.
7786 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7788 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7789 Alan Hayward <alan.hayward@arm.com>
7790 David Sherwood <david.sherwood@arm.com>
7792 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7793 (vect_worthwhile_without_simd_p): Declare.
7794 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7795 (vectorizable_reduction): Use it.
7796 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7797 (vectorizable_operation): Likewise.
7799 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7800 Alan Hayward <alan.hayward@arm.com>
7801 David Sherwood <david.sherwood@arm.com>
7803 * tree-vectorizer.h (vect_get_num_copies): New function.
7804 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7805 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7806 (vectorizable_induction): Likewise.
7807 (vectorizable_live_operation): Likewise.
7808 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7809 (vectorizable_bswap): Likewise.
7810 (vectorizable_call): Likewise.
7811 (vectorizable_conversion): Likewise.
7812 (vectorizable_assignment): Likewise.
7813 (vectorizable_shift): Likewise.
7814 (vectorizable_operation): Likewise.
7815 (vectorizable_store): Likewise.
7816 (vectorizable_load): Likewise.
7817 (vectorizable_condition): Likewise.
7818 (vectorizable_comparison): Likewise.
7819 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7821 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7822 Alan Hayward <alan.hayward@arm.com>
7823 David Sherwood <david.sherwood@arm.com>
7825 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7826 of vect_init_vector.
7828 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7829 Alan Hayward <alan.hayward@arm.com>
7830 David Sherwood <david.sherwood@arm.com>
7832 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7833 an inline wrapper that provides a location.
7834 (gimple_build_vector): Likewise.
7835 * gimple-fold.c (gimple_build_vector_from_val): New function.
7836 (gimple_build_vector): Likewise.
7837 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7838 functions to build the initial value. Always return a gimple value.
7839 (get_initial_defs_for_reduction): Likewise. Only compute
7841 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7842 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7843 (vectorizable_induction): Use gimple_build_vector rather than
7846 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7847 Alan Hayward <alan.hayward@arm.com>
7848 David Sherwood <david.sherwood@arm.com>
7850 * target.h (vec_perm_indices): New typedef.
7851 (auto_vec_perm_indices): Likewise.
7852 * optabs-query.h: Include target.h
7853 (can_vec_perm_p): Take a vec_perm_indices *.
7854 * optabs-query.c (can_vec_perm_p): Likewise.
7855 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7856 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7857 * tree-vect-generic.c (lower_vec_perm): Likewise.
7858 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7859 (vect_grouped_load_supported): Likewise.
7860 (vect_shift_permute_load_chain): Likewise.
7861 (vect_permute_store_chain): Use auto_vec_perm_indices.
7862 (vect_permute_load_chain): Likewise.
7863 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7864 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7865 Update uses of can_vec_perm_p.
7866 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7867 mode with a number of elements. Take a vec_perm_indices *.
7868 (vect_create_epilog_for_reduction): Update accordingly.
7869 Use auto_vec_perm_indices.
7870 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7871 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7872 (vect_transform_slp_perm_load): Likewise.
7873 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7874 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7875 (vect_gen_perm_mask_checked): Likewise.
7876 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7877 (vect_gen_perm_mask_checked): Likewise.
7878 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7879 (vectorizable_store): Likewise.
7880 (vectorizable_load): Likewise.
7881 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7882 (vectorizable_bswap): Likewise.
7884 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7885 Alan Hayward <alan.hayward@arm.com>
7886 David Sherwood <david.sherwood@arm.com>
7888 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7889 * tree.c (build_vector): Likewise.
7890 (build_vector_from_ctor): Update accordingly.
7891 (build_vector_from_val): Likewise.
7892 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7893 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7894 * tree-vect-generic.c (add_rshift): Likewise.
7895 (expand_vector_divmod): Likewise.
7896 (optimize_vector_constructor): Likewise.
7897 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7898 (vect_transform_slp_perm_load): Likewise.
7899 (vect_schedule_slp_instance): Likewise.
7900 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7901 (vectorizable_call): Likewise.
7902 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7903 * expmed.c (make_tree): Likewise.
7904 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7905 a vector passed to build_vector.
7906 (fold_convert_const): Likewise.
7907 (exact_inverse): Likewise.
7908 (fold_ternary_loc): Likewise.
7909 (fold_relational_const): Likewise.
7910 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7911 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7912 (const_unop): Likewise. Store the reduction accumulator in a
7913 variable rather than an array.
7914 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7915 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7916 the new vector, rather than constructing it after the input arrays.
7917 (native_interpret_vector): Use auto_vec<tree> when building
7918 a vector passed to build_vector. Add elements in order.
7919 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7920 auto_vec<tree> when building a vector passed to build_vector.
7921 (vect_create_epilog_for_reduction): Likewise.
7922 (vectorizable_induction): Likewise.
7923 (get_initial_def_for_reduction): Likewise. Fix indentation of
7925 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7927 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7928 passed to build_vector.
7930 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7931 Alan Hayward <alan.hayward@arm.com>
7932 David Sherwood <david.sherwood@arm.com>
7934 * tree-core.h (tree_base::u): Add an "nelts" field.
7935 (tree_vector): Use VECTOR_CST_NELTS as the length.
7936 * tree.c (tree_size): Likewise.
7937 (make_vector): Initialize VECTOR_CST_NELTS.
7938 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7939 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7940 TYPE_VECTOR_SUBPARTS.
7941 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7942 as the number of units, setting it from VECTOR_CST_NELTS.
7943 (const_vector_from_tree): Likewise.
7944 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7945 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7946 (fold_negate_expr_1): Likewise.
7947 (fold_convert_const): Likewise.
7948 (const_binop): Likewise. Differentiate the number of output and
7950 (const_unop): Likewise.
7951 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7952 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7953 in cases that did the opposite.
7955 2017-09-14 Richard Biener <rguenther@suse.de>
7957 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7960 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7962 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7964 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7967 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7968 if and where to split a bb, except for splitting before debug insn
7969 sequences followed by non-label real insn. Delete debug insns
7970 in between basic blocks.
7972 * combine.c (make_compound_operation_int): Formatting fixes.
7974 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7975 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7976 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7977 * config/sol2.h (LINK_EH_SPEC): Likewise.
7978 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7979 * config/s390/linux.h (LINK_SPEC): Likewise.
7980 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7981 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7982 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7983 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7984 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7985 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7986 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7987 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7989 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7991 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7993 (ENDFILE_LINUX_SPEC): Likewise.
7994 (LINK_EH_SPEC): Likewise.
7995 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7996 (LINK_OS_LINUX_SPEC32): Likewise.
7997 (LINK_OS_LINUX_SPEC64): Likewise.
7998 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7999 (LINK_OS_LINUX_SPEC): Likewise.
8001 2017-09-13 Martin Liska <mliska@suse.cz>
8004 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8005 CASE_HIGH is NULL_TREE.
8007 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8008 Alan Hayward <alan.hayward@arm.com>
8009 David Sherwood <david.sherwood@arm.com>
8011 * target.def (secondary_memory_needed): New hook.
8012 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8013 instead of SECONDARY_MEMORY_NEEDED.
8014 (secondary_memory_needed_mode): Likewise.
8015 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8016 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8017 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8018 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8019 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8020 * doc/tm.texi: Regenerate.
8021 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8022 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8023 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8024 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8025 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8026 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8027 mode argument first and change the reg_class arguments to reg_class_t.
8028 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8029 Make static. Update the call to inline_secondary_memory_needed.
8030 (ix86_register_move_cost): Update the call to
8031 inline_secondary_memory_needed.
8032 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8033 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8035 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8036 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8038 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8039 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8040 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8041 and match hook interface. Add comment from mips.h.
8042 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8043 * config/mmix/mmix.md (truncdfsf2): Refer to
8044 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8046 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8047 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8048 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8049 (pa_secondary_memory_needed): New function.
8050 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8051 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8052 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8053 (pdp11_secondary_memory_needed): Make static and match hook interface.
8054 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8055 * config/powerpcspe/powerpcspe-protos.h
8056 (rs6000_secondary_memory_needed_ptr): Delete.
8057 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8059 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8060 (rs6000_option_override_internal): Assign to
8061 targetm.secondary_memory_needed rather than
8062 rs6000_secondary_memory_needed_ptr.
8063 (rs6000_secondary_memory_needed): Match hook interface.
8064 (rs6000_debug_secondary_memory_needed): Likewise.
8065 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8066 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8067 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8068 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8069 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8070 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8072 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8073 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8074 (rs6000_option_override_internal): Assign to
8075 targetm.secondary_memory_needed rather than
8076 rs6000_secondary_memory_needed_ptr.
8077 (rs6000_secondary_memory_needed): Match hook interface.
8078 (rs6000_debug_secondary_memory_needed): Likewise.
8079 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8080 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8081 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8082 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8083 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8084 (sparc_secondary_memory_needed): New function.
8085 * lra-constraints.c (check_and_process_move): Refer to
8086 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8088 (curr_insn_transform): Likewise.
8089 (process_alt_operands): Use targetm.secondary_memory_needed
8090 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8091 (check_secondary_memory_needed_p): Likewise.
8092 (choose_split_class): Likewise.
8093 * reload.c: Unconditionally include code that was previously
8094 conditional on SECONDARY_MEMORY_NEEDED.
8095 (push_secondary_reload): Use targetm.secondary_memory_needed
8096 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8097 (push_reload): Likewise.
8098 * reload1.c: Unconditionally include code that was previously
8099 conditional on SECONDARY_MEMORY_NEEDED.
8100 (choose_reload_regs): Use targetm.secondary_memory_needed
8101 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8102 (gen_reload): Likewise.
8103 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8105 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8106 Alan Hayward <alan.hayward@arm.com>
8107 David Sherwood <david.sherwood@arm.com>
8109 * target.def (secondary_memory_needed_mode): New hook:
8110 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8111 * targhooks.h (default_secondary_memory_needed_mode): New function.
8112 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8113 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8114 * doc/tm.texi: Regenerate.
8115 * lra-constraints.c (check_and_process_move): Use
8116 targetm.secondary_memory_needed_mode instead of
8117 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8118 (curr_insn_transform): Likewise.
8119 * reload.c (get_secondary_mem): Likewise.
8120 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8121 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8123 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8124 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8125 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8126 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8127 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8129 * config/powerpcspe/powerpcspe-protos.h
8130 (rs6000_secondary_memory_needed_mode): Delete.
8131 * config/powerpcspe/powerpcspe.c
8132 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8133 (rs6000_secondary_memory_needed_mode): Make static.
8134 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8135 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8137 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8139 (rs6000_secondary_memory_needed_mode): Make static.
8140 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8141 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8142 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8143 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8144 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8146 (sparc_secondary_memory_needed_mode): New function.
8147 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8149 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8151 * config/aarch64/constraints.md (Umq): New constraint.
8152 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8154 (mov<mode>): Update condition.
8156 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8158 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8159 when bitposition is the same.
8161 2017-09-13 Richard Biener <rguenther@suse.de>
8163 * dwarf2out.c (output_die_symbol): Remove.
8164 (output_die): Do not output a DIEs symbol.
8166 2017-09-13 Richard Biener <rguenther@suse.de>
8169 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8170 default-def to avoid breaking iterator update with the weird
8171 interaction with cgraph_update_edges_for_call_stmt_node.
8173 2017-09-13 Richard Biener <rguenther@suse.de>
8175 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8176 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8177 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8178 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8180 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8182 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8183 Disable pc relative literal load irrespective of
8184 TARGET_FIX_ERR_A53_84341 for default.
8186 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8188 * config/sparc/sparc.c (output_return): Output the source location of
8189 the insn in the delay slot, if any.
8190 (output_sibcall): Likewise.
8192 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8195 * common.opt (-static-pie): New alias.
8196 (shared): Negate static-pie.
8197 (-no-pie): Update help text.
8199 (static-pie): New option.
8200 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8201 -static-pie support.
8202 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8203 (LINK_EH_SPEC): Likewise.
8204 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8205 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8206 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8207 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8208 (init_gcc_specs): Likewise.
8209 (init_spec): Likewise.
8210 (display_help): Update help message for -pie.
8211 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8214 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8216 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8217 (movdi_aarch64): Likewise.
8218 (movti_aarch64): Likewise.
8220 2017-09-12 Simon Wright <simon@pushface.org>
8223 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8224 calculation of the minor version, always output as 0.
8226 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8229 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8230 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8231 on it early, rather than manual conversion late. For
8232 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8233 instead of performing manual conversion.
8235 2017-09-12 Carl Love <cel@us.ibm.com>
8237 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8238 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8240 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8241 VMULOSW): Add definitions.
8242 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8243 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8244 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8245 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8246 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8248 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8250 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8252 (movti_aarch64): Likewise.
8253 (movdf_aarch64): Likewise.
8254 (movtf_aarch64): Likewise.
8255 (load_pairdi): Likewise.
8256 (store_pairdi): Likewise.
8257 (load_pairdf): Likewise.
8258 (store_pairdf): Likewise.
8259 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8260 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8261 (ldr_got_small_<mode>): Likewise.
8262 (ldr_got_small_28k_<mode>): Likewise.
8263 (ldr_got_tiny): Likewise.
8264 * config/aarch64/iterators.md (ldst_sz): New.
8265 (ldpstp_sz): Likewise.
8266 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8268 (thunderx_load): Split load_8 to load_16.
8269 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8271 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8272 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8273 (xgene1_store_pair): Split store_8 to store_16.
8274 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8275 (falkor_st_0_st_sd): Split store_8 to store_16.
8277 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8279 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8280 and store1/2/3/4 to store_4/8/12/16.
8281 * config/aarch64/aarch64.md: Update for rename.
8282 * config/arm/arm.md: Likewise.: Likewise.
8283 * config/arm/arm.c: Likewise.
8284 * config/arm/thumb1.md: Likewise.
8285 * config/arm/thumb2.md: Likewise.
8286 * config/arm/vfp.md: Likewise.
8287 * config/arm/arm-generic.md: Likewise.
8288 * config/arm/arm1020e.md: Likewise.
8289 * config/arm/arm1026ejs.md: Likewise.
8290 * config/arm/arm1136jfs.md: Likewise.
8291 * config/arm/arm926ejs.md: Likewise.
8292 * config/arm/cortex-a15.md: Likewise.
8293 * config/arm/cortex-a17.md: Likewise.
8294 * config/arm/cortex-a5.md: Likewise.
8295 * config/arm/cortex-a53.md: Likewise.
8296 * config/arm/cortex-a57.md: Likewise.
8297 * config/arm/cortex-a7.md: Likewise.
8298 * config/arm/cortex-a8.md: Likewise.
8299 * config/arm/cortex-a9.md: Likewise.
8300 * config/arm/cortex-m4.md: Likewise.
8301 * config/arm/cortex-m7.md: Likewise.
8302 * config/arm/cortex-r4.md: Likewise.
8303 * config/arm/exynos-m1.md: Likewise.
8304 * config/arm/fa526.md: Likewise.
8305 * config/arm/fa606te.md: Likewise.
8306 * config/arm/fa626te.md: Likewise.
8307 * config/arm/fa726te.md: Likewise.
8308 * config/arm/fmp626.md: Likewise.
8309 * config/arm/iwmmxt.md: Likewise.
8310 * config/arm/ldmstm.md: Likewise.
8311 * config/arm/marvell-pj4.md: Likewise.
8312 * config/arm/xgene1.md: Likewise.
8313 * config/aarch64/thunderx.md: Likewise.
8314 * config/aarch64/thunderx2t99.md: Likewise.
8315 * config/aarch64/falkor.md: Likewise.
8317 2017-09-12 Martin Liska <mliska@suse.cz>
8319 * attribs.c (private_lookup_attribute): New function.
8320 * attribs.h (private_lookup_attribute): Declared here.
8321 (lookup_attribute): Called from this place.
8323 2017-09-12 Richard Biener <rguenther@suse.de>
8325 PR tree-optimization/82157
8326 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8327 stmts with side-effects.
8329 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8330 Alan Hayward <alan.hayward@arm.com>
8331 David Sherwood <david.sherwood@arm.com>
8333 * target.def (hard_regno_nregs): New hook.
8334 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8335 * targhooks.h (default_hard_regno_nregs): Declare.
8336 * targhooks.c (default_hard_regno_nregs): New function.
8337 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8338 (TARGET_HARD_REGNO_NREGS): ...this hook.
8339 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8340 (CLASS_MAX_NREGS): Likewise.
8341 * doc/tm.texi: Regenerate.
8342 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8343 instead of HARD_REGNO_NREGS.
8344 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8345 HARD_REGNO_NREGS in the comment.
8346 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8347 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8348 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8349 Return an unsigned int.
8350 (TARGET_HARD_REGNO_NREGS): Redefine.
8351 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8352 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8353 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8354 (arc_hard_regno_nregs): New function.
8355 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8356 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8357 (arm_hard_regno_nregs): New function.
8358 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8359 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8360 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8361 (TARGET_HARD_REGNO_NREGS): Redefine.
8362 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8363 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8364 (HARD_REGNO_NREGS): Delete.
8365 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8366 (cr16_hard_regno_nregs): New function.
8367 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8368 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8369 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8370 (cris_hard_regno_nregs): New function.
8371 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8372 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8373 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8374 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8375 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8376 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8377 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8378 (frv_hard_regno_nregs): Make static. Take and return an
8380 (frv_class_max_nregs): Remove outdated copy of documentation.
8381 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8382 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8383 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8384 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8385 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8386 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8387 (TARGET_HARD_REGNO_NREGS): Redefine.
8388 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8389 (CLASS_MAX_NREGS): Update comment.
8390 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8391 (ia64_hard_regno_nregs): New function.
8392 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8393 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8394 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8395 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8396 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8398 (m32c_hard_regno_nregs): Likewise. Make static.
8399 (TARGET_HARD_REGNO_NREGS): Redefine.
8400 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8401 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8402 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8403 (m68k_hard_regno_nregs): New function.
8404 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8405 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8406 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8407 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8408 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8409 Take and return an unsigned int.
8410 (TARGET_HARD_REGNO_NREGS): Redefine.
8411 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8412 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8413 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8414 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8415 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8416 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8417 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8418 (msp430_hard_regno_nregs): Make static. Take and return an
8420 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8421 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8422 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8423 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8424 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8425 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8426 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8427 (TARGET_HARD_REGNO_NREGS): Redefine.
8428 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8429 (PA_HARD_REGNO_NREGS): ...this.
8430 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8431 (PA_HARD_REGNO_NREGS): ...this.
8432 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8433 (pa_hard_regno_nregs): New function.
8434 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8435 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8436 (pdp11_hard_regno_nregs): New function.
8437 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8438 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8439 (rs6000_hard_regno_nregs_hook): New function.
8440 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8441 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8442 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8443 Take and return an unsigned int. Move earlier in file.
8444 (TARGET_HARD_REGNO_NREGS): Redefine.
8445 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8446 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8447 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8448 (rl78_hard_regno_nregs): Make static. Take and return an
8450 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8451 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8452 (rs6000_hard_regno_nregs_hook): New function.
8453 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8454 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8455 (TARGET_HARD_REGNO_NREGS): Redefine.
8456 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8457 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8458 instead of HARD_REGNO_NREGS.
8459 (s390_hard_regno_nregs): New function.
8460 (s390_hard_regno_mode_ok): Add comment from s390.h.
8461 (TARGET_HARD_REGNO_NREGS): Redefine.
8462 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8463 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8464 (sh_hard_regno_nregs): New function.
8465 (sh_pass_in_reg_p): Use it.
8466 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8467 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8468 (sparc_hard_regno_nregs): New function.
8469 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8470 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8471 (spu_function_arg_advance): Use it, supplying a valid register number.
8472 (TARGET_HARD_REGNO_NREGS): Redefine.
8473 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8474 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8475 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8476 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8477 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8478 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8479 (CLASS_MAX_NREGS): Remove copy of old documentation.
8480 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8481 (visium_hard_regno_nregs): New function.
8482 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8483 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8484 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8485 xtensa_hard_regno_nregs): New function.
8486 * system.h (HARD_REGNO_NREGS): Poison.
8488 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8490 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8491 hard_regno_nregs instead of HARD_REGNO_NREGS.
8492 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8493 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8494 (c6x_expand_epilogue): Likewise.
8495 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8496 (frv_read_iacc_argument): Likewise.
8497 * config/sh/sh.c: Include regs.h.
8498 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8499 (regs_used): Likewise.
8500 (output_stack_adjust): Likewise.
8501 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8502 * expmed.c: Include regs.h.
8503 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8504 * ree.c: Include regs.h.
8505 (combine_reaching_defs): Use hard_regno_nregs instead of
8507 (add_removable_extension): Likewise.
8509 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8511 * regs.h (hard_regno_nregs): Turn into a function.
8512 (end_hard_regno): Update accordingly.
8513 * caller-save.c (setup_save_areas): Likewise.
8514 (save_call_clobbered_regs): Likewise.
8515 (replace_reg_with_saved_mem): Likewise.
8516 (insert_restore): Likewise.
8517 (insert_save): Likewise.
8518 * combine.c (can_change_dest_mode): Likewise.
8519 (move_deaths): Likewise.
8520 (distribute_notes): Likewise.
8521 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8522 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8523 (rs6000_split_multireg_move): Likewise.
8524 (rs6000_register_move_cost): Likewise.
8525 (rs6000_memory_move_cost): Likewise.
8526 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8527 (rs6000_split_multireg_move): Likewise.
8528 (rs6000_register_move_cost): Likewise.
8529 (rs6000_memory_move_cost): Likewise.
8530 * cselib.c (cselib_reset_table): Likewise.
8531 (cselib_lookup_1): Likewise.
8532 * emit-rtl.c (set_mode_and_regno): Likewise.
8533 * function.c (aggregate_value_p): Likewise.
8534 * ira-color.c (setup_profitable_hard_regs): Likewise.
8535 (check_hard_reg_p): Likewise.
8536 (calculate_saved_nregs): Likewise.
8537 (assign_hard_reg): Likewise.
8538 (improve_allocation): Likewise.
8539 (calculate_spill_cost): Likewise.
8540 * ira-emit.c (modify_move_list): Likewise.
8541 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8542 (ira_hard_reg_in_set_p): Likewise.
8543 * ira.c (setup_reg_mode_hard_regset): Likewise.
8544 (clarify_prohibited_class_mode_regs): Likewise.
8545 (check_allocation): Likewise.
8546 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8547 (lra_setup_reg_renumber): Likewise.
8548 (setup_try_hard_regno_pseudos): Likewise.
8549 (spill_for): Likewise.
8550 (assign_hard_regno): Likewise.
8551 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8552 * lra-constraints.c (in_class_p): Likewise.
8553 (lra_constraint_offset): Likewise.
8554 (simplify_operand_subreg): Likewise.
8555 (lra_constraints): Likewise.
8556 (split_reg): Likewise.
8557 (split_if_necessary): Likewise.
8558 (invariant_p): Likewise.
8559 (inherit_in_ebb): Likewise.
8560 * lra-lives.c (process_bb_lives): Likewise.
8561 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8562 (get_hard_regs): Likewise.
8563 (do_remat): Likewise.
8564 * lra-spills.c (assign_spill_hard_regs): Likewise.
8565 * mode-switching.c (create_pre_exit): Likewise.
8566 * postreload.c (reload_combine_recognize_pattern): Likewise.
8567 * recog.c (peep2_find_free_register): Likewise.
8568 * regcprop.c (kill_value_regno): Likewise.
8569 (set_value_regno): Likewise.
8570 (copy_value): Likewise.
8571 (maybe_mode_change): Likewise.
8572 (find_oldest_value_reg): Likewise.
8573 (copyprop_hardreg_forward_1): Likewise.
8574 * regrename.c (check_new_reg_p): Likewise.
8575 (regrename_do_replace): Likewise.
8576 * reload.c (push_reload): Likewise.
8577 (combine_reloads): Likewise.
8578 (find_dummy_reload): Likewise.
8579 (operands_match_p): Likewise.
8580 (find_reloads): Likewise.
8581 (find_equiv_reg): Likewise.
8582 (reload_adjust_reg_for_mode): Likewise.
8583 * reload1.c (count_pseudo): Likewise.
8584 (count_spilled_pseudo): Likewise.
8585 (find_reg): Likewise.
8586 (clear_reload_reg_in_use): Likewise.
8587 (free_for_value_p): Likewise.
8588 (allocate_reload_reg): Likewise.
8589 (choose_reload_regs): Likewise.
8590 (reload_adjust_reg_for_temp): Likewise.
8591 (emit_reload_insns): Likewise.
8592 (delete_output_reload): Likewise.
8593 * rtlanal.c (subreg_get_info): Likewise.
8594 * sched-deps.c (sched_analyze_reg): Likewise.
8595 * sel-sched.c (init_regs_for_mode): Likewise.
8596 (mark_unavailable_hard_regs): Likewise.
8597 (choose_best_reg_1): Likewise.
8598 (verify_target_availability): Likewise.
8599 * valtrack.c (dead_debug_insert_temp): Likewise.
8600 * var-tracking.c (track_loc_p): Likewise.
8601 (emit_note_insn_var_location): Likewise.
8602 * varasm.c (make_decl_rtl): Likewise.
8603 * reginfo.c (choose_hard_reg_mode): Likewise.
8604 (init_reg_modes_target): Refer directly to
8605 this_target_regs->x_hard_regno_nregs.
8607 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8609 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8610 instead of hard_regno_nregs.
8612 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8614 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8615 end_hard_regno instead of hard_regno_nregs.
8616 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8617 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8618 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8619 * ira-color.c (improve_allocation): Likewise.
8620 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8621 * lra-lives.c (mark_regno_live): Likewise.
8622 (mark_regno_dead): Likewise.
8623 * lra-remat.c (operand_to_remat): Likewise.
8624 * lra.c (collect_non_operand_hard_regs): Likewise.
8625 * postreload.c (reload_combine_note_store): Likewise.
8626 (move2add_valid_value_p): Likewise.
8627 * reload.c (regno_clobbered_p): Likewise.
8629 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8631 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8633 * config/v850/v850.c (v850_reorg): Likewise.
8634 * reload.c (refers_to_regno_for_reload_p): Likewise.
8635 (find_equiv_reg): Likewise.
8636 * reload1.c (reload_reg_reaches_end_p): Likewise.
8638 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8640 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8642 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8643 * config/arm/arm.c (output_move_neon): Likewise.
8644 (arm_attr_length_move_neon): Likewise.
8645 (neon_split_vcombine): Likewise.
8646 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8647 (c6x_mark_reg_written): Likewise.
8648 (c6x_dwarf_register_span): Likewise.
8649 * config/i386/i386.c (ix86_save_reg): Likewise.
8650 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8651 (rws_access_reg): Likewise.
8652 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8653 * mode-switching.c (create_pre_exit): Likewise.
8654 * ree.c (combine_reaching_defs): Likewise.
8655 (add_removable_extension): Likewise.
8656 * regcprop.c (find_oldest_value_reg): Likewise.
8657 (copyprop_hardreg_forward_1): Likewise.
8658 * reload.c (reload_inner_reg_of_subreg): Likewise.
8659 (push_reload): Likewise.
8660 (combine_reloads): Likewise.
8661 (find_dummy_reload): Likewise.
8662 (reload_adjust_reg_for_mode): Likewise.
8663 * reload1.c (find_reload_regs): Likewise.
8664 (forget_old_reloads_1): Likewise.
8665 (reload_reg_free_for_value_p): Likewise.
8666 (reload_adjust_reg_for_temp): Likewise.
8667 (emit_reload_insns): Likewise.
8668 (delete_output_reload): Likewise.
8669 * sel-sched.c (choose_best_reg_1): Likewise.
8670 (choose_best_pseudo_reg): Likewise.
8672 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8673 Alan Hayward <alan.hayward@arm.com>
8674 David Sherwood <david.sherwood@arm.com>
8676 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8677 * target.def (slow_unaligned_access): New hook.
8678 * targhooks.h (default_slow_unaligned_access): Declare.
8679 * targhooks.c (default_slow_unaligned_access): New function.
8680 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8681 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8682 * doc/tm.texi: Regenerate.
8683 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8684 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8685 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8687 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8688 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8690 (rs6000_slow_unaligned_access): New function.
8691 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8692 (expand_block_compare): Likewise.
8693 (expand_strn_compare): Likewise.
8694 (rs6000_rtx_costs): Likewise.
8695 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8696 (riscv_slow_unaligned_access): Likewise.
8697 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8698 (riscv_slow_unaligned_access_p): ...this and make static.
8699 (riscv_option_override): Update accordingly.
8700 (riscv_slow_unaligned_access): New function.
8701 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8702 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8703 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8704 (rs6000_slow_unaligned_access): New function.
8705 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8706 (rs6000_rtx_costs): Likewise.
8707 * config/rs6000/rs6000-string.c (expand_block_compare)
8708 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8709 of SLOW_UNALIGNED_ACCESS.
8710 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8711 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8712 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8713 of SLOW_UNALIGNED_ACCESS.
8714 * expmed.c (simple_mem_bitfield_p): Likewise.
8715 * expr.c (alignment_for_piecewise_move): Likewise.
8716 (emit_group_load_1): Likewise.
8717 (emit_group_store): Likewise.
8718 (copy_blkmode_from_reg): Likewise.
8719 (emit_push_insn): Likewise.
8720 (expand_assignment): Likewise.
8721 (store_field): Likewise.
8722 (expand_expr_real_1): Likewise.
8723 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8724 * lra-constraints.c (simplify_operand_subreg): Likewise.
8725 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8726 * gimple-ssa-store-merging.c: Likewise in block comment at start
8728 * tree-ssa-strlen.c: Include target.h.
8729 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8730 of SLOW_UNALIGNED_ACCESS.
8731 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8733 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8735 PR rtl-optimization/82185
8736 * expmed.c (emit_store_flag_int): Only test tem if it has been
8739 2017-09-12 Richard Biener <rguenther@suse.de>
8742 * match.pd ((FTYPE) N CMP CST): Fix typo.
8744 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8746 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8748 (mips_near_type_p): Add 'short_call' attribute as a synonym
8750 * doc/extend.texi (short_call): Document new function attribute.
8752 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8755 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8756 assertion check that in the condition.
8757 (get_atomic_generic_size): Likewise. Before testing if parameter
8758 has pointer type, if it has array type, call for C++
8759 default_conversion to perform array-to-pointer conversion.
8761 2017-09-12 Richard Biener <rguenther@suse.de>
8763 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8764 for operations we cannot scalarize.
8766 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8768 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8769 vectors heap vectors. Clean up comments.
8770 Make visited_bbs a reference.
8771 (profitable_jump_thread_path): Make GC
8772 vectors heap vectors. Clean up comments.
8774 (convert_and_register_jump_thread_path): Make GC vectors heap
8776 (check_subpath_and_update_thread_path): Same. Clean up comments.
8777 Make visited_bbs a reference.
8778 (handle_phi): Abstract common code to to
8779 register_jump_thread_path_if_profitable.
8780 Rename VAR_BB to DEF_BB.
8782 Make GC vectors heap vectors.
8783 Make visited_bbs a reference.
8784 (handle_assignment): Same.
8785 (register_jump_thread_path_if_profitable): New.
8786 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8788 Make GC vectors heap vectors. Clean up comments.
8789 Make visited_bbs a reference.
8790 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8791 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8794 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8797 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8798 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8800 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8802 Revert r251800 and r251799.
8804 2017-09-11 Martin Jambor <mjambor@suse.cz>
8807 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8808 arguments in advance.
8809 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8810 use it to find predecessor edges.
8811 (naive_outof_ssa): Collect vector of predecessors.
8813 2017-09-08 Jason Merrill <jason@redhat.com>
8815 PR c++/70029 - ICE with ref-qualifier and -flto
8816 * langhooks.h (struct lang_hooks_for_types): Add
8817 copy_lang_qualifiers.
8818 * attribs.c (build_type_attribute_qual_variant): Use it.
8819 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8821 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8822 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8824 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8827 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8828 (*mulsi3_sp64): New instruction.
8829 (mulsi3): New expander.
8831 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8833 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8835 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8837 * sancov.c: Include memmodel.h.
8839 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8842 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8845 2017-09-07 Carl Love <cel@us.ibm.com>
8847 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8848 the sldi instruction.
8850 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8852 * sancov.c: Include tm_p.h.
8854 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8857 * output.h (switch_to_other_text_partition): New declaration.
8858 * varasm.c (switch_to_other_text_partition): New function.
8859 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8860 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8861 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8862 to the other text partition before emitting LCL label and switch back
8863 after emitting the word after it.
8865 2017-09-07 Richard Biener <rguenther@suse.de>
8867 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8868 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8869 critical edges here, after loop init.
8870 (pass_data_pre): Remove PROP_no_crit_edges flags.
8871 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8872 for valueization of call args to avoid leaking VN_TOP.
8873 (visit_use): Assert we do not visit default defs.
8874 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8875 Use error_mark_node to more easily detect leaking VN_TOP.
8876 All default-defs are varying, not VN_TOP. Mark them visited.
8877 (run_scc_vn): Make code match comment.
8879 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8881 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8882 OPTION_MASK_FLOAT128_KEYWORD.
8883 (POWERPC_MASKS): Likewise.
8884 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8885 support for the -mfloat128-type option, and make -mfloat128
8886 default on PowerPC Linux systems. Define or undefine
8887 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8888 Define __float128 to be __ieee128 if IEEE 128-bit support is
8889 enabled, or undefine it.
8890 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8891 Delete defining __FLOAT128_TYPE__.
8892 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8893 -mfloat128-type option and make -mfloat128 default on PowerPC
8895 (TARGET_FLOAT128_TYPE): Likewise.
8896 (-mfloat128-type): Likewise.
8897 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8898 Delete the -mfloat128-type option and make -mfloat128 default on
8899 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8900 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8901 128-bit floating point is enabled. Change tests from using
8902 -mfloat128-type to -mfloat128.
8903 (rs6000_mangle_type): Use the correct mangling for the __float128
8904 type even if normal long double is restricted to 64-bits.
8905 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8907 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8908 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8909 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8911 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8912 documentation for -mfloat128.
8914 2017-09-06 Olivier Hainque <hainque@adacore.com>
8916 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8918 2017-09-06 Wish Wu <wishwu007@gmail.com>
8919 Jakub Jelinek <jakub@redhat.com>
8921 * asan.c (initialize_sanitizer_builtins): Add
8922 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8923 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8924 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8925 BT_FN_VOID_UINT64_PTR variables.
8926 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8927 (BT_FN_VOID_UINT16_UINT16): Likewise.
8928 (BT_FN_VOID_UINT32_UINT32): Likewise.
8929 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8930 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8931 (BT_FN_VOID_UINT64_PTR): Likewise.
8932 * common.opt (flag_sanitize_coverage): New variable.
8933 (fsanitize-coverage=trace-pc): Remove.
8934 (fsanitize-coverage=): Add.
8935 * flag-types.h (enum sanitize_coverage_code): New enum.
8936 * fold-const.c (fold_range_test): Disable non-short-circuit
8937 optimization if flag_sanitize_coverage.
8938 (fold_truth_andor): Likewise.
8939 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8940 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8941 (coverage_sanitizer_opts): New array.
8942 (get_closest_sanitizer_option): Add OPTS argument, handle also
8943 OPT_fsanitize_coverage_.
8944 (parse_sanitizer_options): Adjusted to also handle
8945 OPT_fsanitize_coverage_.
8946 (common_handle_option): Add OPT_fsanitize_coverage_.
8947 * sancov.c (instrument_comparison, instrument_switch): New function.
8948 (sancov_pass): Add trace-cmp support.
8949 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8950 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8951 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8952 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8953 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8954 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8955 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8956 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8957 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8958 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8960 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8962 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8963 error. Only quit immediately if parsing is complete.
8964 (BEGIN): Initialize fatal_err and parse_done.
8965 (begin fpu, end fpu): Check number of arguments.
8966 (begin arch, end arch): Likewise.
8967 (begin cpu, end cpu): Likewise.
8968 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8969 (optalias): Likewise.
8971 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8973 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8974 * config/arm/arm-isa.h: Delete. Move definitions to ...
8975 * arm-cpus.in: ... here. Use new feature and fgroup values.
8976 * config/arm/arm.c (arm_option_override): Use lower case for feature
8978 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8979 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8980 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8982 (print_isa_bits_for): New function.
8983 (gen_isa): New function.
8984 (gen_comm_data): Use print_isa_bits_for.
8985 (define feature): New keyword.
8986 (define fgroup): New keyword.
8987 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8988 (arm-isa.h): Add rule to generate file.
8989 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8990 case for feature bit names.
8992 2017-09-06 Richard Biener <rguenther@suse.de>
8994 * tree-ssa-pre.c (NECESSARY): Remove.
8995 (create_expression_by_pieces): Do not touch pass-local flags.
8996 (insert_into_preds_of_block): Likewise.
8997 (do_pre_regular_insertion): Likewise.
8998 (eliminate_insert): Likewise.
8999 (eliminate_dom_walker::before_dom_children): Likewise.
9000 (fini_eliminate): Do not look at inserted_exprs.
9001 (mark_operand_necessary): Remove.
9002 (remove_dead_inserted_code): Replace with simple work-list
9003 algorithm based on inserted_exprs and SSA uses.
9004 (pass_pre::execute): Re-order fini_eliminate and
9005 remove_dead_inserted_code.
9007 2017-09-06 Olivier Hainque <hainque@adacore.com>
9009 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9010 for VxWorks 7. Adjust surrounding comments.
9012 2017-09-06 Richard Biener <rguenther@suse.de>
9014 * gimple-ssa-strength-reduction.c
9015 (find_candidates_dom_walker::before_dom_children): Also allow
9018 2017-09-06 Richard Biener <rguenther@suse.de>
9020 PR tree-optimization/82108
9021 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9022 for gap in the non-permutation SLP case.
9024 2017-09-06 Martin Jambor <mjambor@suse.cz>
9026 PR tree-optimization/82078
9027 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9028 add_access_to_work_queue...
9029 (build_accesses_from_assign): ...here.
9030 (propagate_all_subaccesses): Make sure racc is the group
9031 representative, if there is one.
9033 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9036 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9039 2017-09-06 Richard Biener <rguenther@suse.de>
9041 * gimple-ssa-strength-reduction.c
9042 (find_candidates_dom_walker::before_doom_children): Use a
9043 type and not a mode check.
9045 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9048 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9050 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9052 2017-09-05 Jeff Law <law@redhat.com>
9054 PR tree-optimization/64910
9055 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9056 cases where we have 3 or more operands.
9058 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9061 * omp-low.c (lower_omp_for): Recompute tree invariant if
9062 gimple_omp_for_initial/final is ADDR_EXPR.
9065 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9066 into gimple val before gimplification fo the COND_EXPR.
9068 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9070 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9071 REGION_COPY argument.
9072 (thread_through_all_blocks): Remove unused argument to
9073 duplicate_thread_path.
9075 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9076 Alan Hayward <alan.hayward@arm.com>
9077 David Sherwood <david.sherwood@arm.com>
9079 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9080 Take a scalar_mode rather than a machine_mode.
9081 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9082 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9083 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9084 (aarch64_gen_adjusted_ldpstp): Likewise.
9085 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9087 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9088 Alan Hayward <alan.hayward@arm.com>
9089 David Sherwood <david.sherwood@arm.com>
9091 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9092 Take a scalar_int_mode instead of a machine_mode.
9093 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9094 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9095 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9096 (aarch64_simd_attr_length_rglist): Delete.
9097 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9098 a scalar_int_mode instead of a machine_mode.
9099 (aarch64_add_offset): Likewise.
9100 (aarch64_internal_mov_immediate): Likewise
9101 (aarch64_add_constant_internal): Likewise.
9102 (aarch64_add_constant): Likewise.
9103 (aarch64_movw_imm): Likewise.
9104 (aarch64_rtx_arith_op_extract_p): Likewise.
9105 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9106 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9107 Remove assert that the mode isn't a vector.
9108 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9109 (aarch64_expand_mov_immediate): Update calls after above changes.
9110 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9111 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9112 (aarch64_move_imm): Likewise.
9113 (aarch64_can_const_movi_rtx_p): Likewise.
9114 (aarch64_strip_extend): Likewise.
9115 (aarch64_extr_rtx_p): Likewise.
9116 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9117 a CONST_INT when the mode parameter is VOIDmode.
9118 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9120 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9122 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9123 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9124 (bitwise_type_for_mode): Update accordingly.
9126 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9128 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9129 * stor-layout.c (mode_for_size_tree): Likewise.
9130 (mode_for_array): Update accordingly.
9131 (layout_decl): Likewise.
9132 (compute_record_mode): Likewise. Only set the mode once.
9134 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9136 * target.def (get_mask_mode): Change return type to opt_mode.
9138 * doc/tm.texi: Regenerate.
9139 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9140 * targhooks.c (default_get_mask_mode): Likewise.
9141 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9142 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9143 targetm.get_mask_mode.
9144 * tree.c (build_truth_vector_type): Likewise.
9146 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9148 * machmode.h (mode_for_vector): Return an opt_mode.
9149 * stor-layout.c (mode_for_vector): Likewise.
9150 (mode_for_int_vector): Update accordingly.
9151 (layout_type): Likewise.
9152 * config/i386/i386.c (emit_memmov): Likewise.
9153 (ix86_expand_set_or_movmem): Likewise.
9154 (ix86_expand_vector_init): Likewise.
9155 (ix86_get_mask_mode): Likewise.
9156 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9158 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9159 * expmed.c (extract_bit_field_1): Likewise.
9160 * expr.c (expand_expr_real_2): Likewise.
9161 * optabs-query.c (can_vec_perm_p): Likewise.
9162 (can_vec_mask_load_store_p): Likewise.
9163 * optabs.c (expand_vec_perm): Likewise.
9164 * targhooks.c (default_get_mask_mode): Likewise.
9165 * tree-vect-stmts.c (vectorizable_store): Likewise.
9166 (vectorizable_load): Likewise.
9167 (get_vectype_for_scalar_type_and_size): Likewise.
9169 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9171 * machmode.h (mode_for_int_vector): New function.
9172 * stor-layout.c (mode_for_int_vector): Likewise.
9173 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9174 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9175 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9176 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9177 (s390_expand_vcond): Likewise.
9179 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9181 * machmode.h (opt_machine_mode): New type.
9182 (opt_mode<T>): Allow construction from anything that can be
9184 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9185 (mode_for_size): Return an opt_machine_mode.
9186 * stor-layout.c (mode_for_size): Likewise.
9187 (mode_for_size_tree): Update call accordingly.
9188 (bitwise_mode_for_mode): Likewise.
9189 (make_fract_type): Likewise.
9190 (make_accum_type): Likewise.
9191 * caller-save.c (replace_reg_with_saved_mem): Update call
9193 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9194 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9195 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9196 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9197 * expmed.c (extract_bit_field_1): Likewise.
9198 * reload.c (get_secondary_mem): Likewise.
9199 * varasm.c (assemble_integer): Likewise.
9200 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9203 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9205 * machmode.h (decimal_float_mode_for_size): New function.
9206 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9207 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9208 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9209 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9210 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9211 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9213 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9215 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9216 (get_builtin_sync_mode): Likewise.
9217 (expand_ifn_atomic_compare_exchange): Likewise.
9218 (expand_builtin_atomic_clear): Likewise.
9219 (expand_builtin_atomic_test_and_set): Likewise.
9220 (fold_builtin_atomic_always_lock_free): Likewise.
9221 * calls.c (compute_argument_addresses): Likewise.
9222 (emit_library_call_value_1): Likewise.
9223 (store_one_arg): Likewise.
9224 * combine.c (combine_instructions): Likewise.
9225 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9226 * config/arm/arm.c (arm_function_value): Likewise.
9227 (aapcs_allocate_return_reg): Likewise.
9228 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9229 * config/i386/i386.c (construct_container): Likewise.
9230 (ix86_gimplify_va_arg): Likewise.
9231 (ix86_expand_sse_cmp): Likewise.
9232 (emit_memmov): Likewise.
9233 (emit_memset): Likewise.
9234 (expand_small_movmem_or_setmem): Likewise.
9235 (ix86_expand_pextr): Likewise.
9236 (ix86_expand_pinsr): Likewise.
9237 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9238 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9240 * config/mips/mips.c (mips_function_value_1) Likewise.
9241 (mips_block_move_straight): Likewise.
9242 (mips_expand_ins_as_unaligned_store): Likewise.
9243 * config/powerpcspe/powerpcspe.c
9244 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9245 (rs6000_darwin64_record_arg_flush): Likewise.
9246 * config/rs6000/rs6000.c
9247 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9248 (rs6000_darwin64_record_arg_flush): Likewise.
9249 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9250 (sparc_function_value_1): Likewise.
9251 * config/spu/spu.c (adjust_operand): Likewise.
9252 (spu_emit_branch_or_set): Likewise.
9253 (arith_immediate_p): Likewise.
9254 * emit-rtl.c (gen_lowpart_common): Likewise.
9255 * expr.c (expand_expr_real_1): Likewise.
9256 * function.c (assign_parm_setup_block): Likewise.
9257 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9258 * reload1.c (alter_reg): Likewise.
9259 * stor-layout.c (mode_for_vector): Likewise.
9260 (layout_type): Likewise.
9262 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9264 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9265 (spu_convert_move): Likewise.
9266 * lower-subreg.c (resolve_simple_move): Likewise.
9268 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9271 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9272 define_insn to a define_expand.
9273 (altivec_vsum2sws_direct): New define_insn.
9274 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9276 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9278 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9281 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9286 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9287 realign stack if stack alignment needed is less than incoming
9290 2017-09-05 Marek Polacek <polacek@redhat.com>
9293 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9296 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9298 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9300 2017-09-05 Richard Biener <rguenther@suse.de>
9302 PR tree-optimization/82084
9303 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9305 2017-09-05 Richard Biener <rguenther@suse.de>
9307 PR tree-optimization/82102
9308 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9310 2017-09-05 Martin Liska <mliska@suse.cz>
9312 PR tree-optimization/82032
9313 * tree-cfg.c (generate_range_test): New function.
9314 * tree-cfg.h (generate_range_test): Declared here.
9315 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9316 (cleanup_control_expr_graph): Use it.
9317 * tree-switch-conversion.c (try_switch_expansion): Remove
9319 (emit_case_nodes): Use generate_range_test.
9321 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9324 * config/i386/i386.md (*<btsc><mode>_mask): Add
9325 TARGET_USE_BT to insn constraint.
9326 (*btr<mode>_mask): Ditto.
9328 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9330 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9331 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9333 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9336 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9337 TARGET_NEON and TARGET_IWMMXT.
9338 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9339 TARGET_NEON and TARGET_IWMMXT.
9340 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9342 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9344 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9345 (ix86_rewrite_tls_address): Ditto.
9346 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9347 (ix86_rewrite_tls_address_1): Ditto.
9348 (ix86_rewrite_tls_address): Ditto.
9349 * config/i386/predicates.md (tls_address_pattern): New predicate.
9350 * config/i386/i386.md (TLS address splitter): New splitter.
9352 2017-09-04 Richard Biener <rguenther@suse.de>
9354 PR tree-optimization/82084
9355 * fold-const.h (can_native_encode_string_p): Declare.
9356 * fold-const.c (can_native_encode_string_p): Factor out from ...
9357 (native_encode_string): ... here.
9358 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9359 vectorizing stores from constants we later cannot handle.
9361 2017-09-04 Marek Polacek <polacek@redhat.com>
9364 * doc/invoke.texi: Update -Wtautological-compare documentation.
9366 2017-09-04 Jeff Law <law@redhat.com>
9368 PR tree-optimization/64910
9369 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9370 swap the first and last operand if the last is a constant.
9372 2017-09-04 Marek Polacek <polacek@redhat.com>
9375 * convert.c (do_narrow): When sanitizing signed integer overflows,
9376 bail out for signed types.
9377 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9379 2017-09-04 Richard Biener <rguenther@suse.de>
9381 PR tree-optimization/82060
9382 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9383 Move devirtualization after stmt folding and before EH/AB/noreturn
9384 cleanup to get the stmt refs canonicalized. Use a bool instead
9385 of gimple_modified_p since that doesn't work for NOPs. Schedule
9386 NOPs generated by folding for removal.
9388 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9389 Alan Hayward <alan.hayward@arm.com>
9390 David Sherwood <david.sherwood@arm.com>
9392 * coretypes.h (pad_direction): New enum.
9393 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9394 (FUNCTION_ARG_PADDING): Likewise.
9395 * target.def (function_arg_padding): New hook.
9396 * targhooks.h (default_function_arg_padding): Declare.
9397 * targhooks.c (default_function_arg_padding): New function.
9398 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9399 (TARGET_FUNCTION_ARG_PADDING): ...this.
9400 * doc/tm.texi: Regenerate.
9401 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9402 instead of direction.
9403 (compute_argument_addresses): Likewise.
9404 (load_register_parameters): Likewise.
9405 (emit_library_call_value_1): Likewise.
9406 (store_one_arg): Use targetm.calls.function_arg_padding instead
9407 of FUNCTION_ARG_PADDING.
9408 (must_pass_in_stack_var_size_or_pad): Likewise.
9409 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9410 (emit_group_store): Likewise.
9411 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9412 instead of FUNCTION_ARG_PADDING.
9413 (emit_push_insn): Likewise, and propagate enum change throughout
9415 * function.h (direction): Delete.
9416 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9418 * function.c (assign_parm_find_stack_rtl): Likewise.
9419 (assign_parm_setup_block_p): Likewise.
9420 (assign_parm_setup_block): Likewise.
9421 (gimplify_parameters): Likewise.
9422 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9423 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9425 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9426 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9427 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9428 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9429 (aarch64_function_arg_padding): ...this new function.
9430 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9431 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9432 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9433 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9434 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9435 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9436 (arm_pad_arg_upward): Replace with...
9437 (arm_function_arg_padding): ...this new function.
9438 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9440 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9441 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9442 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9443 (ia64_hpux_function_arg_padding): Replace with...
9444 (ia64_function_arg_padding): ...this new function. Use pad_direction
9445 instead of direction. Check for TARGET_HPUX.
9446 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9447 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9448 (iq2000_function_arg_padding): New function.
9449 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9450 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9451 (mips_function_arg_padding): ...this new function.
9452 (mips_pad_reg_upward): Update accordingly.
9453 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9454 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9455 targetm.calls.function_arg_padding.
9456 (FUNCTION_ARG_PADDING): Delete.
9457 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9458 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9459 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9460 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9461 (nios2_block_reg_padding): Return pad_direction instead of direction.
9462 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9463 instead of direction.
9464 (nios2_function_arg_padding): Likewise. Make static.
9465 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9466 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9467 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9468 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9469 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9470 (pa_function_arg_padding): Make static. Return pad_direction instead
9472 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9473 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9474 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9475 instead of direction. Use targetm.calls.function_arg_padding.
9476 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9477 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9478 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9479 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9480 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9482 (function_arg_padding): Rename to...
9483 (rs6000_function_arg_padding): ...this. Make static. Return
9484 pad_direction instead of direction.
9485 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9486 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9487 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9488 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9489 instead of direction. Use targetm.calls.function_arg_padding.
9490 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9491 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9492 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9493 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9494 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9495 (function_arg_padding): Rename to...
9496 (rs6000_function_arg_padding): ...this. Make static. Return
9497 pad_direction instead of direction.
9498 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9499 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9500 * config/s390/s390.c (s390_function_arg_padding): New function.
9501 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9502 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9503 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9504 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9505 (function_arg_padding): Rename to...
9506 (sparc_function_arg_padding): ...this. Make static. Return
9507 pad_direction instead of direction.
9508 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9509 * config/spu/spu.c (spu_function_arg_padding): New function.
9510 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9511 * system.h (FUNCTION_ARG_PADDING): Poison.
9513 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9514 Alan Hayward <alan.hayward@arm.com>
9515 David Sherwood <david.sherwood@arm.com>
9517 * target.def (modes_tieable_p): New hook.
9518 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9519 (TARGET_MODES_TIEABLE_P): ...this.
9520 * doc/tm.texi.in: Regenerate.
9521 * hooks.h (hook_bool_mode_mode_true): Declare.
9522 * hooks.c (hook_bool_mode_mode_true): New function.
9523 * combine.c (subst): Use targetm.modes_tieable_p instead of
9525 * dse.c (find_shift_sequence): Likewise.
9526 * expmed.c (extract_low_bits): Likewise.
9527 * lower-subreg.c: Include target.h.
9528 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9530 * rtlanal.c (rtx_cost): Likewise.
9531 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9532 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9533 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9534 (TARGET_MODES_TIEABLE_P): Redefine.
9535 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9536 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9537 (TARGET_MODES_TIEABLE_P): Redefine.
9538 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9539 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9540 (arc_modes_tieable_p): New function.
9541 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9542 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9543 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9544 (arm_modes_tieable_p): Make static.
9545 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9546 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9547 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9548 (TARGET_MODES_TIEABLE_P): Redefine.
9549 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9550 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9551 (TARGET_MODES_TIEABLE_P): Redefine.
9552 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9553 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9554 (cr16_modes_tieable_p): New function.
9555 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9556 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9557 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9558 (TRULY_NOOP_TRUNCATION): Update comment.
9559 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9560 (TRULY_NOOP_TRUNCATION): Update comment.
9561 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9562 (frv_modes_tieable_p): New function.
9563 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9564 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9565 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9566 (TARGET_MODES_TIEABLE_P): Redefine.
9567 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9568 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9569 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9570 (TARGET_MODES_TIEABLE_P): Redefine.
9571 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9572 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9573 (ia64_modes_tieable_p): New function.
9574 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9575 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9576 (iq2000_modes_tieable_p): New function.
9577 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9578 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9579 (lm32_modes_tieable_p): New function.
9580 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9581 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9582 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9583 (TARGET_MODES_TIEABLE_P): Redefine.
9584 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9585 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9586 (m32r_modes_tieable_p): New function.
9587 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9588 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9589 (m68k_modes_tieable_p): New function.
9590 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9591 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9592 (mcore_modes_tieable_p): New function.
9593 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9594 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9596 (TARGET_MODES_TIEABLE_P): Redefine.
9597 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9598 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9599 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9600 (TARGET_MODES_TIEABLE_P): Redefine.
9601 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9602 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9603 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9604 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9605 (mn10300_modes_tieable_p): ...this and make static.
9606 (TARGET_MODES_TIEABLE_P): Redefine.
9607 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9608 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9609 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9610 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9611 (msp430_modes_tieable_p): Make static.
9612 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9613 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9614 (TARGET_MODES_TIEABLE_P): Redefine.
9615 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9616 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9617 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9618 (TARGET_MODES_TIEABLE_P): Redefine.
9619 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9620 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9621 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9622 (TARGET_MODES_TIEABLE_P): Redefine.
9623 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9624 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9625 (pdp11_modes_tieable_p): New function.
9626 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9627 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9628 (rs6000_modes_tieable_p): New function.
9629 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9630 * config/powerpcspe/powerpcspe.md: Update comment.
9631 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9632 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9633 (TARGET_MODES_TIEABLE_P): Redefine.
9634 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9635 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9636 (rl78_modes_tieable_p): New function.
9637 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9638 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9639 (rs6000_modes_tieable_p): New function.
9640 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9641 * config/rs6000/rs6000.md: Update comment.
9642 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9643 * config/rx/rx.c (rx_modes_tieable_p): New function.
9644 (TARGET_MODES_TIEABLE_P): Redefine.
9645 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9646 * config/s390/s390.c (s390_modes_tieable_p): New function.
9647 (TARGET_MODES_TIEABLE_P): Redefine.
9648 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9649 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9650 (sh_modes_tieable_p): New function.
9651 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9652 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9653 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9654 (sparc_modes_tieable_p): Make static.
9655 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9656 * config/spu/spu.c (spu_modes_tieable_p): New function.
9657 (TARGET_MODES_TIEABLE_P): Redefine.
9658 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9659 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9660 (TARGET_MODES_TIEABLE_P): Redefine.
9661 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9662 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9663 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9664 * config/v850/v850.c (v850_modes_tieable_p): New function.
9665 (TARGET_MODES_TIEABLE_P): Redefine.
9666 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9667 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9668 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9669 (visium_modes_tieable_p): New function.
9670 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9671 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9672 (xtensa_modes_tieable_p): New function.
9673 * system.h (MODES_TIEABLE_P): Poison.
9675 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9676 Alan Hayward <alan.hayward@arm.com>
9677 David Sherwood <david.sherwood@arm.com>
9679 * target.def (hard_regno_mode_ok): New hook.
9680 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9681 (TARGET_HARD_REGNO_MODE_OK): ...this.
9682 * doc/tm.texi.in: Regenerate.
9683 * hooks.h (hook_bool_uint_mode_true): Declare.
9684 * hooks.c (hook_bool_uint_mode_true): New function.
9685 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9687 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9688 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9689 instead of HARD_REGNO_MODE_OK.
9690 * caller-save.c: Include target.h.
9691 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9693 * combine.c (can_combine_p): Likewise.
9694 (combinable_i3pat): Likewise.
9695 (can_change_dest_mode): Likewise.
9696 * expr.c (init_expr_target): Likewise.
9697 (convert_move): Likewise.
9698 (convert_modes): Likewise.
9699 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9700 (setup_prohibited_mode_move_regs): Likewise.
9701 * ira.h (target_ira): Likewise.
9702 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9703 * lra-constraints.c (process_alt_operands): Likewise.
9704 (split_reg): Likewise.
9705 * recog.c (peep2_find_free_register): Likewise.
9706 * ree.c (combine_reaching_defs): Likewise.
9707 * regcprop.c (maybe_mode_change): Likewise.
9708 * reginfo.c (init_reg_sets_1): Likewise.
9709 (choose_hard_reg_mode): Likewise.
9710 (simplifiable_subregs): Likewise.
9711 * regrename.c (check_new_reg_p): Likewise.
9712 * reload.c (find_valid_class): Likewise.
9713 (find_valid_class_1): Likewise.
9714 (reload_inner_reg_of_subreg): Likewise.
9715 (push_reload): Likewise.
9716 (combine_reloads): Likewise.
9717 (find_dummy_reload): Likewise.
9718 (find_reloads): Likewise.
9719 * reload1.c (find_reg): Likewise.
9720 (set_reload_reg): Likewise.
9721 (allocate_reload_reg): Likewise.
9722 (choose_reload_regs): Likewise.
9723 (reload_adjust_reg_for_temp): Likewise.
9724 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9725 (simplify_subreg_regno): Likewise.
9726 * sel-sched.c (init_regs_for_mode): Likewise.
9727 * varasm.c (make_decl_rtl): Likewise.
9728 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9729 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9731 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9732 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9733 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9734 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9735 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9736 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9737 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9738 (arc_mode_class): Delete.
9739 (HARD_REGNO_MODE_OK): Delete.
9740 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9741 (arc_hard_regno_mode_ok): Rename old array to...
9742 (arc_hard_regno_mode_ok_modes): ...this.
9743 (arc_conditional_register_usage): Update accordingly.
9744 (arc_mode_class): Make static.
9745 (arc_hard_regno_mode_ok): New function.
9746 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9747 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9748 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9749 (arm_hard_regno_mode_ok): Make static.
9750 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9752 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9753 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9754 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9756 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9757 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9758 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9759 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9760 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9761 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9762 * config/bfin/predicates.md (valid_reg_operand): Use
9763 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9764 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9765 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9766 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9767 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9768 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9769 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9770 (cr16_hard_regno_mode_ok): Make static and return a bool.
9771 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9772 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9773 (cris_hard_regno_mode_ok): New function.
9774 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9775 (epiphany_mode_class): Delete.
9776 (HARD_REGNO_MODE_OK): Delete.
9777 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9778 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9779 (hard_regno_mode_ok): Rename to...
9780 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9781 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9782 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9784 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9785 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9786 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9787 (frv_hard_regno_mode_ok): Make static and return a bool.
9788 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9790 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9791 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9792 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9793 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9795 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9796 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9797 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9798 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9800 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9801 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9802 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9803 (ia64_hard_regno_mode_ok): New function.
9804 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9805 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9806 (iq2000_hard_regno_mode_ok): New function.
9807 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9808 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9809 (lm32_hard_regno_mode_ok): New function.
9810 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9811 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9812 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9813 instead of HARD_REGNO_MODE_OK.
9814 (m32c_hard_regno_ok): Rename to...
9815 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9816 (m32c_cannot_change_mode_class): Update accordingly.
9817 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9818 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9819 (m32r_mode_class): Delete.
9820 (HARD_REGNO_MODE_OK): Delete.
9821 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9822 (m32r_hard_regno_mode_ok): Rename to...
9823 (m32r_hard_regno_modes): ...this.
9824 (m32r_mode_class): Make static.
9825 (m32r_hard_regno_mode_ok): New function.
9826 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9827 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9828 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9829 (m68k_hard_regno_mode_ok): Make static.
9830 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9831 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9832 (mcore_hard_regno_mode_ok): New function.
9833 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9834 (HARD_REGNO_MODE_OK): Delete.
9835 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9837 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9838 (microblaze_hard_regno_mode_ok): New function.
9839 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9840 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9841 (mips_hard_regno_mode_ok): Delete.
9842 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9843 (mips_hard_regno_mode_ok_p): ...this and make static.
9844 (mips_hard_regno_mode_ok_p): Rename to...
9845 (mips_hard_regno_mode_ok_uncached): ...this.
9846 (mips_hard_regno_mode_ok): New function.
9847 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9848 of HARD_REGNO_MODE_OK.
9849 (mips_option_override): Update after above name changes.
9850 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9851 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9852 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9853 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9854 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9855 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9856 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9857 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9858 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9859 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9860 (msp430_hard_regno_mode_ok): Make static and return a bool.
9861 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9862 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9863 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9865 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9866 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9867 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9868 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9869 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9870 (PA_HARD_REGNO_MODE_OK): ...this
9871 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9872 (PA_HARD_REGNO_MODE_OK): ...this.
9873 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9874 (pa_hard_regno_mode_ok): New function.
9875 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9876 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9877 (pdp11_hard_regno_mode_ok): New function.
9878 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9879 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9881 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9883 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9884 (rs6000_hard_regno_mode_ok): Rename to...
9885 (rs6000_hard_regno_mode_ok_uncached): ...this.
9886 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9887 (rs6000_hard_regno_mode_ok): New function.
9888 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9889 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9890 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9891 (riscv_hard_regno_mode_ok): ...this and make static.
9892 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9893 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9894 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9895 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9896 (rl78_hard_regno_mode_ok): Make static and return bool.
9897 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9898 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9900 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9901 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9902 (rs6000_hard_regno_mode_ok): Rename to...
9903 (rs6000_hard_regno_mode_ok_uncached): ...this.
9904 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9905 (rs6000_hard_regno_mode_ok): New function.
9906 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9907 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9908 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9909 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9910 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9911 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9912 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9913 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9914 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9915 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9916 (sh_hard_regno_mode_ok): Make static.
9917 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9918 instead of HARD_REGNO_MODE_OK.
9919 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9920 (sparc_mode_class): Delete.
9921 (HARD_REGNO_MODE_OK): Delete.
9922 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9923 (hard_regno_mode_classes): Make static.
9924 (sparc_mode_class): Likewise.
9925 (sparc_hard_regno_mode_ok): New function.
9926 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9927 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9928 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9930 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9931 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9932 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9933 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9934 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9935 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9936 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9937 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9938 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9939 (visium_hard_regno_mode_ok): New function.
9940 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9941 instead of HARD_REGNO_MODE_OK.
9942 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9943 (HARD_REGNO_MODE_OK): Delete.
9944 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9945 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9946 (xtensa_option_override): Update accordingly.
9947 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9948 (xtensa_hard_regno_mode_ok): New function.
9949 * system.h (HARD_REGNO_MODE_OK): Poison.
9951 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9952 Alan Hayward <alan.hayward@arm.com>
9953 David Sherwood <david.sherwood@arm.com>
9955 * target.def (hard_regno_call_part_clobbered): New hook.
9956 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9957 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9958 * doc/tm.texi: Regenerate.
9959 * hooks.h (hook_bool_uint_mode_false): Declare.
9960 * hooks.c (hook_bool_uint_mode_false): New function.
9961 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9962 * cselib.c (cselib_process_insn): Use
9963 targetm.hard_regno_call_part_clobbered instead of
9964 HARD_REGNO_CALL_PART_CLOBBERED.
9965 * ira-conflicts.c (ira_build_conflicts): Likewise.
9966 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9967 * lra-constraints.c (need_for_call_save_p): Likewise.
9968 * lra-lives.c: Include target.h.
9969 (check_pseudos_live_through_calls): Use
9970 targetm.hard_regno_call_part_clobbered instead of
9971 HARD_REGNO_CALL_PART_CLOBBERED.
9972 * regcprop.c: Include target.h.
9973 (copyprop_hardreg_forward_1): Use
9974 targetm.hard_regno_call_part_clobbered instead of
9975 HARD_REGNO_CALL_PART_CLOBBERED.
9976 * reginfo.c (choose_hard_reg_mode): Likewise.
9977 * regrename.c (check_new_reg_p): Likewise.
9978 * reload.c (find_equiv_reg): Likewise.
9979 * reload1.c (emit_reload_insns): Likewise.
9980 * sched-deps.c (deps_analyze_insn): Likewise.
9981 * sel-sched.c (init_regs_for_mode): Likewise.
9982 (mark_unavailable_hard_regs): Likewise.
9983 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9984 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9985 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9987 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9988 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9989 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9991 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9993 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9994 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9995 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9997 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9998 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9999 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10001 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10002 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10004 * config/powerpcspe/powerpcspe.c
10005 (rs6000_hard_regno_call_part_clobbered): New function.
10006 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10007 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10008 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10010 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10011 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10012 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10014 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10015 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10016 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10018 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10019 Alan Hayward <alan.hayward@arm.com>
10020 David Sherwood <david.sherwood@arm.com>
10022 * rtl.h (subreg_memory_offset): Declare.
10023 * emit-rtl.c (subreg_memory_offset): New function.
10024 * expmed.c (store_bit_field_1): Use it.
10025 * expr.c (undefined_operand_subword_p): Likewise.
10026 * simplify-rtx.c (simplify_subreg): Likewise.
10028 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10030 PR rtl-optimization/57448
10033 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10034 using atomic_load pattern.
10035 (expand_atomic_store): Likewise.
10037 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10040 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10041 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10042 handling. Use replace_call_with_value with NULL instead of
10043 gsi_replace, unlink_stmt_vdef and release_defs.
10045 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10048 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10050 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10053 * rtl.h (emit_library_call_value_1): Declare.
10054 (emit_library_call): Replace declaration with a series of overloads.
10055 Remove the parameter count argument.
10056 (emit_library_call_value): Likewise.
10057 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10058 with an "rtx_mode_t *".
10059 (emit_library_call_value): Delete.
10060 (emit_library_call): Likewise.
10061 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10062 (asan_emit_allocas_unpoison): Likewise.
10063 * builtins.c (expand_builtin_powi): Likewise.
10064 (expand_asan_emit_allocas_unpoison): Likewise.
10065 * cfgexpand.c (expand_main_function): Likewise.
10066 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10067 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10068 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10069 * config/arm/arm.c (arm_trampoline_init): Likewise.
10070 (arm_call_tls_get_addr): Likewise.
10071 (arm_expand_divmod_libfunc): Likewise.
10072 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10073 (smulsi3_highpart): Likewise.
10074 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10075 (c6x_expand_compare): Likewise.
10076 (c6x_expand_movmem): Likewise.
10077 * config/frv/frv.c (frv_trampoline_init): Likewise.
10078 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10079 (ix86_expand_divmod_libfunc): Likewise.
10080 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10081 (ia64_expand_compare): Likewise.
10082 (ia64_profile_hook): Likewise.
10083 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10084 (nonlocal_goto): Likewise.
10085 (restore_stack_nonlocal): Likewise.
10086 * config/m32r/m32r.c (block_move_call): Likewise.
10087 (m32r_trampoline_init): Likewise.
10088 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10089 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10090 (m68k_call_m68k_read_tp): Likewise.
10091 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10092 (microblaze_expand_divide): Likewise.
10093 * config/mips/mips.h (mips_args): Likewise.
10094 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10095 (MIPS_ICACHE_SYNC): Likewise.
10096 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10097 (nios2_trampoline_init): Likewise.
10098 * config/pa/pa.c (hppa_tls_call): Likewise.
10099 (pa_trampoline_init): Likewise.
10100 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10101 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10102 (expand_strn_compare): Likewise.
10103 (rs6000_generate_compare): Likewise.
10104 (rs6000_expand_float128_convert): Likewise.
10105 (output_profile_hook): Likewise.
10106 (rs6000_trampoline_init): Likewise.
10107 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10108 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10109 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10110 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10111 (rs6000_generate_compare): Likewise.
10112 (rs6000_expand_float128_convert): Likewise.
10113 (output_profile_hook): Likewise.
10114 (rs6000_trampoline_init): Likewise.
10115 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10116 * config/sh/sh.c (sh_trampoline_init): Likewise.
10117 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10118 (sparc_emit_float_lib_cmp): Likewise.
10119 (sparc32_initialize_trampoline): Likewise.
10120 (sparc64_initialize_trampoline): Likewise.
10121 (sparc_profile_hook): Likewise.
10122 * config/spu/spu.c (ea_load_store): Likewise.
10123 * config/spu/spu.md (floatunssidf2): Likewise.
10124 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10125 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10126 * config/visium/visium.c (expand_block_move_4): Likewise.
10127 (expand_block_move_2): Likewise.
10128 (expand_block_move_1): Likewise.
10129 (expand_block_set_4): Likewise.
10130 (expand_block_set_2): Likewise.
10131 (expand_block_set_1): Likewise.
10132 (visium_trampoline_init): Likewise.
10133 (visium_profile_hook): Likewise.
10134 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10135 (xtensa_setup_frame_addresses): Likewise.
10136 (xtensa_trampoline_init): Likewise.
10137 * except.c (sjlj_emit_function_enter): Likewise.
10138 (sjlj_emit_function_exit): Likewise.
10139 * explow.c (allocate_dynamic_stack_space): Likewise.
10140 (probe_stack_range): Likewise.
10141 * expr.c (convert_mode_scalar): Likewise.
10142 * optabs.c (expand_binop): Likewise.
10143 (expand_twoval_binop_libfunc): Likewise.
10144 (expand_unop): Likewise.
10145 (prepare_cmp_insn): Likewise.
10146 (prepare_float_lib_cmp): Likewise.
10147 (expand_float): Likewise.
10148 (expand_fix): Likewise.
10149 (expand_fixed_convert): Likewise.
10150 (maybe_emit_sync_lock_test_and_set): Likewise.
10151 (expand_atomic_compare_and_swap): Likewise.
10152 (expand_mem_thread_fence): Likewise.
10153 (expand_atomic_fetch_op): Likewise.
10155 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10157 * doc/generic.texi (OpenACC): Adjust URL.
10158 * doc/invoke.texi (C Dialect Options): Ditto.
10160 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10162 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10163 predicate for operand 1. Add (m,<S>) constraint.
10164 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10165 Prevent memory operand 1 with register operand 2.
10167 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10169 PR rtl-optimization/82024
10170 * combine.c (try_combine): If the combination result is a PARALLEL,
10171 and we only need to retain the SET in there that would be placed
10172 at I2, check that we can place that at I3 instead, before doing so.
10174 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10177 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10179 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10180 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10183 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10184 Jeff Law <law@redhat.com>
10186 * varasm.c (bss_initializer_p): Do not put constants into .bss
10187 (categorize_decl_for_section): Handle bss_initializer_p returning
10188 false when DECL_INITIAL is NULL.
10190 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10193 * config/s390/s390.c (s390_can_inline_p): New function.
10195 2017-09-01 Jeff Law <law@redhat.com>
10197 PR tree-optimization/82052
10198 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10199 Always initialize the returned slot after a hash table miss
10200 when INSERT is true.
10202 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10204 * config/s390/s390.md (mem_signal_fence): Remove.
10205 * doc/md.texi (mem_signal_fence): Remove.
10206 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10208 * target-insns.def (mem_signal_fence): Remove.
10210 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10213 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10216 * asan.c (create_odr_indicator): Strip name encoding from assembler
10217 name before appending it after __odr_asan_.
10219 2017-09-01 Martin Liska <mliska@suse.cz>
10221 PR tree-optimization/82059
10222 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10223 frequency only when an edge is redirected.
10225 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10227 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10228 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10229 (arc_conditional_register_usage): Remove ARC600 lp_count
10231 (arc_file_start): Emit Tag_ARC_CPU_variation.
10232 (arc_can_use_doloop_p): New conditions to use ZOLs.
10233 (hwloop_fail): New function.
10234 (hwloop_optimize): Likewise.
10235 (hwloop_pattern_reg): Likewise.
10236 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10237 (arc_reorg_loops): New function, calls reorg_loops.
10238 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10239 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10241 (arc_loop_hazard): Remove function, functionality moved into
10243 (arc_hazard): Remove arc_loop_hazard call.
10244 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10245 into hwloop_optimize.
10246 (arc_label_align): Remove ZOL handling.
10247 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10248 * config/arc/arc.md (doloop_begin): Remove pattern.
10249 (doloop_begin_i): Likewise.
10250 (doloop_end_i): Likewise.
10251 (doloop_fallback): Likewise.
10252 (doloop_fallback_m): Likewise.
10253 (doloop_end): Reimplement expand.
10254 (arc_lp): New pattern for LP instruction.
10255 (loop_end): New pattern.
10256 (loop_fail): Likewise.
10257 (decrement_and_branch_until_zero): Likewise.
10258 * config/arc/arc.opt (mlpc-width): New option.
10259 * doc/invoke.texi (mlpc-width): Document option.
10261 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10263 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10264 (arc_ccfsm_advance): Fix checking for delay slots.
10265 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10267 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10269 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10271 (movhi_insn): Update store instruction constraint which are saving
10272 6-bit short immediates.
10273 (movsi_insn): Consider also short scaled load operations.
10274 (zero_extendhisi2_i): Use Usd constraint instead of T.
10275 (extendhisi2_i): Add q constraint.
10276 (arc_clzsi2): Add type and length attributes.
10277 (arc_ctzsi2): Likewise.
10278 * config/arc/constraints.md (Usc): Update constraint, the
10279 assembler can parse two relocations for a single instruction.
10281 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10283 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10284 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10286 2017-08-31 Olivier Hainque <hainque@adacore.com>
10288 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10289 match as powerpc-wrs-vxworks*.
10291 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10293 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10294 register constraint for by-element operand.
10295 (aarch64_mls_elt_merge<mode>): Likewise.
10297 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10299 * config/arc/arc.c (arc_can_follow_jump): Check for short
10302 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10304 * config.gcc: Use g.opt for arc.
10305 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10306 functionality moved to ...
10307 (legitimate_scaled_address_p): New function, ...here.
10308 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10309 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10310 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10312 (arc_override_options): Handle G option.
10313 (arc_output_pic_addr_const): Correct function definition.
10314 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10315 (arc_decl_anon_ns_mem_p): Delete.
10316 (arc_in_small_data_p): Overhaul this function to take into
10317 consideration the value given via G option.
10318 (arc_rewrite_small_data_1): Renamed and corrected old
10319 arc_rewrite_small_data function.
10320 (arc_rewrite_small_data): New function.
10321 (small_data_pattern): Don't use pic_offset_table_rtx.
10322 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10323 * config/arc/simdext.md (movmisalignv2hi): Use
10324 prepare_move_operands function.
10326 (movmisalign*): Likewise.
10327 * doc/invoke.texi (ARC options): Document -G option.
10329 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10331 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10333 * config/arc/arc.c (arc_print_operand): Output scalled address for
10334 sdata whenever is possible.
10335 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10336 load/stores are available.
10337 (compact_sda_memory_operand): Check for the alignment required by
10338 code density instructions.
10339 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10341 * config/arc/constraints.md (Usd): Update constraint.
10342 (Us0): New constraint.
10343 (Usc): Update constraint.
10345 2017-08-31 Richard Biener <rguenther@suse.de>
10347 PR middle-end/82054
10348 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10349 function only once.
10351 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10353 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10354 Resize type_signature.
10356 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10357 Alan Hayward <alan.hayward@arm.com>
10358 David Sherwood <david.sherwood@arm.com>
10360 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10361 subregs whose inner modes can be stored in GPRs.
10362 (aarch64_classify_index): Likewise.
10364 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10365 Alan Hayward <alan.hayward@arm.com>
10366 David Sherwood <david.sherwood@arm.com>
10368 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10369 (V_INT_EQUIV): ...this.
10370 (v_cmp_result): Rename to...
10371 (v_int_equiv): ...this.
10372 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10373 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10374 (copysign<mode>3): Likewise.
10375 (aarch64_simd_bsl<mode>_internal): Likewise.
10376 (aarch64_simd_bsl<mode>): Likewise.
10377 (vec_cmp<mode><mode>): Likewise.
10378 (vcond<mode><mode>): Likewise.
10379 (vcond<v_cmp_mixed><mode>): Likewise.
10380 (vcondu<mode><v_cmp_mixed>): Likewise.
10381 (aarch64_cm<optab><mode>): Likewise.
10382 (aarch64_cmtst<mode>): Likewise.
10383 (aarch64_fac<optab><mode>): Likewise.
10384 (vec_perm_const<mode>): Likewise.
10385 (vcond_mask_<mode><v_cmp_result>): Rename to...
10386 (vcond_mask_<mode><v_int_equiv>): ...this.
10387 (vec_cmp<mode><v_cmp_result>): Rename to...
10388 (vec_cmp<mode><v_int_equiv>): ...this.
10390 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10391 Alan Hayward <alan.hayward@arm.com>
10392 David Sherwood <david.sherwood@arm.com>
10394 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10396 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10397 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10398 (UNSPEC_LD4_DREG): New unspecs.
10399 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10400 (aarch64_ld2<mode>_dreg_be): Replace with...
10401 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10403 (aarch64_ld3<mode>_dreg_le)
10404 (aarch64_ld3<mode>_dreg_be): Replace with...
10405 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10407 (aarch64_ld4<mode>_dreg_le)
10408 (aarch64_ld4<mode>_dreg_be): Replace with...
10409 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10412 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10414 PR tree-optimization/81987
10415 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10416 insert an initializer in a location not dominated by the stride
10419 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10421 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10422 on the entire header of the finally block in the fallthru case.
10424 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10426 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10428 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10430 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10431 rs6000_emit_move_from_cr and call renamed function.
10432 (rs6000_emit_prologue): Call renamed functions.
10433 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10434 movesi_from_cr, remove volatile CRs.
10436 2017-08-30 Jon Beniston <jon@beniston.com>
10437 Richard Biener <rguenther@suse.de>
10439 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10440 of VECTOR_MODE_P check.
10441 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10442 element vector types.
10444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10446 * df.h (df_read_modify_subreg_p): Remove in favor of...
10447 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10448 const_rtx instead of an rtx.
10449 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10450 * df-problems.c (df_word_lr_mark_ref): Likewise.
10451 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10452 (mark_pseudo_reg_dead): Likewise.
10453 (mark_ref_dead): Likewise.
10454 * reginfo.c (init_subregs_of_mode): Likewise.
10455 * sched-deps.c (sched_analyze_1): Likewise.
10456 * df-scan.c (df_def_record_1): Likewise.
10457 (df_uses_record): Likewise.
10458 (df_read_modify_subreg_p): Remove in favor of...
10459 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10460 const_rtx instead of an rtx.
10462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10463 Alan Hayward <alan.hayward@arm.com>
10464 David Sherwood <david.sherwood@arm.com>
10466 * rtl.h (partial_subreg_p): New function.
10467 * caller-save.c (save_call_clobbered_regs): Use it.
10468 * calls.c (expand_call): Likewise.
10469 * combine.c (combinable_i3pat): Likewise.
10470 (simplify_set): Likewise.
10471 (make_extraction): Likewise.
10472 (make_compound_operation_int): Likewise.
10473 (gen_lowpart_or_truncate): Likewise.
10474 (force_to_mode): Likewise.
10475 (make_field_assignment): Likewise.
10476 (reg_truncated_to_mode): Likewise.
10477 (record_truncated_value): Likewise.
10478 (move_deaths): Likewise.
10479 * cse.c (record_jump_cond): Likewise.
10480 (cse_insn): Likewise.
10481 * cselib.c (cselib_lookup_1): Likewise.
10482 * expmed.c (extract_bit_field_using_extv): Likewise.
10483 * function.c (assign_parm_setup_reg): Likewise.
10484 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10485 * ira-build.c (create_insn_allocnos): Likewise.
10486 * lra-coalesce.c (merge_pseudos): Likewise.
10487 * lra-constraints.c (match_reload): Likewise.
10488 (simplify_operand_subreg): Likewise.
10489 (curr_insn_transform): Likewise.
10490 * lra-lives.c (process_bb_lives): Likewise.
10491 * lra.c (new_insn_reg): Likewise.
10492 (lra_substitute_pseudo): Likewise.
10493 * regcprop.c (mode_change_ok): Likewise.
10494 (maybe_mode_change): Likewise.
10495 (copyprop_hardreg_forward_1): Likewise.
10496 * reload.c (push_reload): Likewise.
10497 (find_reloads): Likewise.
10498 (find_reloads_subreg_address): Likewise.
10499 * reload1.c (alter_reg): Likewise.
10500 (eliminate_regs_1): Likewise.
10501 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10503 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10505 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10506 back to if statements, including unpack.
10508 2017-08-30 Martin Liska <mliska@suse.cz>
10510 PR inline-asm/82001
10511 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10513 (func_checker::compare_asm_inputs_outputs): ... this function.
10514 (func_checker::compare_gimple_asm): Use the function to compare
10515 also ASM constrains.
10516 * ipa-icf-gimple.h: Rename the function.
10518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10519 Alan Hayward <alan.hayward@arm.com>
10520 David Sherwood <david.sherwood@arm.com>
10522 * coretypes.h (complex_mode): New type.
10523 * gdbhooks.py (build_pretty_printer): Handle it.
10524 * machmode.h (complex_mode): New class.
10525 (complex_mode::includes_p): New function.
10526 (is_complex_int_mode): Likewise.
10527 (is_complex_float_mode): Likewise.
10528 * genmodes.c (get_mode_class): Handle complex mode classes.
10529 * function.c (expand_function_end): Use is_complex_int_mode.
10531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10532 Alan Hayward <alan.hayward@arm.com>
10533 David Sherwood <david.sherwood@arm.com>
10535 * coretypes.h (scalar_mode_pod): New typedef.
10536 * gdbhooks.py (build_pretty_printer): Handle it.
10537 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10538 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10539 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10540 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10541 as_a <scalar_mode>.
10543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10544 Alan Hayward <alan.hayward@arm.com>
10545 David Sherwood <david.sherwood@arm.com>
10547 * machmode.h (mode_for_vector): Take a scalar_mode instead
10549 * stor-layout.c (mode_for_vector): Likewise.
10550 * explow.c (promote_mode): Use as_a <scalar_mode>.
10551 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10554 Alan Hayward <alan.hayward@arm.com>
10555 David Sherwood <david.sherwood@arm.com>
10557 * target.def (preferred_simd_mode): Take a scalar_mode
10558 instead of a machine_mode.
10559 * targhooks.h (default_preferred_simd_mode): Likewise.
10560 * targhooks.c (default_preferred_simd_mode): Likewise.
10561 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10562 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10563 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10564 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10565 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10566 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10567 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10568 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10570 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10571 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10572 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10573 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10574 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10575 * doc/tm.texi: Regenerate.
10576 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10580 Alan Hayward <alan.hayward@arm.com>
10581 David Sherwood <david.sherwood@arm.com>
10583 * target.def (scalar_mode_supported_p): Take a scalar_mode
10584 instead of a machine_mode.
10585 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10586 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10587 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10588 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10589 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10590 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10591 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10592 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10593 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10594 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10595 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10596 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10597 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10598 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10600 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10601 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10602 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10603 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10604 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10606 * doc/tm.texi: Regenerate.
10608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10609 Alan Hayward <alan.hayward@arm.com>
10610 David Sherwood <david.sherwood@arm.com>
10612 * coretypes.h (opt_scalar_mode): New typedef.
10613 * gdbhooks.py (build_pretty_printers): Handle it.
10614 * machmode.h (mode_iterator::get_2xwider): Add overload for
10616 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10618 * expr.c (convert_mode_scalar): Likewise.
10619 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10620 * optabs.c (expand_float): Likewise.
10621 (expand_fix): Likewise.
10622 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10625 Alan Hayward <alan.hayward@arm.com>
10626 David Sherwood <david.sherwood@arm.com>
10628 * optabs.c (expand_float): Explicitly check for scalars before
10629 using a branching expansion.
10630 (expand_fix): Likewise.
10632 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10633 Alan Hayward <alan.hayward@arm.com>
10634 David Sherwood <david.sherwood@arm.com>
10636 * expr.c (convert_mode): Split scalar handling out into...
10637 (convert_mode_scalar): ...this new function. Treat the modes
10640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10641 Alan Hayward <alan.hayward@arm.com>
10642 David Sherwood <david.sherwood@arm.com>
10644 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10646 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10649 Alan Hayward <alan.hayward@arm.com>
10650 David Sherwood <david.sherwood@arm.com>
10652 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10653 rather than a machine_mode.
10654 (fixed_from_string): Likewise.
10655 (fixed_convert): Likewise.
10656 (fixed_convert_from_int): Likewise.
10657 (fixed_convert_from_real): Likewise.
10658 (real_convert_from_fixed): Likewise.
10659 * fixed-value.c (fixed_from_double_int): Likewise.
10660 (fixed_from_string): Likewise.
10661 (fixed_convert): Likewise.
10662 (fixed_convert_from_int): Likewise.
10663 (fixed_convert_from_real): Likewise.
10664 (real_convert_from_fixed): Likewise.
10665 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10668 Alan Hayward <alan.hayward@arm.com>
10669 David Sherwood <david.sherwood@arm.com>
10671 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10672 of separate mode class checks. Do not allow vector modes here.
10673 (immed_wide_int_const): Use as_a <scalar_mode>.
10674 * explow.c (trunc_int_for_mode): Likewise.
10675 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10676 (wi::shwi): Likewise.
10677 (wi::min_value): Likewise.
10678 (wi::max_value): Likewise.
10679 * dwarf2out.c (loc_descriptor): Likewise.
10680 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10681 for CONST_WIDE_INT.
10683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10684 Alan Hayward <alan.hayward@arm.com>
10685 David Sherwood <david.sherwood@arm.com>
10687 * tree.h (SCALAR_TYPE_MODE): New macro.
10688 * expr.c (expand_expr_addr_expr_1): Use it.
10689 (expand_expr_real_2): Likewise.
10690 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10691 (fold_convert_const_fixed_from_int): Likewise.
10692 (fold_convert_const_fixed_from_real): Likewise.
10693 (native_encode_fixed): Likewise
10694 (native_encode_complex): Likewise
10695 (native_encode_vector): Likewise.
10696 (native_interpret_fixed): Likewise.
10697 (native_interpret_real): Likewise.
10698 (native_interpret_complex): Likewise.
10699 (native_interpret_vector): Likewise.
10700 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10701 (simd_clone_adjust_argument_types): Likewise.
10702 (simd_clone_init_simd_arrays): Likewise.
10703 (simd_clone_adjust): Likewise.
10704 * stor-layout.c (layout_type): Likewise.
10705 * tree.c (build_minus_one_cst): Likewise.
10706 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10707 * tree-inline.c (estimate_move_cost): Likewise.
10708 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10709 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10710 (vectorizable_reduction): Likewise.
10711 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10712 (vect_recog_mixed_size_cond_pattern): Likewise.
10713 (check_bool_pattern): Likewise.
10714 (adjust_bool_pattern): Likewise.
10715 (search_type_for_mask_1): Likewise.
10716 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10717 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10718 (vectorizable_load): Likewise.
10719 (vectorizable_store): Likewise.
10720 * ubsan.c (ubsan_encode_value): Likewise.
10721 * varasm.c (output_constant): Likewise.
10723 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10724 Alan Hayward <alan.hayward@arm.com>
10725 David Sherwood <david.sherwood@arm.com>
10727 * coretypes.h (scalar_mode): New class.
10728 * machmode.h (scalar_mode): Likewise.
10729 (scalar_mode::includes_p): New function.
10730 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10731 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10732 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10733 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10734 * expmed.c (store_bit_field_1): Likewise.
10735 (extract_bit_field_1): Likewise.
10736 * expr.c (write_complex_part): Likewise.
10737 (read_complex_part): Likewise.
10738 (emit_move_complex_push): Likewise.
10739 (expand_expr_real_2): Likewise.
10740 * function.c (assign_parm_setup_reg): Likewise.
10741 (assign_parms_unsplit_complex): Likewise.
10742 * optabs.c (expand_binop): Likewise.
10743 * rtlanal.c (subreg_get_info): Likewise.
10744 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10745 * varasm.c (output_constant_pool_2): Likewise.
10747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10748 Alan Hayward <alan.hayward@arm.com>
10749 David Sherwood <david.sherwood@arm.com>
10751 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10753 (expmed_mult_highpart_optab): Likewise.
10754 (expmed_mult_highpart): Likewise.
10756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10757 Alan Hayward <alan.hayward@arm.com>
10758 David Sherwood <david.sherwood@arm.com>
10760 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10761 instead of a machine_mode.
10762 (builtin_memset_read_str): Likewise.
10763 * builtins.c (c_readstr): Likewise.
10764 (builtin_memcpy_read_str): Likewise.
10765 (builtin_strncpy_read_str): Likewise.
10766 (builtin_memset_read_str): Likewise.
10767 (builtin_memset_gen_str): Likewise.
10768 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10769 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10770 instead of a machine_mode.
10771 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10773 (make_extraction): Likewise.
10774 (try_widen_shift_mode): Take and return scalar_int_modes instead
10776 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10777 a scalar_int_mode instead of a machine_mode.
10778 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10779 (avr_addr_space_pointer_mode): Likewise.
10780 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10781 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10782 (msp430_unwind_word_mode): Likewise.
10783 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10784 (spu_addr_space_pointer_mode): Likewise.
10785 (spu_addr_space_address_mode): Likewise.
10786 (spu_libgcc_cmp_return_mode): Likewise.
10787 (spu_libgcc_shift_count_mode): Likewise.
10788 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10789 (rl78_addr_space_pointer_mode): Likewise.
10790 (fl78_unwind_word_mode): Likewise.
10791 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10793 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10794 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10795 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10796 (mips_valid_pointer_mode): Likewise.
10797 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10798 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10799 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10801 (ft32_addr_space_address_mode): Likewise.
10802 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10803 scalar_int_mode instead of a machine_mode.
10804 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10806 (m32c_addr_space_address_mode): Likewise.
10807 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10808 (rs6000_eh_return_filter_mode): Likewise.
10809 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10810 (rs6000_eh_return_filter_mode): Likewise.
10811 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10812 (s390_libgcc_shift_count_mode): Likewise.
10813 (s390_unwind_word_mode): Likewise.
10814 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10816 * target.def (mode_rep_extended): Likewise.
10817 (valid_pointer_mode): Likewise.
10818 (addr_space.valid_pointer_mode): Likewise.
10819 (eh_return_filter_mode): Return a scalar_int_mode rather than
10821 (libgcc_cmp_return_mode): Likewise.
10822 (libgcc_shift_count_mode): Likewise.
10823 (unwind_word_mode): Likewise.
10824 (addr_space.pointer_mode): Likewise.
10825 (addr_space.address_mode): Likewise.
10826 * doc/tm.texi: Regenerate.
10827 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10829 (do_jump): Use scalar_int_mode for local variables.
10830 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10831 rather than a machine_mode.
10832 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10833 (scompare_loc_descriptor_wide): Likewise.
10834 (scompare_loc_descriptor_narrow): Likewise.
10835 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10837 * except.c (sjlj_emit_dispatch_table): Likewise.
10838 (expand_builtin_eh_copy_values): Likewise.
10839 * explow.c (convert_memory_address_addr_space_1): Likewise.
10840 Take a scalar_int_mode rather than a machine_mode.
10841 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10842 than a machine_mode.
10843 (memory_address_addr_space): Use scalar_int_mode for local variables.
10844 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10845 rather than a machine_mode.
10846 * expmed.c (mask_rtx): Likewise.
10847 (init_expmed_one_conv): Likewise.
10848 (expand_mult_highpart_adjust): Likewise.
10849 (extract_high_half): Likewise.
10850 (expmed_mult_highpart_optab): Likewise.
10851 (expmed_mult_highpart): Likewise.
10852 (expand_smod_pow2): Likewise.
10853 (expand_sdiv_pow2): Likewise.
10854 (emit_store_flag_int): Likewise.
10855 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10857 (extract_low_bits): Likewise.
10858 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10860 * expr.c (pieces_addr::adjust): Likewise.
10861 (can_store_by_pieces): Likewise.
10862 (store_by_pieces): Likewise.
10863 (clear_by_pieces_1): Likewise.
10864 (expand_expr_addr_expr_1): Likewise.
10865 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10866 (expand_expr_real_1): Likewise.
10867 (try_casesi): Likewise.
10868 * final.c (shorten_branches): Likewise.
10869 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10870 type of "mode" to machine_mode.
10871 * internal-fn.c (expand_arith_overflow_result_store): Take a
10872 scalar_int_mode rather than a machine_mode.
10873 (expand_mul_overflow): Use scalar_int_mode for local variables.
10874 * loop-doloop.c (doloop_modify): Likewise.
10875 (doloop_optimize): Likewise.
10876 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10877 than a machine_mode.
10878 (expand_doubleword_shift_condmove): Likewise.
10879 (expand_doubleword_shift): Likewise.
10880 (expand_doubleword_clz): Likewise.
10881 (expand_doubleword_popcount): Likewise.
10882 (expand_doubleword_parity): Likewise.
10883 (expand_absneg_bit): Use scalar_int_mode for local variables.
10884 (prepare_float_lib_cmp): Likewise.
10885 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10886 rather than a machine_mode.
10887 (convert_memory_address_addr_space): Likewise.
10888 (get_mode_bounds): Likewise.
10889 (get_address_mode): Return a scalar_int_mode rather than a
10891 * rtlanal.c (get_address_mode): Likewise.
10892 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10893 than a machine_mode.
10894 * targhooks.c (default_mode_rep_extended): Likewise.
10895 (default_valid_pointer_mode): Likewise.
10896 (default_addr_space_valid_pointer_mode): Likewise.
10897 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10898 than a machine_mode.
10899 (default_libgcc_cmp_return_mode): Likewise.
10900 (default_libgcc_shift_count_mode): Likewise.
10901 (default_unwind_word_mode): Likewise.
10902 (default_addr_space_pointer_mode): Likewise.
10903 (default_addr_space_address_mode): Likewise.
10904 * targhooks.h (default_eh_return_filter_mode): Likewise.
10905 (default_libgcc_cmp_return_mode): Likewise.
10906 (default_libgcc_shift_count_mode): Likewise.
10907 (default_unwind_word_mode): Likewise.
10908 (default_addr_space_pointer_mode): Likewise.
10909 (default_addr_space_address_mode): Likewise.
10910 (default_mode_rep_extended): Take a scalar_int_mode rather than
10912 (default_valid_pointer_mode): Likewise.
10913 (default_addr_space_valid_pointer_mode): Likewise.
10914 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10916 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10917 rather than a machine_mode.
10918 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10919 for local variables.
10920 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10921 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10922 than a machine_mode.
10924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10925 Alan Hayward <alan.hayward@arm.com>
10926 David Sherwood <david.sherwood@arm.com>
10928 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10929 the mode argument to scalar_int_mode.
10930 (do_jump_by_parts_zero_rtx): Likewise.
10931 (do_jump_by_parts_equality_rtx): Likewise.
10932 (do_jump_by_parts_greater): Take a mode argument.
10933 (do_jump_by_parts_equality): Likewise.
10934 (do_jump_1): Update calls accordingly.
10936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10937 Alan Hayward <alan.hayward@arm.com>
10938 David Sherwood <david.sherwood@arm.com>
10940 * is-a.h (safe_dyn_cast): New function.
10941 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10942 (jump_table_for_label): Likewise.
10943 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10944 instead of an rtx_insn *.
10945 (shorten_branches): Use dyn_cast instead of LABEL_P and
10946 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10947 rtx_jump_table_data::get_data_mode.
10948 (final_scan_insn): Likewise.
10950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10951 Alan Hayward <alan.hayward@arm.com>
10952 David Sherwood <david.sherwood@arm.com>
10954 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10955 trying to combine a full-register integer set with a subreg
10958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10959 Alan Hayward <alan.hayward@arm.com>
10960 David Sherwood <david.sherwood@arm.com>
10962 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10963 that is always either address_mode or pointer_mode.
10965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10966 Alan Hayward <alan.hayward@arm.com>
10967 David Sherwood <david.sherwood@arm.com>
10969 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10970 when the two are known to be equal.
10972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10973 Alan Hayward <alan.hayward@arm.com>
10974 David Sherwood <david.sherwood@arm.com>
10976 * simplify-rtx.c (simplify_const_unary_operation): Use
10977 is_a <scalar_int_mode> instead of checking for a nonzero
10978 precision. Forcibly convert op_mode to a scalar_int_mode
10979 in that case. More clearly differentiate the operand and
10980 result modes and use the former when deciding what the value
10981 of a count-bits operation should be. Use is_int_mode instead
10982 of checking for a MODE_INT. Remove redundant check for whether
10983 this mode has a zero precision.
10985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10986 Alan Hayward <alan.hayward@arm.com>
10987 David Sherwood <david.sherwood@arm.com>
10989 * optabs.c (widen_leading): Change the type of the mode argument
10990 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10991 (widen_bswap): Likewise.
10992 (expand_parity): Likewise.
10993 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10994 (expand_ffs): Likewise.
10995 (epand_unop): Check for scalar integer modes before calling the
10998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10999 Alan Hayward <alan.hayward@arm.com>
11000 David Sherwood <david.sherwood@arm.com>
11002 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11004 (expand_expr_real_1): Update call accordingly.
11006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11007 Alan Hayward <alan.hayward@arm.com>
11008 David Sherwood <david.sherwood@arm.com>
11010 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11011 value_mode arguments. Use scalar_int_mode internally.
11012 (store_bit_field_1): Rename the new integer mode from imode
11013 to op0_mode and use it instead of GET_MODE (op0). Update calls
11014 to store_split_bit_field, store_bit_field_using_insv and
11015 store_fixed_bit_field.
11016 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11017 Use scalar_int_mode internally. Use a bit count rather than a mode
11018 when calculating the largest bit size for get_best_mode.
11019 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11020 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11021 Remove assertion that OP0 has a scalar integer mode.
11022 (store_split_bit_field): Add op0_mode and value_mode arguments.
11023 Update calls to extract_fixed_bit_field.
11024 (extract_bit_field_using_extv): Add an op0_mode argument.
11025 Use scalar_int_mode internally.
11026 (extract_bit_field_1): Rename the new integer mode from imode to
11027 op0_mode and use it instead of GET_MODE (op0). Update calls to
11028 extract_split_bit_field, extract_bit_field_using_extv and
11029 extract_fixed_bit_field.
11030 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11031 to extract_split_bit_field and extract_fixed_bit_field_1.
11032 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11033 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11034 on the target mode.
11035 (extract_split_bit_field): Add an op0_mode argument. Update call
11036 to extract_fixed_bit_field.
11038 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11039 Alan Hayward <alan.hayward@arm.com>
11040 David Sherwood <david.sherwood@arm.com>
11042 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11043 * explow.c (hard_function_value): Likewise.
11044 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11045 convert_to_mode call outside the loop.
11046 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11047 for the mode iterator. Require the mode specified by max_pieces
11049 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11051 (copy_blkmode_to_reg): Likewise.
11052 (set_storage_via_setmem): Likewise.
11053 * optabs.c (prepare_cmp_insn): Likewise.
11054 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11055 * stor-layout.c (finish_bitfield_representative): Likewise.
11057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11058 Alan Hayward <alan.hayward@arm.com>
11059 David Sherwood <david.sherwood@arm.com>
11061 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11062 * expr.c (convert_move): Use them.
11063 (convert_modes): Likewise.
11064 (store_expr_with_bounds): Likewise.
11066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11067 Alan Hayward <alan.hayward@arm.com>
11068 David Sherwood <david.sherwood@arm.com>
11070 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11071 parameter for the mode of "x". Remove the "known_x", "known_mode"
11072 and "known_ret" arguments. Change the type of the mode argument
11073 to scalar_int_mode.
11074 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11075 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11076 (reg_num_sign_bit_copies_for_combine): Likewise.
11077 * rtlanal.c (nonzero_bits1): Likewise.
11078 (num_sign_bit_copies1): Likewise.
11079 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11080 (reg_num_sign_bit_copies_general): Likewise.
11081 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11082 (reg_nonzero_bits_general): Likewise.
11084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11085 Alan Hayward <alan.hayward@arm.com>
11086 David Sherwood <david.sherwood@arm.com>
11088 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11089 than in subroutines. Return 1 for non-integer modes.
11090 (cached_num_sign_bit_copies): Change the type of the mode parameter
11091 to scalar_int_mode.
11092 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11093 classes. Handle CONST_INT_P first and then check whether X also
11094 has a scalar integer mode. Check the same thing for inner registers
11095 of a SUBREG and for values that are being extended or truncated.
11097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11098 Alan Hayward <alan.hayward@arm.com>
11099 David Sherwood <david.sherwood@arm.com>
11101 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11102 in subroutines. Return the mode mask for non-integer modes.
11103 (cached_nonzero_bits): Change the type of the mode parameter
11104 to scalar_int_mode.
11105 (nonzero_bits1): Likewise. Remove early exit for other mode
11106 classes. Handle CONST_INT_P first and then check whether X
11107 also has a scalar integer mode.
11109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11110 Alan Hayward <alan.hayward@arm.com>
11111 David Sherwood <david.sherwood@arm.com>
11113 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11114 Return a scalar_int_mode and assert that the size is greater than
11116 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11117 (op_by_pieces_d::op_by_pieces_d): Likewise.
11118 (op_by_pieces_d::run): Likewise.
11119 (can_store_by_pieces): Likewise.
11121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11122 Alan Hayward <alan.hayward@arm.com>
11123 David Sherwood <david.sherwood@arm.com>
11125 * combine.c (extract_left_shift): Add a mode argument and update
11127 (make_compound_operation_int): Change the type of the mode parameter
11128 to scalar_int_mode and update the call to extract_left_shift.
11130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11131 Alan Hayward <alan.hayward@arm.com>
11132 David Sherwood <david.sherwood@arm.com>
11134 * combine.c (simplify_and_const_int): Change the type of the mode
11135 parameter to scalar_int_mode.
11136 (simplify_and_const_int_1): Likewise. Update recursive call.
11138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11139 Alan Hayward <alan.hayward@arm.com>
11140 David Sherwood <david.sherwood@arm.com>
11142 * combine.c (simplify_compare_const): Check that the mode is a
11143 scalar_int_mode (rather than VOIDmode) before testing its
11145 (simplify_comparison): Move COMPARISON_P handling out of the
11146 loop and restrict the latter part of the loop to scalar_int_modes.
11147 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11148 and when considering SUBREG_REGs. Use is_int_mode instead of
11149 checking GET_MODE_CLASS against MODE_INT.
11151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11152 Alan Hayward <alan.hayward@arm.com>
11153 David Sherwood <david.sherwood@arm.com>
11155 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11156 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11157 shift_unit_mode and for modes involved in scalar shifts.
11159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11160 Alan Hayward <alan.hayward@arm.com>
11161 David Sherwood <david.sherwood@arm.com>
11163 * combine.c (force_int_to_mode): New function, split out from...
11164 (force_to_mode): ...here. Keep xmode up-to-date and use it
11165 instead of GET_MODE (x).
11167 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11168 Alan Hayward <alan.hayward@arm.com>
11169 David Sherwood <david.sherwood@arm.com>
11171 * optabs-query.h (extraction_insn::struct_mode): Change type to
11172 opt_scalar_int_mode and update comment.
11173 (extraction_insn::field_mode): Change type to scalar_int_mode.
11174 (extraction_insn::pos_mode): Likewise.
11175 * combine.c (make_extraction): Update accordingly.
11176 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11177 (get_optab_extraction_insn): Likewise.
11178 * recog.c (simplify_while_replacing): Likewise.
11179 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11180 parameter to opt_scalar_int_mode.
11182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11183 Alan Hayward <alan.hayward@arm.com>
11184 David Sherwood <david.sherwood@arm.com>
11186 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11187 to a scalar_int_mode instead of a machine_mode.
11188 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11189 (get_best_mode): Return a boolean and use a pointer argument to store
11190 the selected mode. Replace the limit mode parameter with a bit limit.
11191 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11192 for the values returned by bit_field_mode_iterator::next_mode.
11193 (store_bit_field): Update call to get_best_mode.
11194 (store_fixed_bit_field): Likewise.
11195 (extract_fixed_bit_field): Likewise.
11196 * expr.c (optimize_bitfield_assignment_op): Likewise.
11197 * fold-const.c (optimize_bit_field_compare): Likewise.
11198 (fold_truth_andor_1): Likewise.
11199 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11200 Update for new type of m_mode.
11201 (get_best_mode): As above.
11203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11204 Alan Hayward <alan.hayward@arm.com>
11205 David Sherwood <david.sherwood@arm.com>
11207 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11208 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11209 (store_bit_field): Check is_a <scalar_int_mode> before calling
11210 strict_volatile_bitfield_p.
11211 (extract_bit_field): Likewise.
11213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11214 Alan Hayward <alan.hayward@arm.com>
11215 David Sherwood <david.sherwood@arm.com>
11217 * target.def (cstore_mode): Return a scalar_int_mode.
11218 * doc/tm.texi: Regenerate.
11219 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11220 * targhooks.h (default_cstore_mode): Likewise.
11221 * targhooks.c (default_cstore_mode): Likewise, using a forced
11223 * expmed.c (emit_cstore): Expect the target of the cstore to be
11226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11227 Alan Hayward <alan.hayward@arm.com>
11228 David Sherwood <david.sherwood@arm.com>
11230 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11232 (niter_desc): Likewise mode.
11233 (iv_analyze): Add a mode parameter.
11235 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11236 and change its type to scalar_int_mode.
11237 * loop-iv.c: Update commentary at head of file.
11238 (iv_constant): Pass the mode paraeter before the rtx it describes
11239 and change its type to scalar_int_mode. Remove VOIDmode handling.
11240 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11241 (iv_extend): Likewise.
11242 (shorten_into_mode): Likewise.
11243 (iv_add): Use scalar_int_mode.
11244 (iv_mult): Likewise.
11245 (iv_shift): Likewise.
11246 (canonicalize_iv_subregs): Likewise.
11247 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11248 it describes and change its mode to scalar_int_mode. Also change
11249 the type of the returned inner_mode to scalar_int_mode.
11250 (get_biv_step): Likewise, turning outer_mode from a pointer
11251 into a direct parameter. Update call to get_biv_step_1.
11252 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11253 iv_constant and get_biv_step.
11254 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11255 and change its type to scalar_int_mode. Don't initialise iv->mode
11256 to VOIDmode and remove later checks for its still being VOIDmode.
11257 Update calls to iv_analyze_op and iv_analyze_expr. Check
11258 is_a <scalar_int_mode> when changing the mode under consideration.
11259 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11260 Update call to iv_analyze_expr.
11261 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11262 inner register is not also a scalar_int_mode. Update call to
11264 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11265 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11266 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11267 separate mode class checks. Update calls to iv_analyze. Remove
11268 fix-up of VOIDmodes after iv_analyze_biv.
11269 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11270 don't have a scalar_int_mode. Update call to biv_p.
11272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11273 Alan Hayward <alan.hayward@arm.com>
11274 David Sherwood <david.sherwood@arm.com>
11276 * cfgexpand.c (convert_debug_memory_address): Use
11277 as_a <scalar_int_mode>.
11278 * combine.c (expand_compound_operation): Likewise.
11279 (make_extraction): Likewise.
11280 (change_zero_ext): Likewise.
11281 (simplify_comparison): Likewise.
11282 * cse.c (cse_insn): Likewise.
11283 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11284 (mem_loc_descriptor): Likewise.
11285 (loc_descriptor): Likewise.
11286 * expmed.c (init_expmed_one_mode): Likewise.
11287 (synth_mult): Likewise.
11288 (emit_store_flag_1): Likewise.
11289 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11290 of a comparison with size.
11291 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11292 (reduce_to_bit_field_precision): Likewise.
11293 * function.c (expand_function_end): Likewise.
11294 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11295 * loop-doloop.c (doloop_modify): Likewise.
11296 * optabs.c (expand_binop): Likewise.
11297 (expand_unop): Likewise.
11298 (expand_copysign_absneg): Likewise.
11299 (prepare_cmp_insn): Likewise.
11300 (maybe_legitimize_operand): Likewise.
11301 * recog.c (const_scalar_int_operand): Likewise.
11302 * rtlanal.c (get_address_mode): Likewise.
11303 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11304 (simplify_cond_clz_ctz): Likewise.
11305 * tree-nested.c (get_nl_goto_field): Likewise.
11306 * tree.c (build_vector_type_for_mode): Likewise.
11307 * var-tracking.c (use_narrower_mode): Likewise.
11309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11310 Alan Hayward <alan.hayward@arm.com>
11311 David Sherwood <david.sherwood@arm.com>
11313 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11314 * builtins.c (expand_builtin_signbit): Use it.
11315 * cfgexpand.c (expand_debug_expr): Likewise.
11316 * dojump.c (do_jump): Likewise.
11317 (do_compare_and_jump): Likewise.
11318 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11319 * expmed.c (make_tree): Likewise.
11320 * expr.c (expand_expr_real_2): Likewise.
11321 (expand_expr_real_1): Likewise.
11322 (try_casesi): Likewise.
11323 * fold-const-call.c (fold_const_call_ss): Likewise.
11324 * fold-const.c (unextend): Likewise.
11325 (extract_muldiv_1): Likewise.
11326 (fold_single_bit_test): Likewise.
11327 (native_encode_int): Likewise.
11328 (native_encode_string): Likewise.
11329 (native_interpret_int): Likewise.
11330 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11331 * internal-fn.c (expand_addsub_overflow): Likewise.
11332 (expand_neg_overflow): Likewise.
11333 (expand_mul_overflow): Likewise.
11334 (expand_arith_overflow): Likewise.
11335 * match.pd: Likewise.
11336 * stor-layout.c (layout_type): Likewise.
11337 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11338 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11339 * tree-ssanames.c (get_range_info): Likewise.
11340 * tree-switch-conversion.c (array_value_type) Likewise.
11341 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11342 (vect_recog_divmod_pattern): Likewise.
11343 (vect_recog_mixed_size_cond_pattern): Likewise.
11344 * tree-vrp.c (extract_range_basic): Likewise.
11345 (simplify_float_conversion_using_ranges): Likewise.
11346 * tree.c (int_fits_type_p): Likewise.
11347 * ubsan.c (instrument_bool_enum_load): Likewise.
11348 * varasm.c (mergeable_string_section): Likewise.
11349 (narrowing_initializer_constant_valid_p): Likewise.
11350 (output_constant): Likewise.
11352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11353 Alan Hayward <alan.hayward@arm.com>
11354 David Sherwood <david.sherwood@arm.com>
11356 * machmode.h (NARROWEST_INT_MODE): New macro.
11357 * expr.c (alignment_for_piecewise_move): Use it instead of
11358 GET_CLASS_NARROWEST_MODE (MODE_INT).
11359 (push_block): Likewise.
11360 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11362 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11365 Alan Hayward <alan.hayward@arm.com>
11366 David Sherwood <david.sherwood@arm.com>
11368 * postreload.c (move2add_valid_value_p): Change the type of the
11369 mode parameter to scalar_int_mode.
11370 (move2add_use_add2_insn): Add a mode parameter and use it instead
11372 (move2add_use_add3_insn): Likewise.
11373 (reload_cse_move2add): Update accordingly.
11375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11376 Alan Hayward <alan.hayward@arm.com>
11377 David Sherwood <david.sherwood@arm.com>
11379 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11381 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11382 * optabs.c (expand_unop): Likewise.
11384 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11385 Alan Hayward <alan.hayward@arm.com>
11386 David Sherwood <david.sherwood@arm.com>
11388 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11389 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11390 (popcount_loc_descriptor): Likewise.
11391 (bswap_loc_descriptor): Likewise.
11392 (rotate_loc_descriptor): Likewise.
11393 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11394 calling the functions above.
11396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11397 Alan Hayward <alan.hayward@arm.com>
11398 David Sherwood <david.sherwood@arm.com>
11400 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11402 (try_combine): Likewise.
11403 (simplify_if_then_else): Likewise.
11404 * cse.c (cse_insn): Likewise.
11405 * dwarf2out.c (mem_loc_descriptor): Likewise.
11406 * emit-rtl.c (gen_lowpart_common): Likewise.
11407 * simplify-rtx.c (simplify_truncation): Likewise.
11408 (simplify_binary_operation_1): Likewise.
11409 (simplify_const_relational_operation): Likewise.
11410 (simplify_ternary_operation): Likewise.
11411 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11414 Alan Hayward <alan.hayward@arm.com>
11415 David Sherwood <david.sherwood@arm.com>
11417 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11418 * reload.c (push_reload): Likewise.
11419 (find_reloads): Likewise.
11421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11422 Alan Hayward <alan.hayward@arm.com>
11423 David Sherwood <david.sherwood@arm.com>
11425 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11426 (make_compound_operation_int): Likewise.
11427 (change_zero_ext): Likewise.
11428 * expr.c (convert_move): Likewise.
11429 (convert_modes): Likewise.
11430 * fwprop.c (forward_propagate_subreg): Likewise.
11431 * loop-iv.c (get_biv_step_1): Likewise.
11432 * optabs.c (widen_operand): Likewise.
11433 * postreload.c (move2add_valid_value_p): Likewise.
11434 * recog.c (simplify_while_replacing): Likewise.
11435 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11436 (simplify_binary_operation_1): Likewise. Remove redundant
11437 mode equality check.
11439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11440 Alan Hayward <alan.hayward@arm.com>
11441 David Sherwood <david.sherwood@arm.com>
11443 * combine.c (combine_simplify_rtx): Add checks for
11444 is_a <scalar_int_mode>.
11445 (simplify_if_then_else): Likewise.
11446 (make_field_assignment): Likewise.
11447 (simplify_comparison): Likewise.
11448 * ifcvt.c (noce_try_bitop): Likewise.
11449 * loop-invariant.c (canonicalize_address_mult): Likewise.
11450 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11453 Alan Hayward <alan.hayward@arm.com>
11454 David Sherwood <david.sherwood@arm.com>
11456 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11457 is_a <scalar_int_mode> instead of != BLKmode.
11459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11460 Alan Hayward <alan.hayward@arm.com>
11461 David Sherwood <david.sherwood@arm.com>
11463 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11464 instead of != VOIDmode.
11465 * combine.c (if_then_else_cond): Likewise.
11466 (change_zero_ext): Likewise.
11467 * dwarf2out.c (mem_loc_descriptor): Likewise.
11468 (loc_descriptor): Likewise.
11469 * rtlanal.c (canonicalize_condition): Likewise.
11470 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11473 Alan Hayward <alan.hayward@arm.com>
11474 David Sherwood <david.sherwood@arm.com>
11476 * simplify-rtx.c (simplify_binary_operation_1): Use
11477 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11480 Alan Hayward <alan.hayward@arm.com>
11481 David Sherwood <david.sherwood@arm.com>
11483 * wide-int.h (int_traits<unsigned char>) New class.
11484 (int_traits<unsigned short>) Likewise.
11485 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11486 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11488 * combine.c (set_nonzero_bits_and_sign_copies): Use
11489 is_a <scalar_int_mode>.
11490 (find_split_point): Likewise.
11491 (combine_simplify_rtx): Likewise.
11492 (simplify_logical): Likewise.
11493 (expand_compound_operation): Likewise.
11494 (expand_field_assignment): Likewise.
11495 (make_compound_operation): Likewise.
11496 (extended_count): Likewise.
11497 (change_zero_ext): Likewise.
11498 (simplify_comparison): Likewise.
11499 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11500 (ucompare_loc_descriptor): Likewise.
11501 (minmax_loc_descriptor): Likewise.
11502 (mem_loc_descriptor): Likewise.
11503 (loc_descriptor): Likewise.
11504 * expmed.c (init_expmed_one_mode): Likewise.
11505 * lra-constraints.c (lra_constraint_offset): Likewise.
11506 * optabs.c (prepare_libcall_arg): Likewise.
11507 * postreload.c (move2add_note_store): Likewise.
11508 * reload.c (operands_match_p): Likewise.
11509 * rtl.h (load_extend_op): Likewise.
11510 * rtlhooks.c (gen_lowpart_general): Likewise.
11511 * simplify-rtx.c (simplify_truncation): Likewise.
11512 (simplify_unary_operation_1): Likewise.
11513 (simplify_binary_operation_1): Likewise.
11514 (simplify_const_binary_operation): Likewise.
11515 (simplify_const_relational_operation): Likewise.
11516 (simplify_subreg): Likewise.
11517 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11518 * var-tracking.c (adjust_mems): Likewise.
11519 (prepare_call_arguments): Likewise.
11521 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11522 Alan Hayward <alan.hayward@arm.com>
11523 David Sherwood <david.sherwood@arm.com>
11525 * machmode.h (is_int_mode): New fuction.
11526 * combine.c (find_split_point): Use it.
11527 (combine_simplify_rtx): Likewise.
11528 (simplify_if_then_else): Likewise.
11529 (simplify_set): Likewise.
11530 (simplify_shift_const_1): Likewise.
11531 (simplify_comparison): Likewise.
11532 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11533 * cse.c (notreg_cost): Likewise.
11534 (cse_insn): Likewise.
11535 * cselib.c (cselib_lookup_1): Likewise.
11536 * dojump.c (do_jump_1): Likewise.
11537 (do_compare_rtx_and_jump): Likewise.
11538 * dse.c (get_call_args): Likewise.
11539 * dwarf2out.c (rtl_for_decl_init): Likewise.
11540 (native_encode_initializer): Likewise.
11541 * expmed.c (emit_store_flag_1): Likewise.
11542 (emit_store_flag): Likewise.
11543 * expr.c (convert_modes): Likewise.
11544 (store_field): Likewise.
11545 (expand_expr_real_1): Likewise.
11546 * fold-const.c (fold_read_from_constant_string): Likewise.
11547 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11548 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11549 * optabs.c (expand_binop): Likewise.
11550 (expand_unop): Likewise.
11551 (expand_abs_nojump): Likewise.
11552 (expand_one_cmpl_abs_nojump): Likewise.
11553 * simplify-rtx.c (mode_signbit_p): Likewise.
11554 (val_signbit_p): Likewise.
11555 (val_signbit_known_set_p): Likewise.
11556 (val_signbit_known_clear_p): Likewise.
11557 (simplify_relational_operation_1): Likewise.
11558 * tree.c (vector_type_mode): Likewise.
11560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11561 Alan Hayward <alan.hayward@arm.com>
11562 David Sherwood <david.sherwood@arm.com>
11564 * machmode.h (smallest_mode_for_size): Fix formatting.
11565 (smallest_int_mode_for_size): New function.
11566 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11567 instead of smallest_mode_for_size.
11568 * combine.c (make_extraction): Likewise.
11569 * config/arc/arc.c (arc_expand_movmem): Likewise.
11570 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11571 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11572 * config/s390/s390.c (s390_expand_insv): Likewise.
11573 * config/sparc/sparc.c (assign_int_registers): Likewise.
11574 * config/spu/spu.c (spu_function_value): Likewise.
11575 (spu_function_arg): Likewise.
11576 * coverage.c (get_gcov_type): Likewise.
11577 (get_gcov_unsigned_t): Likewise.
11578 * dse.c (find_shift_sequence): Likewise.
11579 * expmed.c (store_bit_field_1): Likewise.
11580 * expr.c (convert_move): Likewise.
11581 (store_field): Likewise.
11582 * internal-fn.c (expand_arith_overflow): Likewise.
11583 * optabs-query.c (get_best_extraction_insn): Likewise.
11584 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11585 * stor-layout.c (layout_type): Likewise.
11586 (initialize_sizetypes): Likewise.
11587 * targhooks.c (default_get_mask_mode): Likewise.
11588 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11591 Alan Hayward <alan.hayward@arm.com>
11592 David Sherwood <david.sherwood@arm.com>
11594 * machmode.h (opt_mode::else_blk): New function.
11595 (int_mode_for_mode): Declare.
11596 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11597 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11599 * cfgexpand.c (expand_debug_expr): Likewise.
11600 * combine.c (gen_lowpart_or_truncate): Likewise.
11601 (gen_lowpart_for_combine): Likewise.
11602 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11603 * config/avr/avr.c (avr_to_int_mode): Likewise.
11604 (avr_out_plus_1): Likewise.
11605 (avr_out_plus): Likewise.
11606 (avr_out_round): Likewise.
11607 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11608 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11609 (s390_expand_vcond): Likewise.
11610 * config/spu/spu.c (spu_split_immediate): Likewise.
11611 (spu_expand_mov): Likewise.
11612 * dse.c (get_stored_val): Likewise.
11613 * expmed.c (store_bit_field_1): Likewise.
11614 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11616 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11617 (extract_low_bits): Likewise.
11618 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11619 handling rather than repeating the check.
11620 (emit_group_store): Likewise.
11621 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11622 * optabs.c (expand_absneg_bit): Likewise.
11623 (expand_copysign_absneg): Likewise.
11624 (expand_copysign_bit): Likewise.
11625 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11626 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11627 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11628 * var-tracking.c (prepare_call_arguments): Likewise.
11629 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11630 int_mode_for_mode instead of mode_for_size.
11631 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11634 Alan Hayward <alan.hayward@arm.com>
11635 David Sherwood <david.sherwood@arm.com>
11637 * machmode.h (int_mode_for_size): New function.
11638 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11639 instead of mode_for_size.
11640 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11642 * combine.c (expand_field_assignment): Use int_mode_for_size
11643 instead of mode_for_size.
11644 (make_extraction): Likewise.
11645 (simplify_shift_const_1): Likewise.
11646 (simplify_comparison): Likewise.
11647 * dojump.c (do_jump): Likewise.
11648 * dwarf2out.c (mem_loc_descriptor): Likewise.
11649 * emit-rtl.c (init_derived_machine_modes): Likewise.
11650 * expmed.c (flip_storage_order): Likewise.
11651 (convert_extracted_bit_field): Likewise.
11652 * expr.c (copy_blkmode_from_reg): Likewise.
11653 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11654 * internal-fn.c (expand_mul_overflow): Likewise.
11655 * lower-subreg.c (simple_move): Likewise.
11656 * optabs-libfuncs.c (init_optabs): Likewise.
11657 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11658 * tree.c (vector_type_mode): Likewise.
11659 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11660 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11661 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11662 * tree-vect-stmts.c (vectorizable_load): Likewise.
11663 (vectorizable_store): Likewise.
11665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11666 Alan Hayward <alan.hayward@arm.com>
11667 David Sherwood <david.sherwood@arm.com>
11669 * coretypes.h (pod_mode): New type.
11670 (scalar_int_mode_pod): New typedef.
11671 * machmode.h (pod_mode): New class.
11672 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11673 * genmodes.c (emit_mode_int_n): Update accordingly.
11674 * lower-subreg.h (target_lower_subreg): Change type to
11675 scalar_int_mode_pod.
11676 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11677 scalar_int_mode_pod.
11679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11680 Alan Hayward <alan.hayward@arm.com>
11681 David Sherwood <david.sherwood@arm.com>
11683 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11684 machine_mode to scalar_int_mode.
11685 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11686 (rs6000_option_override_internal): Remove cast to int.
11687 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11688 machine_mode to scalar_int_mode.
11689 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11690 (rs6000_option_override_internal): Remove cast to int.
11691 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11692 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11694 * config/s390/s390.c (s390_expand_builtin): Likewise.
11695 * coretypes.h (scalar_int_mode): New type.
11696 (opt_scalar_int_mode): New typedef.
11697 * machmode.h (scalar_int_mode): New class.
11698 (scalar_int_mode::includes_p): New function.
11699 (byte_mode): Change type to scalar_int_mode.
11700 (word_mode): Likewise.
11701 (ptr_mode): Likewise.
11702 * emit-rtl.c (byte_mode): Likewise.
11703 (word_mode): Likewise.
11704 (ptr_mode): Likewise.
11705 (init_derived_machine_modes): Update accordingly.
11706 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11707 and MODE_PARTIAL_INT.
11708 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11709 opt_scalar_int_mode.
11711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11712 Alan Hayward <alan.hayward@arm.com>
11713 David Sherwood <david.sherwood@arm.com>
11715 * target.def (libgcc_floating_mode_supported_p): Take a
11717 * doc/tm.texi: Regenerate.
11718 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11720 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11721 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11725 Alan Hayward <alan.hayward@arm.com>
11726 David Sherwood <david.sherwood@arm.com>
11728 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11729 * doc/tm.texi: Regenerate.
11730 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11731 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11732 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11733 * targhooks.h (default_floatn_mode): Likewise.
11734 * targhooks.c (default_floatn_mode): Likewise.
11735 * tree.c (build_common_tree_nodes): Update accordingly.
11737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11738 Alan Hayward <alan.hayward@arm.com>
11739 David Sherwood <david.sherwood@arm.com>
11741 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11742 (mode_iterator::iterate_p): Likewise.
11743 (mode_iterator::get_wider): Likewise.
11744 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11747 Alan Hayward <alan.hayward@arm.com>
11748 David Sherwood <david.sherwood@arm.com>
11750 * coretypes.h (opt_scalar_float_mode): New typedef.
11751 * machmode.h (float_mode_for_size): New function.
11752 * emit-rtl.c (double_mode): Delete.
11753 (init_emit_once): Use float_mode_for_size.
11754 * stor-layout.c (layout_type): Likewise.
11755 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11758 Alan Hayward <alan.hayward@arm.com>
11759 David Sherwood <david.sherwood@arm.com>
11761 * output.h (assemble_real): Take a scalar_float_mode.
11762 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11763 * config/arm/arm.md (consttable_4): Likewise.
11764 (consttable_8): Likewise.
11765 (consttable_16): Likewise.
11766 * config/mips/mips.md (consttable_float): Likewise.
11767 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11768 * varasm.c (assemble_real): Take a scalar_float_mode.
11769 (output_constant_pool_2): Update accordingly.
11770 (output_constant): Likewise.
11772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11773 Alan Hayward <alan.hayward@arm.com>
11774 David Sherwood <david.sherwood@arm.com>
11776 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11777 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11778 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11779 (native_encode_real): Likewise.
11780 (native_interpret_real): Likewise.
11781 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11782 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11785 Alan Hayward <alan.hayward@arm.com>
11786 David Sherwood <david.sherwood@arm.com>
11788 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11789 <scalar_float_mode>. Simplify.
11790 (gen_extend_conv_libfunc): Likewise.
11792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11793 Alan Hayward <alan.hayward@arm.com>
11794 David Sherwood <david.sherwood@arm.com>
11796 * coretypes.h (scalar_float_mode): New type.
11797 * machmode.h (mode_traits::from_int): Use machine_mode if
11798 USE_ENUM_MODES is defined.
11799 (is_a): New function.
11801 (dyn_cast): Likewise.
11802 (scalar_float_mode): New class.
11803 (scalar_float_mode::includes_p): New function.
11804 (is_float_mode): Likewise.
11805 * gdbhooks.py (MachineModePrinter): New class.
11806 (build_pretty_printer): Use it for scalar_float_mode.
11807 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11808 (format_helper::format_helper): Turn into a template.
11809 * genmodes.c (get_mode_class): New function.
11810 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11811 or machine_mode if none.
11812 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11813 as_a <scalar_float_mode>.
11814 * dwarf2out.c (mem_loc_descriptor): Likewise.
11815 (insert_float): Likewise.
11816 (add_const_value_attribute): Likewise.
11817 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11818 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11819 (expand_unop): Update accordingly.
11820 (expand_abs_nojump): Likewise.
11821 (expand_copysign_absneg): Take a scalar_float_mode.
11822 (expand_copysign_bit): Likewise.
11823 (expand_copysign): Update accordingly.
11825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11826 Alan Hayward <alan.hayward@arm.com>
11827 David Sherwood <david.sherwood@arm.com>
11829 * coretypes.h (opt_mode): New class.
11830 * machmode.h (opt_mode): Likewise.
11831 (opt_mode::else_void): New function.
11832 (opt_mode::require): Likewise.
11833 (opt_mode::exists): Likewise.
11834 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11835 (GET_MODE_2XWIDER_MODE): Likewise.
11836 (mode_iterator::get_wider): Update accordingly.
11837 (mode_iterator::get_2xwider): Likewise.
11838 (mode_iterator::get_known_wider): Likewise, turning into a template.
11839 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11840 forcing a wider mode to exist.
11841 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11842 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11843 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11844 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11845 * lower-subreg.c (init_lower_subreg): Likewise.
11846 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11847 on the final iteration.
11848 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11849 a wider mode exists before asking for a move pattern.
11850 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11851 forcing a wider mode to exist.
11852 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11853 returning false if no such mode exists.
11854 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11855 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11856 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11857 Avoid checking for a MODE_INT if we already know the mode is not a
11859 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11860 forcing a wider mode to exist.
11861 (expmed_mult_highpart_optab): Likewise.
11862 (expmed_mult_highpart): Likewise.
11863 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11865 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11866 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11867 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11868 * internal-fn.c (expand_mul_overflow): Update use of
11869 GET_MODE_2XWIDER_MODE.
11870 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11871 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11872 GET_MODE_WIDER_MODE.
11873 (convert_plusminus_to_widen): Likewise.
11874 * tree-switch-conversion.c (array_value_type): Likewise.
11875 * var-tracking.c (emit_note_insn_var_location): Likewise.
11876 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11877 Return false inside rather than outside the loop if no wider mode
11879 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11880 and GET_MODE_2XWIDER_MODE
11881 (can_compare_p): Use else_void.
11882 * gdbhooks.py (OptMachineModePrinter): New class.
11883 (build_pretty_printer): Use it for opt_mode.
11885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11886 Alan Hayward <alan.hayward@arm.com>
11887 David Sherwood <david.sherwood@arm.com>
11889 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11890 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11893 Alan Hayward <alan.hayward@arm.com>
11894 David Sherwood <david.sherwood@arm.com>
11896 * machmode.h (mode_traits): New structure.
11897 (get_narrowest_mode): New function.
11898 (mode_iterator::start): Likewise.
11899 (mode_iterator::iterate_p): Likewise.
11900 (mode_iterator::get_wider): Likewise.
11901 (mode_iterator::get_known_wider): Likewise.
11902 (mode_iterator::get_2xwider): Likewise.
11903 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11904 (FOR_EACH_MODE): Likewise.
11905 (FOR_EACH_MODE_FROM): Likewise.
11906 (FOR_EACH_MODE_UNTIL): Likewise.
11907 (FOR_EACH_WIDER_MODE): Likewise.
11908 (FOR_EACH_2XWIDER_MODE): Likewise.
11909 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11910 * combine.c (simplify_comparison): Likewise
11911 * config/i386/i386.c (type_natural_mode): Likewise.
11912 * cse.c (cse_insn): Likewise.
11913 * dse.c (find_shift_sequence): Likewise.
11914 * emit-rtl.c (init_derived_machine_modes): Likewise.
11915 (init_emit_once): Likewise.
11916 * explow.c (hard_function_value): Likewise.
11917 * expmed.c (extract_fixed_bit_field_1): Likewise.
11918 (extract_bit_field_1): Likewise.
11919 (expand_divmod): Likewise.
11920 (emit_store_flag_1): Likewise.
11921 * expr.c (init_expr_target): Likewise.
11922 (convert_move): Likewise.
11923 (alignment_for_piecewise_move): Likewise.
11924 (widest_int_mode_for_size): Likewise.
11925 (emit_block_move_via_movmem): Likewise.
11926 (copy_blkmode_to_reg): Likewise.
11927 (set_storage_via_setmem): Likewise.
11928 (compress_float_constant): Likewise.
11929 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11930 * optabs-query.c (get_best_extraction_insn): Likewise.
11931 * optabs.c (expand_binop): Likewise.
11932 (expand_twoval_unop): Likewise.
11933 (expand_twoval_binop): Likewise.
11934 (widen_leading): Likewise.
11935 (widen_bswap): Likewise.
11936 (expand_parity): Likewise.
11937 (expand_unop): Likewise.
11938 (prepare_cmp_insn): Likewise.
11939 (prepare_float_lib_cmp): Likewise.
11940 (expand_float): Likewise.
11941 (expand_fix): Likewise.
11942 (expand_sfix_optab): Likewise.
11943 * postreload.c (move2add_use_add2_insn): Likewise.
11944 * reg-stack.c (reg_to_stack): Likewise.
11945 * reginfo.c (choose_hard_reg_mode): Likewise.
11946 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11947 * stor-layout.c (mode_for_size): Likewise.
11948 (smallest_mode_for_size): Likewise.
11949 (mode_for_vector): Likewise.
11950 (finish_bitfield_representative): Likewise.
11951 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11952 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11953 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11954 * var-tracking.c (prepare_call_arguments): Likewise.
11956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11957 Alan Hayward <alan.hayward@arm.com>
11958 David Sherwood <david.sherwood@arm.com>
11960 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11961 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11962 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11963 * machmode.h (mode_size): Move earlier in file.
11964 (mode_precision): Likewise.
11965 (mode_inner): Likewise.
11966 (mode_nunits): Likewise.
11967 (mode_unit_size): Likewise.
11968 (unit_unit_precision): Likewise.
11969 (mode_wider): Likewise.
11970 (mode_2xwider): Likewise.
11971 (machine_mode): New class.
11972 (mode_to_bytes): New function.
11973 (mode_to_bits): Likewise.
11974 (mode_to_precision): Likewise.
11975 (mode_to_inner): Likewise.
11976 (mode_to_unit_size): Likewise.
11977 (mode_to_unit_precision): Likewise.
11978 (mode_to_nunits): Likewise.
11979 (GET_MODE_SIZE): Use mode_to_bytes.
11980 (GET_MODE_BITSIZE): Use mode_to_bits.
11981 (GET_MODE_PRECISION): Use mode_to_precision.
11982 (GET_MODE_INNER): Use mode_to_inner.
11983 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11984 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11985 (GET_MODE_NUNITS): Use mode_to_nunits.
11986 * system.h (ALWAYS_INLINE): New macro.
11987 * config/powerpcspe/powerpcspe-c.c
11988 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11989 int for arg1_mode and arg2_mode.
11991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11992 Alan Hayward <alan.hayward@arm.com>
11993 David Sherwood <david.sherwood@arm.com>
11995 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11996 Prefix mode names with E_ in case statements.
11997 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11998 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11999 (aarch64_split_simd_move): Likewise.
12000 (aarch64_gen_storewb_pair): Likewise.
12001 (aarch64_gen_loadwb_pair): Likewise.
12002 (aarch64_gen_store_pair): Likewise.
12003 (aarch64_gen_load_pair): Likewise.
12004 (aarch64_get_condition_code_1): Likewise.
12005 (aarch64_constant_pool_reload_icode): Likewise.
12006 (get_rsqrte_type): Likewise.
12007 (get_rsqrts_type): Likewise.
12008 (get_recpe_type): Likewise.
12009 (get_recps_type): Likewise.
12010 (aarch64_gimplify_va_arg_expr): Likewise.
12011 (aarch64_simd_container_mode): Likewise.
12012 (aarch64_emit_load_exclusive): Likewise.
12013 (aarch64_emit_store_exclusive): Likewise.
12014 (aarch64_expand_compare_and_swap): Likewise.
12015 (aarch64_gen_atomic_cas): Likewise.
12016 (aarch64_emit_bic): Likewise.
12017 (aarch64_emit_atomic_swap): Likewise.
12018 (aarch64_emit_atomic_load_op): Likewise.
12019 (aarch64_evpc_trn): Likewise.
12020 (aarch64_evpc_uzp): Likewise.
12021 (aarch64_evpc_zip): Likewise.
12022 (aarch64_evpc_ext): Likewise.
12023 (aarch64_evpc_rev): Likewise.
12024 (aarch64_evpc_dup): Likewise.
12025 (aarch64_gen_ccmp_first): Likewise.
12026 (aarch64_gen_ccmp_next): Likewise.
12027 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12028 (alpha_emit_xfloating_libcall): Likewise.
12029 (emit_insxl): Likewise.
12030 (alpha_arg_type): Likewise.
12031 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12032 (arc_preferred_simd_mode): Likewise.
12033 (arc_secondary_reload): Likewise.
12034 (get_arc_condition_code): Likewise.
12035 (arc_print_operand): Likewise.
12036 (arc_legitimate_constant_p): Likewise.
12037 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12038 * config/arc/arc.md (casesi_load): Likewise.
12039 (casesi_compact_jump): Likewise.
12040 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12041 (cc_use_register): Likewise.
12042 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12043 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12044 (arm_init_iwmmxt_builtins): Likewise.
12045 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12046 (neon_expand_vector_init): Likewise.
12047 (arm_attr_length_move_neon): Likewise.
12048 (maybe_get_arm_condition_code): Likewise.
12049 (arm_emit_vector_const): Likewise.
12050 (arm_preferred_simd_mode): Likewise.
12051 (arm_output_iwmmxt_tinsr): Likewise.
12052 (thumb1_output_casesi): Likewise.
12053 (thumb2_output_casesi): Likewise.
12054 (arm_emit_load_exclusive): Likewise.
12055 (arm_emit_store_exclusive): Likewise.
12056 (arm_expand_compare_and_swap): Likewise.
12057 (arm_evpc_neon_vuzp): Likewise.
12058 (arm_evpc_neon_vzip): Likewise.
12059 (arm_evpc_neon_vrev): Likewise.
12060 (arm_evpc_neon_vtrn): Likewise.
12061 (arm_evpc_neon_vext): Likewise.
12062 (arm_validize_comparison): Likewise.
12063 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12064 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12065 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12066 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12067 (c6x_preferred_simd_mode): Likewise.
12068 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12069 (epiphany_rtx_costs): Likewise.
12070 * config/epiphany/predicates.md (proper_comparison_operator):
12072 * config/frv/frv.c (condexec_memory_operand): Likewise.
12073 (frv_emit_move): Likewise.
12074 (output_move_single): Likewise.
12075 (output_condmove_single): Likewise.
12076 (frv_hard_regno_mode_ok): Likewise.
12077 (frv_matching_accg_mode): Likewise.
12078 * config/h8300/h8300.c (split_adds_subs): Likewise.
12079 (h8300_rtx_costs): Likewise.
12080 (h8300_print_operand): Likewise.
12081 (compute_mov_length): Likewise.
12082 (output_logical_op): Likewise.
12083 (compute_logical_op_length): Likewise.
12084 (compute_logical_op_cc): Likewise.
12085 (h8300_shift_needs_scratch_p): Likewise.
12086 (output_a_shift): Likewise.
12087 (compute_a_shift_length): Likewise.
12088 (compute_a_shift_cc): Likewise.
12089 (expand_a_rotate): Likewise.
12090 (output_a_rotate): Likewise.
12091 * config/i386/i386.c (classify_argument): Likewise.
12092 (function_arg_advance_32): Likewise.
12093 (function_arg_32): Likewise.
12094 (function_arg_64): Likewise.
12095 (function_value_64): Likewise.
12096 (ix86_gimplify_va_arg): Likewise.
12097 (ix86_legitimate_constant_p): Likewise.
12098 (put_condition_code): Likewise.
12099 (split_double_mode): Likewise.
12100 (ix86_avx256_split_vector_move_misalign): Likewise.
12101 (ix86_expand_vector_logical_operator): Likewise.
12102 (ix86_split_idivmod): Likewise.
12103 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12104 (ix86_build_const_vector): Likewise.
12105 (ix86_build_signbit_mask): Likewise.
12106 (ix86_match_ccmode): Likewise.
12107 (ix86_cc_modes_compatible): Likewise.
12108 (ix86_expand_branch): Likewise.
12109 (ix86_expand_sse_cmp): Likewise.
12110 (ix86_expand_sse_movcc): Likewise.
12111 (ix86_expand_int_sse_cmp): Likewise.
12112 (ix86_expand_vec_perm_vpermi2): Likewise.
12113 (ix86_expand_vec_perm): Likewise.
12114 (ix86_expand_sse_unpack): Likewise.
12115 (ix86_expand_int_addcc): Likewise.
12116 (ix86_split_to_parts): Likewise.
12117 (ix86_vectorize_builtin_gather): Likewise.
12118 (ix86_vectorize_builtin_scatter): Likewise.
12119 (avx_vpermilp_parallel): Likewise.
12120 (inline_memory_move_cost): Likewise.
12121 (ix86_tieable_integer_mode_p): Likewise.
12122 (x86_maybe_negate_const_int): Likewise.
12123 (ix86_expand_vector_init_duplicate): Likewise.
12124 (ix86_expand_vector_init_one_nonzero): Likewise.
12125 (ix86_expand_vector_init_one_var): Likewise.
12126 (ix86_expand_vector_init_concat): Likewise.
12127 (ix86_expand_vector_init_interleave): Likewise.
12128 (ix86_expand_vector_init_general): Likewise.
12129 (ix86_expand_vector_set): Likewise.
12130 (ix86_expand_vector_extract): Likewise.
12131 (emit_reduc_half): Likewise.
12132 (ix86_emit_i387_round): Likewise.
12133 (ix86_mangle_type): Likewise.
12134 (ix86_expand_round_sse4): Likewise.
12135 (expand_vec_perm_blend): Likewise.
12136 (canonicalize_vector_int_perm): Likewise.
12137 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12138 (expand_vec_perm_1): Likewise.
12139 (expand_vec_perm_interleave3): Likewise.
12140 (expand_vec_perm_even_odd_pack): Likewise.
12141 (expand_vec_perm_even_odd_1): Likewise.
12142 (expand_vec_perm_broadcast_1): Likewise.
12143 (ix86_vectorize_vec_perm_const_ok): Likewise.
12144 (ix86_expand_vecop_qihi): Likewise.
12145 (ix86_expand_mul_widen_hilo): Likewise.
12146 (ix86_expand_sse2_abs): Likewise.
12147 (ix86_expand_pextr): Likewise.
12148 (ix86_expand_pinsr): Likewise.
12149 (ix86_preferred_simd_mode): Likewise.
12150 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12151 * config/i386/sse.md (*andnot<mode>3): Likewise.
12152 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12153 (*<code><mode>3): Likewise.
12154 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12155 (ia64_expand_atomic_op): Likewise.
12156 (ia64_arg_type): Likewise.
12157 (ia64_mode_to_int): Likewise.
12158 (ia64_scalar_mode_supported_p): Likewise.
12159 (ia64_vector_mode_supported_p): Likewise.
12160 (expand_vec_perm_broadcast): Likewise.
12161 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12162 (iq2000_function_arg_advance): Likewise.
12163 (iq2000_function_arg): Likewise.
12164 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12165 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12166 (m68k_libcall_value): Likewise.
12167 (m68k_function_value): Likewise.
12168 (sched_attr_op_type): Likewise.
12169 * config/mcore/mcore.c (mcore_output_move): Likewise.
12170 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12172 (microblaze_function_arg): Likewise.
12173 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12174 (mips_print_operand): Likewise.
12175 (mips_mode_ok_for_mov_fmt_p): Likewise.
12176 (mips_vector_mode_supported_p): Likewise.
12177 (mips_preferred_simd_mode): Likewise.
12178 (mips_expand_vpc_loongson_even_odd): Likewise.
12179 (mips_expand_vec_unpack): Likewise.
12180 (mips_expand_vi_broadcast): Likewise.
12181 (mips_expand_vector_init): Likewise.
12182 (mips_expand_vec_reduc): Likewise.
12183 (mips_expand_msa_cmp): Likewise.
12184 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12185 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12186 (cc_flags_for_mode): Likewise.
12187 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12188 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12189 (nds32_output_casesi_pc_relative): Likewise.
12190 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12191 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12192 (nvptx_gen_unpack): Likewise.
12193 (nvptx_gen_pack): Likewise.
12194 (nvptx_gen_shuffle): Likewise.
12195 (nvptx_gen_wcast): Likewise.
12196 (nvptx_preferred_simd_mode): Likewise.
12197 * config/pa/pa.c (pa_secondary_reload): Likewise.
12198 * config/pa/predicates.md (base14_operand): Likewise.
12199 * config/powerpcspe/powerpcspe-c.c
12200 (altivec_resolve_overloaded_builtin): Likewise.
12201 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12203 (rs6000_preferred_simd_mode): Likewise.
12204 (output_vec_const_move): Likewise.
12205 (rs6000_expand_vector_extract): Likewise.
12206 (rs6000_split_vec_extract_var): Likewise.
12207 (reg_offset_addressing_ok_p): Likewise.
12208 (rs6000_legitimate_offset_address_p): Likewise.
12209 (rs6000_legitimize_address): Likewise.
12210 (rs6000_emit_set_const): Likewise.
12211 (rs6000_const_vec): Likewise.
12212 (rs6000_emit_move): Likewise.
12213 (spe_build_register_parallel): Likewise.
12214 (rs6000_darwin64_record_arg_recurse): Likewise.
12215 (swap_selector_for_mode): Likewise.
12216 (spe_init_builtins): Likewise.
12217 (paired_init_builtins): Likewise.
12218 (altivec_init_builtins): Likewise.
12219 (do_load_for_compare): Likewise.
12220 (rs6000_generate_compare): Likewise.
12221 (rs6000_expand_float128_convert): Likewise.
12222 (emit_load_locked): Likewise.
12223 (emit_store_conditional): Likewise.
12224 (rs6000_output_function_epilogue): Likewise.
12225 (rs6000_handle_altivec_attribute): Likewise.
12226 (rs6000_function_value): Likewise.
12227 (emit_fusion_gpr_load): Likewise.
12228 (emit_fusion_p9_load): Likewise.
12229 (emit_fusion_p9_store): Likewise.
12230 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12231 (fusion_gpr_mem_load): Likewise.
12232 (fusion_addis_mem_combo_load): Likewise.
12233 (fusion_addis_mem_combo_store): Likewise.
12234 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12235 (fusion_gpr_mem_load): Likewise.
12236 (fusion_addis_mem_combo_load): Likewise.
12237 (fusion_addis_mem_combo_store): Likewise.
12238 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12240 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12241 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12242 (rs6000_preferred_simd_mode): Likewise.
12243 (output_vec_const_move): Likewise.
12244 (rs6000_expand_vector_extract): Likewise.
12245 (rs6000_split_vec_extract_var): Likewise.
12246 (reg_offset_addressing_ok_p): Likewise.
12247 (rs6000_legitimate_offset_address_p): Likewise.
12248 (rs6000_legitimize_address): Likewise.
12249 (rs6000_emit_set_const): Likewise.
12250 (rs6000_const_vec): Likewise.
12251 (rs6000_emit_move): Likewise.
12252 (rs6000_darwin64_record_arg_recurse): Likewise.
12253 (swap_selector_for_mode): Likewise.
12254 (paired_init_builtins): Likewise.
12255 (altivec_init_builtins): Likewise.
12256 (rs6000_expand_float128_convert): Likewise.
12257 (emit_load_locked): Likewise.
12258 (emit_store_conditional): Likewise.
12259 (rs6000_output_function_epilogue): Likewise.
12260 (rs6000_handle_altivec_attribute): Likewise.
12261 (rs6000_function_value): Likewise.
12262 (emit_fusion_gpr_load): Likewise.
12263 (emit_fusion_p9_load): Likewise.
12264 (emit_fusion_p9_store): Likewise.
12265 * config/rx/rx.c (rx_gen_move_template): Likewise.
12266 (flags_from_mode): Likewise.
12267 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12268 (s390_slb_comparison): Likewise.
12269 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12270 (s390_vector_mode_supported_p): Likewise.
12271 (s390_cc_modes_compatible): Likewise.
12272 (s390_match_ccmode_set): Likewise.
12273 (s390_canonicalize_comparison): Likewise.
12274 (s390_emit_compare_and_swap): Likewise.
12275 (s390_branch_condition_mask): Likewise.
12276 (s390_rtx_costs): Likewise.
12277 (s390_secondary_reload): Likewise.
12278 (__SECONDARY_RELOAD_CASE): Likewise.
12279 (s390_expand_cs): Likewise.
12280 (s390_preferred_simd_mode): Likewise.
12281 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12282 * config/sh/sh.c (sh_print_operand): Likewise.
12283 (dump_table): Likewise.
12284 (sh_secondary_reload): Likewise.
12285 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12286 * config/sh/sh.md (casesi_worker_1): Likewise.
12287 (casesi_worker_2): Likewise.
12288 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12289 (fcc_comparison_operator): Likewise.
12290 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12291 (emit_soft_tfmode_cvt): Likewise.
12292 (sparc_preferred_simd_mode): Likewise.
12293 (output_cbranch): Likewise.
12294 (sparc_print_operand): Likewise.
12295 (sparc_expand_vec_perm_bmask): Likewise.
12296 (vector_init_bshuffle): Likewise.
12297 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12298 (spu_vector_mode_supported_p): Likewise.
12299 (spu_expand_insv): Likewise.
12300 (spu_emit_branch_or_set): Likewise.
12301 (spu_handle_vector_attribute): Likewise.
12302 (spu_builtin_splats): Likewise.
12303 (spu_builtin_extract): Likewise.
12304 (spu_builtin_promote): Likewise.
12305 (spu_expand_sign_extend): Likewise.
12306 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12307 (tilegx_simd_int): Likewise.
12308 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12309 (tilepro_simd_int): Likewise.
12310 * config/v850/v850.c (const_double_split): Likewise.
12311 (v850_print_operand): Likewise.
12312 (ep_memory_offset): Likewise.
12313 * config/vax/vax.c (vax_rtx_costs): Likewise.
12314 (vax_output_int_move): Likewise.
12315 (vax_output_int_add): Likewise.
12316 (vax_output_int_subtract): Likewise.
12317 * config/visium/predicates.md (visium_branch_operator): Likewise.
12318 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12319 (visium_print_operand_address): Likewise.
12320 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12321 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12322 (xtensa_expand_conditional_branch): Likewise.
12323 (xtensa_copy_incoming_a7): Likewise.
12324 (xtensa_output_literal): Likewise.
12325 * dfp.c (decimal_real_maxval): Likewise.
12326 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12329 Alan Hayward <alan.hayward@arm.com>
12330 David Sherwood <david.sherwood@arm.com>
12332 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12333 (mode_nunits_inline): Likewise.
12334 (mode_inner_inline): Likewise.
12335 (mode_unit_size_inline): Likewise.
12336 (mode_unit_precision_inline): Likewise.
12337 (emit_insn_modes_h): Likewise. Also emit a #define of the
12339 (emit_mode_wider): Add an E_ prefix to mode names.
12340 (emit_mode_complex): Likewise.
12341 (emit_mode_inner): Likewise.
12342 (emit_mode_adjustments): Likewise.
12343 (emit_mode_int_n): Likewise.
12344 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12345 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12346 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12347 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12348 (CRC32_BUILTIN, ENTRY): Likewise.
12349 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12350 (aarch64_pop_regs): Likewise.
12351 (aarch64_process_components): Likewise.
12352 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12353 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12354 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12355 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12356 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12357 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12359 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12360 (pcmpistr): Likewise.
12361 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12362 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12363 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12365 * config/rl78/rl78.c (mduc_regs): Likewise.
12366 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12367 (htm_expand_builtin): Likewise.
12368 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12369 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12370 * config/xtensa/xtensa.c (print_operand): Likewise.
12371 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12372 (NUM_MODE_VECTOR_INT): Likewise.
12373 * genoutput.c (null_operand): Likewise.
12374 (output_operand_data): Likewise.
12375 * genrecog.c (print_parameter_value): Likewise.
12376 * lra.c (debug_operand_data): Likewise.
12378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12380 * dwarf2out.c (scompare_loc_descriptor_wide)
12381 (scompare_loc_descriptor_narrow): New functions, split out from...
12382 (scompare_loc_descriptor): ...here.
12383 * expmed.c (emit_store_flag_int): New function, split out from...
12384 (emit_store_flag): ...here.
12386 2017-08-30 Richard Biener <rguenther@suse.de>
12388 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12389 (dwarf2out_early_finish): Move setting of AT_pubnames from
12390 early debug output to early finish.
12392 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12394 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12395 and -mdata-region to the assembler.
12397 2017-08-30 Richard Biener <rguenther@suse.de>
12399 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12401 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12402 (dwarf2out_early_global_decl): Only generate a DIE for the
12403 abstract origin if it doesn't already exist or is a declaration DIE.
12404 (resolve_addr): Do not add the linkage name twice when
12405 generating a stub DIE for the DW_TAG_GNU_call_site target.
12407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12409 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12410 Use machine_mode rather than int for arg1_mode.
12412 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12415 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12416 that the second argument of the built-in functions to unpack
12417 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12418 switch statement instead a lot of if statements.
12419 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12420 Allow 64-bit values to be in Altivec registers as well as
12421 traditional floating point registers.
12422 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12424 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12426 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12427 MAX_REGS_PER_ADDRESS == 1.
12429 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12431 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12432 * config/i386/i386.c (ix86_option_override_internal): Simplify
12433 setting of opts->x_flag_entry.
12435 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12436 Jakub Jelinek <jakub@redhat.com>
12437 Richard Biener <rguenther@suse.de>
12439 PR tree-optimization/81503
12440 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12441 folded constant fits in the target type; reorder tests for clarity.
12443 2017-08-29 Martin Liska <mliska@suse.cz>
12445 * passes.def: Include pass_lower_switch.
12446 * stmt.c (dump_case_nodes): Remove and move to
12447 tree-switch-conversion.
12448 (case_values_threshold): Likewise.
12449 (expand_switch_as_decision_tree_p): Likewise.
12450 (emit_case_decision_tree): Likewise.
12451 (expand_case): Likewise.
12452 (balance_case_nodes): Likewise.
12453 (node_has_low_bound): Likewise.
12454 (node_has_high_bound): Likewise.
12455 (node_is_bounded): Likewise.
12456 (emit_case_nodes): Likewise.
12457 (struct simple_case_node): New struct.
12458 (add_case_node): Remove.
12459 (emit_case_dispatch_table): Use vector instead of case_list.
12460 (reset_out_edges_aux): Remove.
12461 (compute_cases_per_edge): Likewise.
12462 (expand_case): Build list of simple_case_node.
12463 (expand_sjlj_dispatch_table): Use it.
12464 * tree-switch-conversion.c (struct case_node): Moved from
12465 stmt.c and adjusted.
12466 (emit_case_nodes): Likewise.
12467 (node_has_low_bound): Likewise.
12468 (node_has_high_bound): Likewise.
12469 (node_is_bounded): Likewise.
12470 (case_values_threshold): Likewise.
12471 (reset_out_edges_aux): Likewise.
12472 (compute_cases_per_edge): Likewise.
12473 (add_case_node): Likewise.
12474 (dump_case_nodes): Likewise.
12475 (balance_case_nodes): Likewise.
12476 (expand_switch_as_decision_tree_p): Likewise.
12477 (emit_jump): Likewise.
12478 (emit_case_decision_tree): Likewise.
12479 (try_switch_expansion): Likewise.
12480 (do_jump_if_equal): Likewise.
12481 (emit_cmp_and_jump_insns): Likewise.
12482 (fix_phi_operands_for_edge): New function.
12483 (record_phi_operand_mapping): Likewise.
12484 (class pass_lower_switch): New pass.
12485 (pass_lower_switch::execute): New function.
12486 (make_pass_lower_switch): Likewise.
12487 (conditional_probability):
12488 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12489 * tree-pass.h: Add make_pass_lower_switch.
12491 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12494 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12497 2017-08-29 Richard Biener <rguenther@suse.de>
12499 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12500 we do not add a DW_AT_inline attribute twice.
12501 (gen_subprogram_die): Remove code setting DW_AT_inline on
12502 DECL_ABSTRACT_P nodes.
12504 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12506 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12507 calls to internal functions.
12508 (gimplify_modify_expr): Likewise.
12509 * tree-call-cdce.c (use_internal_fn): Likewise.
12510 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12511 (convert_to_divmod): Set the nothrow flag.
12512 * tree-if-conv.c (predicate_mem_writes): Likewise.
12513 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12514 (vectorizable_call): Likewise.
12515 (vectorizable_store): Likewise.
12516 (vectorizable_load): Likewise.
12517 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12518 (vect_recog_mask_conversion_pattern): Likewise.
12520 2017-08-29 Martin Liska <mliska@suse.cz>
12523 * gcc.c (driver_handle_option): Add new argument.
12524 * opts-common.c (handle_option): Pass
12525 target_option_override_hook.
12526 * opts-global.c (lang_handle_option): Add new option.
12527 (set_default_handlers): Add new argument.
12528 (decode_options): Likewise.
12529 * opts.c (target_handle_option): Likewise.
12530 (common_handle_option): Call target_option_override_hook.
12531 * opts.h (struct cl_option_handler_func): Add hook for
12532 target option override.
12533 (struct cl_option_handlers): Likewise.
12534 (set_default_handlers): Add new argument.
12535 (decode_options): Likewise.
12536 (common_handle_option): Likewise.
12537 (target_handle_option): Likewise.
12538 * toplev.c (toplev::main): Pass targetm.target_option.override
12541 2017-08-29 Richard Biener <rguenther@suse.de>
12542 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12544 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12545 life to the active subtree.
12547 2017-08-28 Jeff Law <law@redhat.com>
12549 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12550 derive_equivalences.
12551 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12552 Code moved into....
12553 (edge_info::derive_equivalences): New private member function
12555 * tree-ssa-dom.c (class edge_info): Changed from a struct
12556 to a class. Add ctor/dtor, methods and data members.
12557 (edge_info::edge_info): Renamed from allocate_edge_info.
12558 Initialize additional members.
12559 (edge_info::~edge_info): New.
12560 (free_dom_edge_info): Delete the edge info.
12561 (record_edge_info): Use new class & associated member functions.
12562 Tighten forms for testing for edge equivalences.
12563 (record_temporary_equivalences): Iterate over the simple
12564 equivalences rather than assuming there's only one per edge.
12565 (cprop_into_successor_phis): Iterate over the simple
12566 equivalences rather than assuming there's only one per edge.
12567 (optimize_stmt): Use operand_equal_p rather than pointer
12568 equality for mini-DSE code.
12570 2017-08-28 Nathan Sidwell <nathan@acm.org>
12572 * gcc.c (execute): Fold SIGPIPE handling into switch
12573 statement. Adjust internal error message.
12575 2017-08-28 Richard Biener <rguenther@suse.de>
12578 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12579 Do nothing for removed DIEs.
12581 2017-08-28 Richard Biener <rguenther@suse.de>
12583 PR tree-optimization/81977
12584 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12587 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12590 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12591 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12592 using targetm.gen_mem_thread_fence.
12594 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12597 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12598 predicate to register_operand. Reorder operands.
12599 (*btr<mode>): Ditto.
12600 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12601 (*btr<mode>_mask): Ditto.
12603 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12605 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12606 * config/rs6000/xmmintrin.h: New file.
12607 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12609 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12612 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12613 parameter and_insn and return it.
12614 (recombine_lvx_pattern): Insert a copy to ensure availability of
12615 the base register of the copied masking operation at the point of
12616 the instruction replacement.
12617 (recombine_stvx_pattern): Likewise.
12619 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12621 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12622 undocumented switches.
12623 (-mpower9-dform-vector): Likewise.
12624 (-mpower9-dform): Likewise.
12625 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12626 comments to delete references to -mpower9-dform* switches.
12627 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12628 Delete reference to -mpower9-dform* switches, test for
12629 -mpower9-vector instead.
12630 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12631 (OTHER_P9_VECTOR_MASKS): Likewise.
12632 (POWERPC_MASKS): Likewise.
12633 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12634 tests against -mpower9-dform* to -mpower9-vector. Delete code
12635 that checked for -mpower9-dform* consistancy with other options.
12636 Add test for -mpower9-misc to enable other power9 switches.
12637 (rs6000_init_hard_regno_mode_ok): Likewise.
12638 (rs6000_option_override_internal): Likewise.
12639 (rs6000_emit_prologue): Likewise.
12640 (rs6000_emit_epilogue): Likewise.
12641 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12642 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12643 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12645 (emit_fusion_p9_store): Likewise.
12646 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12647 resetting these macros if the assembler does not support ISA 3.0
12649 (TARGET_P9_DFORM_VECTOR): Likewise.
12650 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12651 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12653 2017-08-25 Alan Modra <amodra@gmail.com>
12655 PR rtl-optimization/81747
12656 * cse.c (cse_extended_basic_block): Don't attempt to record
12657 equivalences for degenerate conditional branches that branch
12658 to their fall-through.
12660 2017-08-24 Martin Sebor <msebor@redhat.com>
12662 PR middle-end/81908
12663 * gimple-fold.c (size_must_be_zero_p): New function.
12664 (gimple_fold_builtin_memory_op): Call it.
12666 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12668 * config/rs6000/mm_malloc.h: New file.
12670 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12672 PR tree-optimization/81913
12673 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12674 analysis when either IVs in condition can wrap.
12676 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12678 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12679 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12681 2017-08-24 Richard Biener <rguenther@suse.de>
12684 * targhooks.c (default_target_can_inline_p): Properly
12685 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12686 is present and always compare.
12687 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12688 infer -mfpmath=sse from TARGET_SSE_P.
12689 (ix86_can_inline_p): Properly use target_option_default_node when
12690 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12692 2017-08-24 Richard Biener <rguenther@suse.de>
12695 * dwarf2out.c (output_die): Handle flag_generate_offload like
12697 (output_comp_unit): Likewise.
12698 (gen_array_type_die): Likewise.
12699 (dwarf2out_early_finish): Likewise.
12700 (note_variable_value_in_expr): Likewise.
12701 (dwarf2out_finish): Likewise. Adjust assert.
12702 * cgraphunit.c (symbol_table::compile): Move setting of
12703 flag_generate_offload earlier ...
12704 (symbol_table::finalize_compilation_unit): ... here, before
12705 early debug finalization.
12707 2017-08-24 Richard Biener <rguenther@suse.de>
12709 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12710 and ipa-fnsummary.h.
12711 (ix86_can_inline_p): When ix86_fpmath flags do not match
12712 check whether the callee uses FP math at all.
12714 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12716 PR middle-end/81931
12717 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12718 instead of TYPE_PRECISION.
12720 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12721 Alan Hayward <alan.hayward@arm.com>
12722 David Sherwood <david.sherwood@arm.com>
12724 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12726 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12728 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12730 (ix86_frame::stack_realign_allocate): New field.
12731 (struct machine_frame_state): Modify comments.
12732 (machine_frame_state::sp_realigned_fp_end): New field.
12733 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12734 layout calculation.
12735 (sp_valid_at): Add assertion to assure no attempt to access invalid
12736 offset of a realigned stack.
12737 (fp_valid_at): Likewise.
12738 (choose_baseaddr): Modify comments.
12739 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12740 ix86_expand_prologue.
12741 (ix86_expand_prologue): Modify stack realignment and allocation.
12742 (ix86_expand_epilogue): Modify comments.
12743 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12744 avx2_runtime, avx512f, and avx512f_runtime.
12746 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12748 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12749 (mstackrealign): Do not init to -1.
12750 * config/i386/i386.c (ix86_option_override_internal):
12751 Check opts_set, not opts when setting default value of
12752 opts->x_ix86_force_align_arg_pointer.
12754 2017-08-23 Richard Biener <rguenther@suse.de>
12756 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12757 lang_hooks.decl_printable_name.
12758 * print-rtl-function.c (print_rtx_function): Likewise.
12759 * tree-pretty-print.c (dump_function_header): Likewise.
12761 2017-08-23 Richard Biener <rguenther@suse.de>
12764 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12765 -g0 at compile-time.
12767 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12769 PR middle-end/19706
12770 * doc/sourcebuild.texi (Other hardware attributes):
12773 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12775 PR middle-end/19706
12776 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12777 Fix single-use check.
12779 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12781 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12783 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12785 * doc/install.texi: Modify to add more details on running selected
12788 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12790 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12791 is combined with -mabi=ms.
12792 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12795 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12797 PR tree-optimization/81488
12798 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12799 and cached_basis fields.
12800 (MAX_SPREAD): New constant.
12801 (alloc_cand_and_find_basis): Initialize new fields.
12802 (clear_visited): New function.
12803 (create_phi_basis_1): Rename from create_phi_basis, set visited
12804 and cached_basis fields.
12805 (create_phi_basis): New wrapper function.
12806 (phi_add_costs_1): Rename from phi_add_costs, add spread
12807 parameter, set visited field, short-circuit when limits reached.
12808 (phi_add_costs): New wrapper function.
12809 (record_phi_increments_1): Rename from record_phi_increments, set
12811 (record_phi_increments): New wrapper function.
12812 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12813 (phi_incr_cost): New wrapper function.
12814 (all_phi_incrs_profitable_1): Rename from
12815 all_phi_incrs_profitable, set visited field.
12816 (all_phi_incrs_profitable): New wrapper function.
12818 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12819 Alan Hayward <alan.hayward@arm.com>
12820 David Sherwood <david.sherwood@arm.com>
12822 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12823 that takes the outer and inner modes.
12824 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12825 comparison as the canonical test for a paradoxical subreg.
12826 * combine.c (simplify_set): Use paradoxical_subreg_p.
12827 (make_extraction): Likewise.
12828 (force_to_mode): Likewise.
12829 (rtx_equal_for_field_assignment_p): Likewise.
12830 (gen_lowpart_for_combine): Likewise.
12831 (simplify_comparison): Likewise.
12832 * cse.c (equiv_constant): Likewise.
12833 * expmed.c (store_bit_field_1): Likewise.
12834 * final.c (alter_subreg): Likewise.
12835 * fwprop.c (propagate_rtx): Likewise.
12836 (forward_propagate_subreg): Likewise.
12837 * ira-conflicts.c (ira_build_conflicts): Likewise.
12838 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12839 * lra-constraints.c (curr_insn_transform): Likewise.
12840 (split_reg): Likewise.
12841 * lra-eliminations.c (move_plus_up): Likewise.
12842 (lra_eliminate_regs_1): Likewise.
12843 * recog.c (general_operand): Likewise.
12844 * ree.c (combine_reaching_defs): Likewise.
12845 * reload.c (push_reload): Likewise.
12846 (find_reloads): Likewise.
12847 * reload1.c (elimination_effects): Likewise.
12848 (compute_reload_subreg_offset): Likewise.
12849 (choose_reload_regs): Likewise.
12850 * rtlanal.c (subreg_lsb_1): Likewise.
12851 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12852 (simplify_subreg): Likewise.
12853 * var-tracking.c (track_loc_p): Likewise.
12854 * emit-rtl.c (byte_lowpart_offset): Likewise.
12855 (paradoxical_subreg_p): Delete out-of-line definition.
12857 2017-08-22 Jeff Law <law@redhat.com>
12859 PR tree-optimization/81741
12860 PR tree-optimization/71947
12861 * tree-ssa-dom.c: Include tree-inline.h.
12862 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12863 equivalences if one is more expensive to compute than the other.
12864 * tree-ssa-scopedtables.h (class const_or_copies): Make
12865 record_const_or_copy_raw method private.
12866 (class avail_exprs_stack): New method simplify_binary_operation.
12867 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12868 avail_exprs_stack::simplify_binary_operation as needed.
12869 (avail_exprs_stack::simplify_binary_operation): New function.
12871 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12873 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12874 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12875 (DOT_SYMBOLS): Likewise.
12876 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12877 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12878 (RS6000_ABI_NAME): Likewise.
12879 (TARGET_CMODEL): Likewise.
12880 (TOC_SECTION_ASM_OP): Likewise.
12881 (SET_CMODEL): New macro.
12882 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12884 2017-08-22 Richard Biener <rguenther@suse.de>
12886 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12887 to allow for free-lang-data replacements similar to verify_type_variant.
12889 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12890 Alan Hayward <alan.hayward@arm.com>
12891 David Sherwood <david.sherwood@arm.com>
12893 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12894 VOIDmode for the LABEL_REF.
12896 2017-08-22 Richard Biener <rguenther@suse.de>
12898 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12900 2017-08-22 Richard Biener <rguenther@suse.de>
12902 * common.opt (feliminate-dwarf2-dups): Ignore.
12903 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12904 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12905 same_die_p_wrap, compute_section_prefix,
12906 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12907 (comdat_symbol_id, comdat_symbol_number): Likewise.
12908 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12910 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12911 (output_die): Mark unreachable path unreachable.
12912 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12913 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12914 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12915 (dwarf2out_early_finish): Likewise.
12917 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12919 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12921 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12924 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12925 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12926 (avr_attribute_table) <io, io_low, address>: Initialize
12927 .decl_required with true.
12929 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12931 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12932 undocumented debugging options.
12933 (-mvsx-scalar-double): Likewise.
12934 (-mallow-df-permute): Likewise.
12935 (-mvectorize-builtins): Likewise.
12936 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12937 (rs6000_builtin_vectorized_function): Likewise.
12938 (rs6000_builtin_md_vectorized_function): Likewise.
12939 (rs6000_opt_vars): Likewise.
12941 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12944 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12945 (*btrq_imm): Rename from *btrq.
12946 (*btcq_imm): Rename from *btcq.
12947 (btsc): New code attribute.
12948 (*<btsc><mode>): New insn pattern.
12949 (*btr<mode>): Ditto.
12950 (*<btsc><mode>_mask): New insn_and_split pattern.
12951 (*btr<mode>_mask): Ditto.
12953 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12954 Alan Hayward <alan.hayward@arm.com>
12955 David Sherwood <david.sherwood@arm.com>
12957 * function.c (pad_below): Simplify padding calculation.
12959 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12960 Alan Hayward <alan.hayward@arm.com>
12961 David Sherwood <david.sherwood@arm.com>
12963 * target.def (function_prologue): Remove frame size argument.
12964 (function_epilogue): Likewise.
12965 * doc/tm.texi: Regenerate.
12966 * final.c (final_start_function): Update call to function_prologue.
12967 (final_end_function): Update call to function_epilogue.
12968 (default_function_pro_epilogue): Remove frame size argument.
12969 * output.h (default_function_pro_epilogue): Likewise.
12970 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12971 (arm_output_function_prologue): Likewise.
12972 * config/frv/frv.c (frv_function_prologue): Likewise.
12973 (frv_function_epilogue): Likewise.
12974 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12975 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12976 (ia64_output_function_epilogue): Likewise.
12977 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12978 (m32r_output_function_epilogue): Likewise.
12979 * config/microblaze/microblaze.c (microblaze_function_prologue)
12980 (microblaze_function_epilogue): Likewise.
12981 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12982 (mips_output_function_epilogue): Likewise.
12983 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12984 (mmix_target_asm_function_epilogue): Likewise.
12985 * config/msp430/msp430.c (msp430_start_function): Likewise.
12986 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12987 (nds32_asm_function_epilogue): Likewise.
12988 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12989 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12990 (pa_output_function_epilogue): Likewise.
12991 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12992 (rs6000_output_function_epilogue): Likewise.
12993 * config/rl78/rl78.c (rl78_start_function): Likewise.
12994 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12995 (rs6000_output_function_epilogue): Likewise.
12996 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12997 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12998 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12999 (sparc_asm_function_epilogue): Likewise.
13001 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13003 * tree.h (type_has_mode_precision_p): New function.
13004 * convert.c (convert_to_integer_1): Use it.
13005 * expr.c (expand_expr_real_2): Likewise.
13006 (expand_expr_real_1): Likewise.
13007 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13008 * match.pd: Likewise.
13009 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13010 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13011 * tree-tailcall.c (process_assignment): Likewise.
13012 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13013 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13014 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13015 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13016 (vectorizable_assignment): Likewise.
13017 (vectorizable_shift): Likewise.
13018 (vectorizable_operation): Likewise.
13019 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13021 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13023 * match.pd: Add pow (C, x) simplification.
13025 2017-08-21 Richard Biener <rguenther@suse.de>
13027 PR tree-optimization/81900
13028 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13029 for blocks with abnormal predecessors.
13030 (compute_antic): Do not set visited flag prematurely.
13032 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13035 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13037 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13039 * stor-layout.h (vector_type_mode): Move to...
13040 * tree.h (vector_type_mode): ...here.
13041 * stor-layout.c (vector_type_mode): Move to...
13042 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13044 2017-08-21 Richard Biener <rguenther@suse.de>
13046 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13047 register_external_die hooks.
13048 (debug_false_tree_charstarstar_uhwistar): Declare.
13049 (debug_nothing_tree_charstar_uhwi): Likewise.
13050 * debug.c (do_nothing_debug_hooks): Adjust.
13051 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13052 (debug_nothing_tree_charstar_uhwi): Likewise.
13053 * dbxout.c (dbx_debug_hooks): Adjust.
13054 (xcoff_debug_hooks): Likewise.
13055 * sdbout.c (sdb_debug_hooks): Likewise.
13056 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13057 * dwarf2out.c (macinfo_label_base): New global.
13058 (dwarf2out_register_external_die): New function for the
13059 register_external_die hook.
13060 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13061 (dwarf2_debug_hooks): Use them.
13062 (dwarf2_lineno_debug_hooks): Adjust.
13063 (struct die_struct): Add with_offset flag.
13064 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13065 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13066 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13067 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13068 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13069 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13070 defining section names for the early LTO debug variants.
13071 (reset_indirect_string): New helper.
13072 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13073 (print_dw_val): Add support for offsetted symbol references.
13074 (get_ultimate_context): Split out from is_cxx.
13075 (is_cxx): Use get_ultimate_context.
13076 (is_fortran): Add decl overload.
13077 (compute_comp_unit_symbol): Split out worker from
13078 compute_section_prefix.
13079 (compute_section_prefix): Call compute_comp_unit_symbol and
13080 set comdat_type_p here.
13081 (output_die): Skip DIE symbol output for the LTO added one.
13082 Handle DIE symbol references with offset.
13083 (output_comp_unit): Guard section name mangling properly.
13084 For LTO debug sections emit a symbol at the section beginning
13085 which we use to refer to its DIEs.
13086 (add_abstract_origin_attribute): For DIEs registered via
13087 dwarf2out_register_external_die directly refer to the early
13088 DIE rather than indirectly through the shadow one we created.
13089 Remove obsolete call to dwarf2out_abstract_function for
13090 non-function/block origins.
13091 (gen_array_type_die): When generating early LTO debug do
13092 not emit DW_AT_string_length.
13093 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13094 late when in LTO. As suggested place a gcc_unreachable for
13095 the DECL_ABSTRACT_P case.
13096 (gen_subprogram_die): Avoid another specification DIE
13097 for early built declarations/definitions for the late LTO case.
13098 (gen_variable_die): Add type references for late duplicated VLA dies
13100 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13101 we have the abstract instance already.
13102 (process_scope_var): Adjust decl DIE contexts in LTO which
13103 first puts them in limbo.
13104 (gen_decl_die): Do not generate type DIEs late apart from
13105 types for VLAs or for decls we do not yet have a DIE. Do not
13106 call dwarf2out_abstract_function late.
13107 (dwarf2out_early_global_decl): Make sure to create DIEs
13108 for abstract instances of a decl first.
13109 (dwarf2out_late_global_decl): Adjust comment.
13110 (output_macinfo_op): With multiple macro sections use
13111 macinfo_label_base to distinguish labels.
13112 (output_macinfo): Likewise. Update macinfo_label_base.
13113 Pass in the line info label.
13114 (note_variable_value_in_expr): When generating LTO resolve
13115 all variable values here by generating DIEs as needed.
13116 (init_sections_and_labels): Add early LTO debug flag parameter
13117 and generate different sections and names if set. Add generation
13118 counter for the labels so we can have multiple of them.
13119 (reset_dies): Helper to allow DIEs to be output multiple times.
13120 (dwarf2out_finish): When outputting DIEs to the fat part of an
13121 LTO object first reset DIEs.
13122 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13123 (modified_type_die): Check for decl_ultimate_origin being self
13125 (gen_type_die_with_usage): Likewise.
13126 (gen_typedef_die): Allow decl_ultimate_origin being self.
13127 (set_decl_abstract_flags): Remove.
13128 (set_block_abstract_flags): Likewise.
13129 (dwarf2out_abstract_function): Treat the early generated DIEs
13130 as the abstract copy and only add DW_AT_inline and
13131 DW_AT_artificial here and call set_decl_origin_self.
13132 If the DIE has an abstract origin don't do anything.
13133 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13134 if we have none yet (Go fails to build one, PR78628).
13135 (variably_modified_type_p): Prevent endless recursion for Ada
13136 cyclic pointer types.
13137 * lto-streamer-in.c: Include debug.h.
13138 (dref_queue): New global.
13139 (lto_read_tree_1): Stream in DIE references.
13140 (lto_input_tree): Register DIE references.
13141 (input_function): Stream DECL_DEBUG_ARGS.
13142 * lto-streamer-out.c: Include debug.h.
13143 (lto_write_tree_1): Output DIE references.
13144 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13145 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13146 (output_function): Stream DECL_DEBUG_ARGS.
13147 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13148 Stream DECL_ABSTRACT_ORIGIN.
13149 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13150 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13151 DECL_CONTEXT for file-scope decls.
13152 * lto-streamer.h (struct dref_entry): Declare.
13153 (dref_queue): Likewise.
13154 * cfgexpand.c (pass_expand::execute): Do not call the
13155 outlining_inline_function hook here.
13156 * lto-wrapper.c (debug_obj): New global.
13157 (tool_cleanup): Unlink it if required.
13158 (debug_objcopy): New function.
13159 (run_gcc): Handle early debug sections in the IL files by
13160 extracting them to separate files, partially linkin them and
13161 feeding the result back as result to the linker.
13162 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13163 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13164 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13165 sections into a separate segment.
13166 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13168 (darwin_asm_dwarf_section): Likewise.
13169 (darwin_asm_output_dwarf_offset): Likewise.
13170 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13172 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13173 Alan Hayward <alan.hayward@arm.com>
13174 David Sherwood <david.sherwood@arm.com>
13176 * read-md.h (md_reader::record_potential_iterator_use): Replace
13177 pointer argument with an rtx and an index.
13178 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13179 (apply_mode_iterator): Likewise.
13180 (apply_code_iterator): Likewise.
13181 (apply_int_iterator): Likewise.
13182 (apply_subst_iterator): Likewise.
13183 (record_iterator_use): Likewise.
13184 (record_attribute_use): Likewise.
13185 (md_reader::record_potential_iterator_use): Likewise. Update calls
13186 to record_iterator_use and apply_iterator.
13187 (iterator_use): Replace ptr with x and index.
13188 (attribute_use): Likewise.
13189 (apply_attribute_uses): Update calls to apply_iterator.
13190 (apply_iterators): Likewise. Update initialization of iterator_use.
13191 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13192 and record_potential_iterator_use.
13193 (rtx_reader::read_rtx_operand): Likewise.
13195 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13196 Alan Hayward <alan.hayward@arm.com>
13197 David Sherwood <david.sherwood@arm.com>
13199 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13202 2017-08-21 Richard Biener <rguenther@suse.de>
13204 PR middle-end/81884
13205 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13206 at struct end conservatively when comparing common bases.
13208 2017-08-21 Richard Biener <rguenther@suse.de>
13210 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13211 (mem_ref_in_stmt): Remove.
13212 (determine_max_movement): Use ref index to get at the reference.
13213 (invariantness_dom_walker::before_dom_children): Deal with
13214 lim data already initialized.
13215 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13217 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13219 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13220 (TARGET_ISA_ROUND): Ditto.
13221 (TARGET_ROUND): Ditto.
13222 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13223 * config/i386/i386.md: Ditto.
13224 * config/i386/sse.md: Ditto.
13225 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13226 with OPTION_MASK_ISA_SSE4_1.
13228 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13231 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13232 __builtin_ia32_lzcnt_u16.
13234 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13237 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13238 (rs6000_set_current_function): Rewrite function to use it.
13240 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13243 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13244 and TYPE_WARN_IF_NOT_ALIGN.
13245 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13246 (handle_warn_if_not_align): New.
13247 (place_union_field): Call handle_warn_if_not_align.
13248 (place_field): Call handle_warn_if_not_align.
13249 Copy TYPE_WARN_IF_NOT_ALIGN.
13250 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13251 (layout_type): Likewise.
13252 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13254 (tree_decl_common): Add warn_if_not_align.
13255 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13256 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13257 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13258 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13259 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13260 * doc/extend.texi: Document warn_if_not_aligned attribute.
13261 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13263 2017-08-17 Martin Liska <mliska@suse.cz>
13266 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13267 (get_data_dependence): Use it as pointer type.
13268 (distribute_loop): Likewise.
13270 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13272 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13273 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13274 (p8_vmrgow_<mode>_direct): New define_insn.
13275 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13276 handle endianness for vmrgew and vmrgow permute patterns.
13278 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13280 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13281 * config/rs6000/rs6000-cpus.def: Remove comment.
13282 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13283 (POWERPC_MASKS): Likewise.
13284 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13285 use of TARGET_VSX_TIMODE.
13286 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13287 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13288 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13289 (rs6000_option_override_internal): Remove dead code.
13290 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13291 (rs6000_legitimize_reload_address): Likewise.
13292 (rs6000_legitimate_address_p): Likewise.
13293 (rs6000_opt_masks): Delete "vsx-timode".
13294 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13295 from function comment.
13296 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13297 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13298 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13300 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13301 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13302 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13303 (VSX_TI): Likewise.
13305 (define_peephole2): Likewise.
13307 2017-08-17 Martin Sebor <msebor@redhat.com>
13310 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13311 past the end of an array.
13312 (test_pp_format): Add test cases.
13314 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13316 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13317 missing ECF_NOTHROW flags.
13319 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13322 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13323 operands residing in integer registers.
13324 (*vsx_le_perm_load_<mode>): Likewise.
13325 (*vsx_le_perm_store_<mode>): Likewise.
13326 (define_peephole2): Add peepholes to optimize the above.
13328 2017-08-17 Marek Polacek <polacek@redhat.com>
13330 PR middle-end/81814
13331 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13332 to mimic what shorten_compare did. Change the return type to bool.
13333 (fold_cond_expr_with_comparison): Update call to
13334 operand_equal_for_comparison_p.
13335 (fold_ternary_loc): Likewise.
13337 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13339 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13341 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13343 2017-08-17 Richard Biener <rguenther@suse.de>
13345 * tree-ssa-structalias.c (solve_graph): When propagating
13346 to successors update the graphs succ edges and avoid duplicate work.
13348 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13351 * config/i386/i386.c (ix86_option_override_internal): Save target
13352 specific options after ix86_stack_protector_guard_reg was changed.
13354 2017-08-17 Richard Biener <rguenther@suse.de>
13356 PR tree-optimization/81827
13357 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13358 (new_var_info): Initialize it conservatively.
13359 (get_call_vi): Mark register vars.
13360 (new_scalar_tmp_constraint_exp): Likewise.
13361 (handle_rhs_call): Likewise.
13362 (handle_const_call): Likewise.
13363 (create_function_info_for): Likewise.
13364 (solve_constraints): Sort varinfos to separate register from
13365 non-register vars to pack points-to solution bitmaps during
13368 2017-08-17 Marek Polacek <polacek@redhat.com>
13370 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13372 2017-08-17 Richard Biener <rguenther@suse.de>
13374 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13375 to true when overflow is undefined and we saturated the result.
13377 2017-08-17 Alan Modra <amodra@gmail.com>
13380 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13381 Don't use store multiple if only one reg needs saving.
13382 (interesting_frame_related_regno): New function.
13383 (rs6000_frame_related): Don't emit frame info for regs that
13385 (rs6000_emit_epilogue): Likewise.
13387 2017-08-16 Nathan Sidwell <nathan@acm.org>
13389 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13390 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13391 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13392 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13393 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13394 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13395 (verify_type): Adjust for TYPE_BINFO move.
13396 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13397 process TYPE_BINFO directly.
13398 (hash_tree): Likewise.
13399 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13401 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13404 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13406 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13408 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13411 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13412 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13413 Add ix86_binary_operator_ok to insn constraint.
13414 (*iordi_1_bts): Ditto.
13415 (*xordi_1_btc): Ditto.
13416 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13417 Update corresponding peephole2 pattern.
13421 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13423 PR tree-optimization/81832
13424 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13425 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13427 2017-08-16 Marek Polacek <polacek@redhat.com>
13430 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13431 regarding size_zero_node.
13433 2017-08-16 Martin Liska <mliska@suse.cz>
13436 * config.gcc: Respect previously set extra_objs in case
13439 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13441 PR tree-optimization/81835
13442 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13443 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13444 not depend on the phi.
13446 2017-08-16 Alan Modra <amodra@gmail.com>
13448 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13451 2017-08-16 Alan Modra <amodra@gmail.com>
13453 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13454 (save_reg_p): ..into this. Update all callers.
13455 (first_reg_to_save): Simplify.
13457 2017-08-16 Alan Modra <amodra@gmail.com>
13459 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13462 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13466 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13467 constant count if that count is less than 32.
13469 2017-08-15 Nathan Sidwell <nathan@acm.org>
13471 * gcc.c (execute): Emit friendlier message if inferior is killed
13472 by an external cause.
13474 2017-08-15 Richard Biener <rguenther@suse.de>
13476 PR tree-optimization/81790
13477 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13478 CONSTRUCTORs from simplifying and VN.
13480 2017-08-14 Martin Sebor <msebor@redhat.com>
13482 * builtin-attrs.def: Add comments.
13484 2017-08-14 Martin Sebor <msebor@redhat.com>
13487 * doc/extend.texi (attribute nonstring): Document new attribute.
13489 2017-08-14 Martin Sebor <msebor@redhat.com>
13492 * tree-diagnostic.c (default_tree_printer): Handle %G.
13493 * gimple-pretty-print.h (percent_G_format): Declare new function.
13494 * gimple-pretty-print.c (percent_G_format): Define.
13495 * tree-pretty-print.c (percent_K_format): Add argument.
13497 2017-08-14 Martin Sebor <msebor@redhat.com>
13499 PR translation/79998
13500 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13501 Remove a stray space.
13503 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13506 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13507 (*iordi_1_bts): Ditto.
13508 (*xordi_1_btc): Ditto.
13510 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13513 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13514 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13516 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13517 quoted strings, and make more translator-friendly.
13518 (darwin_rs6000_override_options): Likewise.
13519 (rs6000_option_override_internal): Likewise.
13520 (rs6000_return_in_memory): Fix overlong line.
13521 (init_cmulative_args): Use quoted strings, and make more
13522 translator-friendly.
13523 (rs6000_pass_by_reference): Fix overlong line.
13524 (def_builtin): Use quoted strings.
13525 (altivec_expand_predicate_builtin): Use quoted strings, and make
13526 more translator-friendly.
13527 (htm_expand_builtin): Use quoted strings.
13528 (cpu_expand_builtin): Use quoted strings, and make more
13529 translator-friendly.
13530 (altivec_expand_builtin): Likewise.
13531 (paired_expand_predicate_builtin): Likewise.
13532 (rs6000_invalid_builtin): Likewise.
13533 (builtin_function_type): Use quoted strings.
13534 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13535 more translator-friendly.
13536 (rs6000_trampoline_init): Likewise.
13537 (rs6000_handle_altivec_attribute): Likewise.
13538 (rs6000_inner_target_options): Use quoted strings.
13539 (rs6000_disable_incompatible_switches): Likewise.
13540 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13541 strings, and make more translator-friendly.
13542 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13544 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13546 PR tree-optimization/81799
13547 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13548 cond_expr to simple gimple operand.
13550 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13552 PR middle-end/46932
13553 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13555 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13559 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13560 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13561 TARGET_GASISR_PROLOGUES.
13562 * config/avr/avr.c (avr_option_override): Same.
13563 (avr_pass_pre_proep::execute): Same.
13565 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13568 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13569 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13571 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13573 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13574 define_insn_and_split. Split to a memory load from 0 in
13575 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13576 using PTR mode iterator.
13577 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13578 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13579 (*add_tp_<mode>): Redefine as define_insn_and_split.
13580 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13581 address space. Merge with *add_tp_x32 using PTR mode iterator.
13582 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13583 Split to an add with a memory load from 0 in
13584 DEFAULT_TLS_SEG_REG address space.
13586 2017-08-12 Andrew Pinski <apinski@cavium.com>
13588 * config/aarch64/aarch64-option-extensions.def (rdma):
13589 Fix feature string to what Linux prints out in /proc/cpuinfo.
13591 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13594 * dwarf2out.c (modified_type_die): For C typedef types that have
13595 an ultimate origin, process the ultimate origin instead of the
13597 (gen_typedef_die): Assert that input DECLs have no ultimate
13599 (gen_type_die_with_usage): For typedef variants that have an
13600 ultimate origin, just call gen_decl_die on the original DECL.
13601 (process_scope_var): Avoid creating DIEs for local typedefs and
13602 concrete static variables.
13604 2017-08-12 Alan Modra <amodra@gmail.com>
13608 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13609 match gnu-user.h startfile.
13610 (ENDFILE_LINUX_SPEC): Similarly.
13612 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13615 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13617 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13619 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13620 * config/aarch64/aarch64.md (mov<mode>): Change.
13621 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13622 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13623 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13625 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13627 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13628 for storage order barriers.
13630 2017-08-11 Martin Liska <mliska@suse.cz>
13633 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13634 variables of void type.
13636 2017-08-11 Martin Liska <mliska@suse.cz>
13638 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13639 TARGET_SUPPORTS_ALIASES.
13640 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13641 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13642 (optimize_weakref): Likewise.
13643 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13644 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13645 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13647 2017-08-11 Martin Liska <mliska@suse.cz>
13650 * config/i386/i386.c (make_resolver_func): Do complete
13651 refactoring of the function.
13653 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13656 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13657 * config/i386/i386.c (ix86_stack_protect_guard): Use
13658 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13659 * doc/invoke.texi (x86 Options): Document
13660 -mstack-protector-guard-symbol= option.
13662 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13664 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13665 * config/i386/i386.c (ix86_split_stack_guard): New function.
13666 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13667 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13668 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13669 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13670 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13671 (split_stack_space_check): Call ix86_split_stack_guard.
13673 2017-08-10 Martin Sebor <msebor@redhat.com>
13675 * print-tree.c (print_node): Print location using the established
13677 Replace spaces with colons.
13678 (debug_raw, debug): Ditto.
13680 2017-08-10 Martin Sebor <msebor@redhat.com>
13683 * pretty-print.c (pp_format): Correct the handling of %s precision.
13685 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13688 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13690 (ix86_finalize_stack_frame_flags): This. Also clear
13691 frame_pointer_needed if -fno-omit-frame-pointer is used without
13693 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13694 with ix86_finalize_stack_frame_flags.
13695 (ix86_expand_epilogue): Likewise.
13696 (ix86_expand_split_stack_prologue): Likewise.
13697 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13699 2017-08-10 Martin Liska <mliska@suse.cz>
13702 * c-attribs.c (handle_target_attribute):
13703 Report warning for an empty string argument of target attribute.
13705 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13708 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13710 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13711 or DECL_NONLOCAL labels.
13712 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13713 or DECL_NONLOCAL labels here.
13715 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13717 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13718 to indicate when early gimple folding has been disabled.
13719 (rs6000_gimple_fold_builtin): Add debug content.
13720 (rs6000_invalid_builtin): Fix whitespace.
13721 (rs6000_expand_builtin): Fix whitespace.
13722 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13724 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13727 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13728 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13730 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13732 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13733 (qdf24xx): Likewise.
13734 * config/aarch64/aarch64.md: Include falkor.md.
13735 * config/aarch64/falkor.md: New.
13737 2017-08-09 Marek Polacek <polacek@redhat.com>
13740 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13741 * diagnostic.c (emit_diagnostic): Add a comment.
13742 (emit_diagnostic_valist): New function.
13744 2017-08-09 Marek Polacek <polacek@redhat.com>
13747 * input.c (make_location): New overload.
13748 * input.h (make_location): Declare.
13750 2017-08-08 Alan Modra <amodra@gmail.com>
13751 H.J. Lu <hongjiu.lu@intel.com>
13754 * gcc.c (NO_PIE_SPEC): Delete.
13755 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13757 (LINK_PIE_SPEC): ..to here.
13758 (LINK_COMMAND_SPEC): Support -no-pie.
13759 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13760 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13762 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13763 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13764 (ENDFILE_CRTEND_SPEC): Similarly.
13766 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13769 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13770 (mstack-protector-guard-offset=): Ditto.
13771 * config/i386/i386.c (ix86_option_override): Handle
13772 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13774 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13775 ix86_stack_protect_guard_offset variables.
13776 (TARGET_STACK_PROTECT_GUARD): Always define.
13777 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13778 and -mstack-protector-guard-offset= options.
13780 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13782 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13783 boundary case for the last candidate.
13785 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13787 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13788 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13790 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13792 PR middle-end/19706
13793 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13794 * config/aarch64/aarch64-builtins.c
13795 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13796 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13797 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13799 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13800 Andrew Pinski <pinskia@gmail.com>
13802 PR middle-end/19706
13803 * internal-fn.def (XORSIGN): New.
13804 * optabs.def (xorsign_optab): New.
13805 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13806 (convert_expand_mult_copysign): New.
13807 (pass_optimize_widening_mul::execute): Call
13808 convert_expand_mult_copysign.
13810 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13812 PR tree-optimization/81354
13813 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13814 Insert on edges rather than explicitly creating landing pads.
13815 (analyze_candidates_and_replace): Commit edge inserts.
13817 2017-08-08 Richard Biener <rguenther@suse.de>
13819 PR middle-end/81719
13820 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13821 (expand_simple_operations): Also look through ADDR_EXPRs with
13822 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13824 2017-08-08 Richard Biener <rguenther@suse.de>
13826 PR tree-optimization/81723
13827 * tree-vect-slp.c (struct bst_traits): New hash traits.
13828 (bst_fail): New global.
13829 (vect_build_slp_tree_2): New worker, split out from ...
13830 (vect_build_slp_tree): ... this now wrapping it with using
13831 bst_fail set to cache SLP tree build fails. Properly handle
13833 (vect_analyze_slp_instance): Allocate and free bst_fail.
13835 2017-08-08 Martin Liska <mliska@suse.cz>
13838 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13839 LABEL_DECLs that can be from a different function.
13841 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13843 PR tree-optimization/81744
13844 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13845 loop's number of iterations.
13847 2017-08-08 Martin Liska <mliska@suse.cz>
13849 * asan.c: Include header files.
13850 * attribs.c (build_decl_attribute_variant): New function moved
13852 (build_type_attribute_qual_variant): Likewise.
13853 (cmp_attrib_identifiers): Likewise.
13854 (simple_cst_list_equal): Likewise.
13855 (omp_declare_simd_clauses_equal): Likewise.
13856 (attribute_value_equal): Likewise.
13857 (comp_type_attributes): Likewise.
13858 (build_type_attribute_variant): Likewise.
13859 (lookup_ident_attribute): Likewise.
13860 (remove_attribute): Likewise.
13861 (merge_attributes): Likewise.
13862 (merge_type_attributes): Likewise.
13863 (merge_decl_attributes): Likewise.
13864 (merge_dllimport_decl_attributes): Likewise.
13865 (handle_dll_attribute): Likewise.
13866 (attribute_list_equal): Likewise.
13867 (attribute_list_contained): Likewise.
13868 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13869 (lookup_attribute_by_prefix): Likewise.
13870 * bb-reorder.c: Include header files.
13871 * builtins.c: Likewise.
13872 * calls.c: Likewise.
13873 * cfgexpand.c: Likewise.
13874 * cgraph.c: Likewise.
13875 * cgraphunit.c: Likewise.
13876 * convert.c: Likewise.
13877 * dwarf2out.c: Likewise.
13878 * final.c: Likewise.
13879 * fold-const.c: Likewise.
13880 * function.c: Likewise.
13881 * gimple-expr.c: Likewise.
13882 * gimple-fold.c: Likewise.
13883 * gimple-pretty-print.c: Likewise.
13884 * gimple.c: Likewise.
13885 * gimplify.c: Likewise.
13886 * hsa-common.c: Likewise.
13887 * hsa-gen.c: Likewise.
13888 * internal-fn.c: Likewise.
13889 * ipa-chkp.c: Likewise.
13890 * ipa-cp.c: Likewise.
13891 * ipa-devirt.c: Likewise.
13892 * ipa-fnsummary.c: Likewise.
13893 * ipa-inline.c: Likewise.
13894 * ipa-visibility.c: Likewise.
13896 * lto-cgraph.c: Likewise.
13897 * omp-expand.c: Likewise.
13898 * omp-general.c: Likewise.
13899 * omp-low.c: Likewise.
13900 * omp-offload.c: Likewise.
13901 * omp-simd-clone.c: Likewise.
13902 * opts-global.c: Likewise.
13903 * passes.c: Likewise.
13904 * predict.c: Likewise.
13905 * sancov.c: Likewise.
13906 * sanopt.c: Likewise.
13907 * symtab.c: Likewise.
13908 * toplev.c: Likewise.
13909 * trans-mem.c: Likewise.
13910 * tree-chkp.c: Likewise.
13911 * tree-eh.c: Likewise.
13912 * tree-into-ssa.c: Likewise.
13913 * tree-object-size.c: Likewise.
13914 * tree-parloops.c: Likewise.
13915 * tree-profile.c: Likewise.
13916 * tree-ssa-ccp.c: Likewise.
13917 * tree-ssa-live.c: Likewise.
13918 * tree-ssa-loop.c: Likewise.
13919 * tree-ssa-sccvn.c: Likewise.
13920 * tree-ssa-structalias.c: Likewise.
13921 * tree-ssa.c: Likewise.
13922 * tree-streamer-in.c: Likewise.
13923 * tree-vectorizer.c: Likewise.
13924 * tree-vrp.c: Likewise.
13925 * tsan.c: Likewise.
13926 * ubsan.c: Likewise.
13927 * varasm.c: Likewise.
13928 * varpool.c: Likewise.
13929 * tree.c: Remove functions moved to attribs.[ch].
13930 * tree.h: Likewise.
13931 * config/aarch64/aarch64.c: Add attrs.h header file.
13932 * config/alpha/alpha.c: Likewise.
13933 * config/arc/arc.c: Likewise.
13934 * config/arm/arm.c: Likewise.
13935 * config/avr/avr.c: Likewise.
13936 * config/bfin/bfin.c: Likewise.
13937 * config/c6x/c6x.c: Likewise.
13938 * config/cr16/cr16.c: Likewise.
13939 * config/cris/cris.c: Likewise.
13940 * config/darwin.c: Likewise.
13941 * config/epiphany/epiphany.c: Likewise.
13942 * config/fr30/fr30.c: Likewise.
13943 * config/frv/frv.c: Likewise.
13944 * config/ft32/ft32.c: Likewise.
13945 * config/h8300/h8300.c: Likewise.
13946 * config/i386/winnt.c: Likewise.
13947 * config/ia64/ia64.c: Likewise.
13948 * config/iq2000/iq2000.c: Likewise.
13949 * config/lm32/lm32.c: Likewise.
13950 * config/m32c/m32c.c: Likewise.
13951 * config/m32r/m32r.c: Likewise.
13952 * config/m68k/m68k.c: Likewise.
13953 * config/mcore/mcore.c: Likewise.
13954 * config/microblaze/microblaze.c: Likewise.
13955 * config/mips/mips.c: Likewise.
13956 * config/mmix/mmix.c: Likewise.
13957 * config/mn10300/mn10300.c: Likewise.
13958 * config/moxie/moxie.c: Likewise.
13959 * config/msp430/msp430.c: Likewise.
13960 * config/nds32/nds32-isr.c: Likewise.
13961 * config/nds32/nds32.c: Likewise.
13962 * config/nios2/nios2.c: Likewise.
13963 * config/nvptx/nvptx.c: Likewise.
13964 * config/pa/pa.c: Likewise.
13965 * config/pdp11/pdp11.c: Likewise.
13966 * config/powerpcspe/powerpcspe.c: Likewise.
13967 * config/riscv/riscv.c: Likewise.
13968 * config/rl78/rl78.c: Likewise.
13969 * config/rx/rx.c: Likewise.
13970 * config/s390/s390.c: Likewise.
13971 * config/sh/sh.c: Likewise.
13972 * config/sol2.c: Likewise.
13973 * config/sparc/sparc.c: Likewise.
13974 * config/spu/spu.c: Likewise.
13975 * config/stormy16/stormy16.c: Likewise.
13976 * config/tilegx/tilegx.c: Likewise.
13977 * config/tilepro/tilepro.c: Likewise.
13978 * config/v850/v850.c: Likewise.
13979 * config/vax/vax.c: Likewise.
13980 * config/visium/visium.c: Likewise.
13981 * config/xtensa/xtensa.c: Likewise.
13983 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13986 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13987 constraints since the -mupper-regs-* switches have been
13989 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13990 into a vector from a double word element that was extracted from
13991 another vector, and eliminate extra XXPERMDI instructions.
13992 (vsx_concat_<mode>_2): Likewise.
13993 (vsx_concat_<mode>_3): Likewise.
13994 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13995 concat to allow optimizing inserts from previous extracts.
13997 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13999 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14000 memory reference to a SSP offset in TLS address space.
14001 (ix86_print_operand) <case '@'>: Remove.
14002 (ix86_print_operand_punct_valid_p): Remove '@' code.
14003 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14004 UNSPEC_SP_TLS_TEST.
14005 (stack_tls_protect_set_<mode>): Remove.
14006 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14007 (stack_tls_protect_test_<mode>): Remove.
14008 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14010 2017-08-07 Olivier Hainque <hainque@adacore.com>
14013 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14015 2017-08-07 Douglas Rupp <rupp@adacore.com>
14017 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14018 variable was referenced as multidir in command.
14020 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14023 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14026 2017-08-07 Martin Liska <mliska@suse.cz>
14028 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14029 * config/rl78/rl78.c: Add include of attribs.h.
14030 * config/sh/sh.c: Likewise.
14031 * config/v850/v850.c: Likewise.
14033 2017-08-07 Tom de Vries <tom@codesourcery.com>
14035 PR middle-end/78266
14036 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14038 2017-08-07 Martin Liska <mliska@suse.cz>
14040 * config/mips/mips.c: Include attribs.h.
14042 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14045 * doc/invoke.texi: Document change in behvaior for -Ofast for
14048 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14050 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14052 (aarch64_pop_regs): Likewise.
14053 (aarch64_gen_load_pair): Likewise.
14054 (aarch64_save_callee_saves): Likewise.
14055 (aarch64_restore_callee_saves): Likewise.
14057 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14059 * config/i386/i386.c: Revert the last change.
14061 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14064 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14066 (ix86_finalize_stack_frame_flags): This. Also clear
14067 frame_pointer_needed if -fno-omit-frame-pointer is used without
14069 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14070 with ix86_finalize_stack_frame_flags.
14071 (ix86_expand_epilogue): Likewise.
14072 (ix86_expand_split_stack_prologue): Likewise.
14074 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14077 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14078 to P_AES for Westmere.
14080 2017-08-07 Jonathan Yong <10walls@gmail.com>
14082 * config/i386/mingw.opt (fset-stack-executable): Removed.
14083 * config/i386/cygming.opt (fset-stack-executable): Moved
14085 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14087 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14089 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14091 2017-08-07 Marek Polacek <polacek@redhat.com>
14093 PR middle-end/81737
14094 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14096 2017-08-07 Martin Liska <mliska@suse.cz>
14098 * attribs.h (canonicalize_attr_name): New function.
14099 (cmp_attribs): Move from c-format.c and adjusted.
14100 (is_attribute_p): Moved from tree.h.
14101 * tree-inline.c: Add new includes.
14102 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14103 (private_is_attribute_p): Remove.
14104 (private_lookup_attribute): Likewise.
14105 (private_lookup_attribute_by_prefix): Simplify.
14106 (remove_attribute): Use is_attribute_p.
14107 * tree.h: Remove removed declarations.
14109 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14111 PR middle-end/81698
14112 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14113 instead of computing it in the function. Formatting fix.
14114 (expand_case): Don't rely on default_edge being the first edge,
14115 clear it if removing it, pass default_edge to
14116 emit_case_dispatch_table.
14117 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14120 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14122 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14123 insn in the function, emit NOP after the insn.
14125 2017-08-06 Tom de Vries <tom@codesourcery.com>
14127 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14130 2017-08-06 Tom de Vries <tom@codesourcery.com>
14132 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14135 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14137 PR tree-optimization/57371
14138 * match.pd: New pattern.
14140 2017-08-04 Marek Polacek <polacek@redhat.com>
14142 PR middle-end/81695
14143 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14144 perform the computation in offset_int.
14146 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14148 PR tree-optimization/81136
14149 * tree-vectorizer.h: Include tree-hash-traits.h.
14150 (vec_base_alignments): New typedef.
14151 (vec_info): Add a base_alignments field.
14152 (vect_record_base_alignments): Declare.
14153 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14155 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14156 (create_data_ref): Add an is_conditional_in_stmt argument.
14157 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14158 the is_conditional_in_stmt field.
14159 (data_ref_loc): Add an is_conditional_in_stmt field.
14160 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14161 (find_data_references_in_stmt): Update call to create_data_ref.
14162 (graphite_find_data_references_in_stmt): Likewise.
14163 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14164 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14165 (vect_record_base_alignment): New function.
14166 (vect_record_base_alignments): Likewise.
14167 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14168 for nested statements even if we fail to compute a misalignment.
14169 Use pooled base alignments for unconditional references.
14170 (vect_find_same_alignment_drs): Compare base addresses instead
14172 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14173 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14175 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14177 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14178 Add an explicit name for the enum. Use auto_vec for slp_instances
14179 and grouped_stores.
14180 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14182 (_bb_vec_info): Add a constructor and destructor.
14183 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14184 (destroy_loop_vec_info): Delete.
14185 (vect_destroy_datarefs): Likewise.
14186 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14187 (vec_info::vec_info): New function.
14188 (vec_info::~vec_info): Likewise.
14189 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14190 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14191 destroy_loop_vec_info.
14192 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14193 (_loop_vec_info::_loop_vec_info): ...this.
14194 (destroy_loop_vec_info): Replace with...
14195 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14196 the stmt_vec_infos. Leave handling of vec_info information to its
14197 destructor. Remove explicit vector releases.
14198 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14199 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14200 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14201 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14202 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14203 (destroy_bb_vec_info): Replace with...
14204 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14205 information to its destructor.
14206 (vect_slp_analyze_bb_1): Use new and delete instead of
14207 new_bb_vec_info and destroy_bb_vec_info.
14208 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14211 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14213 * tree-data-ref.h (subscript): Add access_fn field.
14214 (data_dependence_relation): Add could_be_independent_p.
14215 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14216 (same_access_functions): Move to tree-data-ref.c.
14217 * tree-data-ref.c (ref_contains_union_access_p): New function.
14218 (access_fn_component_p): Likewise.
14219 (access_fn_components_comparable_p): Likewise.
14220 (dr_analyze_indices): Add a reference to access_fn_component_p.
14221 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14223 (constant_access_functions): Likewise.
14224 (add_other_self_distances): Likewise.
14225 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14226 (initialize_data_dependence_relation): Use XCNEW and remove
14227 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14228 of access functions that have the same type. Allow the
14229 subsequence to end with different bases in some circumstances.
14230 Record the chosen access functions in SUB_ACCESS_FN.
14231 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14232 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14233 (subscript_dependence_tester_1): Likewise dra and drb.
14234 (build_classic_dist_vector): Update calls accordingly.
14235 (subscript_dependence_tester): Likewise.
14236 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14237 DDR_COULD_BE_INDEPENDENT_P.
14238 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14239 comp_alias_ddrs instead of may_alias_ddrs.
14240 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14242 (vect_analyze_data_ref_dependence): Use it if
14243 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14244 distance vectors if that fails.
14245 (dependence_distance_ge_vf): New function.
14246 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14247 LOOP_VINFO_MAY_ALIAS_DDRS.
14249 2017-08-04 Richard Biener <rguenther@suse.de>
14251 PR middle-end/81705
14252 * fold-const.c (fold_binary_loc): Properly restrict
14253 minus_var0 && minus_var1 case when associating undefined overflow
14256 2017-08-04 Tom de Vries <tom@codesourcery.com>
14258 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14260 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14262 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14263 Don't start diagnostic messages with a capital letter.
14264 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14266 (rs6000_invalid_builtin): Likewise.
14267 (rs6000_trampoline_init): Likewise.
14269 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14272 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14273 after setting changeable df flags.
14275 2017-08-03 Richard Biener <rguenther@suse.de>
14277 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14278 up if the use is in USE - X.
14280 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14282 * toplev.c (dumpfile.h): New include.
14283 (internal_error_reentered): New static function. Use it...
14284 (internal_error_function): ...here to handle reentered internal_error.
14286 2017-08-03 Richard Biener <rguenther@suse.de>
14288 PR middle-end/81148
14289 * fold-const.c (split_tree): Add minus_var and minus_con
14290 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14291 here but always use minus_*.
14292 (associate_trees): Assert we never associate with MINUS_EXPR
14293 and NULL first operand. Do not recurse for PLUS_EXPR operands
14294 when associating as MINUS_EXPR either.
14295 (fold_binary_loc): Track minus_var and minus_con.
14297 2017-08-03 Tom de Vries <tom@codesourcery.com>
14300 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14301 ACCEL_COMPILER, apply finish_options on
14302 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14304 2017-08-03 Tom de Vries <tom@codesourcery.com>
14307 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14308 function_entry_patch_area_size > 0.
14310 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14313 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14314 instead of 10??LU, perform unit multiplication in wide_int,
14315 don't change alloc_object_size_limit if the limit is larger
14318 PR tree-optimization/81655
14319 PR tree-optimization/81588
14320 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14321 the case when ranges[i].low and high are 1 for unsigned type with
14324 PR middle-end/81052
14325 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14326 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14328 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14330 * tree-vrp.h: Add include guard.
14332 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14335 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14336 (ud2): New insn pattern.
14337 * config/i386/i386.c (ix86_expand_epilogue):
14338 For naked functions, generate ud2 instead of trap insn.
14340 2017-08-02 Marek Polacek <polacek@redhat.com>
14343 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14345 2017-08-02 Tom de Vries <tom@codesourcery.com>
14346 Cesar Philippidis <cesar@codesourcery.com>
14348 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14349 Add missing edge probabilities.
14351 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14353 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14354 Correct endianness.
14356 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14358 PR middle-end/79499
14359 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14360 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14361 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14362 prologue_seq sequences - if any.
14364 2017-08-02 Richard Biener <rguenther@suse.de>
14366 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14367 via vectors if supported, integer extracts via punning if supported
14368 or otherwise vector extracts.
14370 2017-08-02 Richard Biener <rguenther@suse.de>
14372 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14374 (bitmap_insert_into_set): ... this.
14376 2017-08-02 Richard Biener <rguenther@suse.de>
14378 PR tree-optimization/81633
14380 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14382 PR tree-optimization/71752
14383 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14385 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14387 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14388 (machine_function::call_ms2sysv_pad_out): Remove field.
14389 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14390 (ix86_compute_frame_layout): Likewise.
14392 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14395 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14396 attribute with interrupt attribute.
14398 2017-08-01 Andrew Pinski <apinski@cavium.com>
14400 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14401 BIT_INSERT_EXPR's operand 1
14402 to see if the types precision matches.
14404 2017-08-01 Martin Liska <mliska@suse.cz>
14406 PR middle-end/70140
14407 * builtins.c (expand_builtin_memcpy_args): Remove.
14408 (expand_builtin_memcpy): Call newly added function
14409 expand_builtin_memory_copy_args.
14410 (expand_builtin_memcpy_with_bounds): Likewise.
14411 (expand_builtin_mempcpy): Remove last argument.
14412 (expand_builtin_mempcpy_with_bounds): Likewise.
14413 (expand_builtin_memory_copy_args): New function created from
14414 expand_builtin_mempcpy_args with small modifications.
14415 (expand_builtin_mempcpy_args): Remove.
14416 (expand_builtin_stpcpy): Remove unused argument.
14417 (expand_builtin): Likewise.
14418 (expand_builtin_with_bounds): Likewise.
14420 2017-08-01 Martin Liska <mliska@suse.cz>
14423 Make mempcpy more optimal (PR middle-end/70140).
14425 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14428 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14429 __builtin_vec_cmpne verify both arguments are compatible vectors
14430 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14431 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14432 move computation of aligned to after checking the argument types.
14436 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14437 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14440 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14441 Doug Rupp <rupp@adacore.com>
14442 Olivier Hainque <hainque@adacore.com>
14444 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14445 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14447 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14448 for TARGET_OS_CPP_BUILTIN.
14449 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14450 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14452 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14453 passing required abi options to the assembler for EABI configurations.
14454 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14455 of .text.hot and .text.unlikely sections for kernel modules when
14456 using ARM style exceptions.
14457 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14458 options. Add EXTRA_CC1_SPEC.
14459 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14461 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14463 (ARM_TARGET2_DWARF_FORMAT): Define.
14464 * config/arm/t-vxworks: Adjust multilib control to removal of the
14465 Diab command line options.
14467 2017-08-01 Martin Liska <mliska@suse.cz>
14469 PR gcov-profile/81561
14470 * gcov.c (unblock): Make unblocking safe as we need to preserve
14471 index correspondence of blocks and block_lists.
14473 2017-08-01 Richard Biener <rguenther@suse.de>
14475 PR tree-optimization/81181
14476 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14477 (compute_antic): ... end of iteration here.
14479 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14481 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14482 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14483 (ftree-slp-vectorize): Likewise.
14484 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14485 can no longer be set independent of flag_tree_loop_vectorize.
14486 * omp-general.c (emp_max_vf): Likewise.
14487 * opts.c (enable_fdo_optimizations): Remove references to
14488 flag_tree_vectorize, these are now implicit.
14489 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14490 and leave it for the options machinery.
14492 2017-08-01 Martin Liska <mliska@suse.cz>
14494 PR middle-end/70140
14495 * builtins.c (expand_builtin_memcpy_args): Remove.
14496 (expand_builtin_memcpy): Call newly added function
14497 expand_builtin_memory_copy_args.
14498 (expand_builtin_memcpy_with_bounds): Likewise.
14499 (expand_builtin_mempcpy): Remove last argument.
14500 (expand_builtin_mempcpy_with_bounds): Likewise.
14501 (expand_builtin_memory_copy_args): New function created from
14502 expand_builtin_mempcpy_args with small modifications.
14503 (expand_builtin_mempcpy_args): Remove.
14504 (expand_builtin_stpcpy): Remove unused argument.
14505 (expand_builtin): Likewise.
14506 (expand_builtin_with_bounds): Likewise.
14508 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14511 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14512 print "ds:" only for immediates in generic address space.
14514 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14517 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14518 (ix86_function_ok_for_sibcall): Return false for naked functions.
14520 2017-08-01 Richard Biener <rguenther@suse.de>
14522 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14523 (compute_antic): Seed worklist with exit block predecessors.
14524 * cfganal.c (dfs_find_deadend): For a cycle return the source
14525 of the edge closing it.
14527 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14529 * config/aarch64/aarch64.c
14530 (aarch64_can_const_movi_rtx_p): Move 0 check.
14532 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14534 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14535 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14537 * match.pd: Ditto in address comparison pattern.
14539 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14541 PR tree-optimization/81627
14542 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14543 closed ssa form for store-store chain.
14545 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14547 PR tree-optimization/81620
14548 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14549 for store-store chain.
14551 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14553 PR tree-optimization/81588
14554 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14555 ranges[i].in_p, invert comparison code ccode. For >/>=,
14556 swap rhs1 and rhs2 and comparison code unconditionally,
14557 for </<= don't do that. Don't swap rhs1/rhs2 again if
14558 ranges[i].in_p, instead invert comparison code ccode if
14559 opcode or oe->rank is BIT_IOR_EXPR.
14562 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14563 a direct optab to conversion optab.
14564 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14565 with GET_MODE_INNER as last argument instead of optab_handler.
14566 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14567 vector extraction if possible and optab is available.
14568 * expr.c (store_constructor): Use convert_optab_handler instead
14569 of optab_handler. Use vector initialization from smaller
14570 vectors if possible and optab is available.
14571 * tree-vect-stmts.c (vectorizable_load): Likewise.
14572 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14573 now have two modes.
14574 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14575 of vec_init from half-sized vectors with the same element mode.
14576 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14577 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14578 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14579 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14580 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14581 after mode in gen_vec_extract* calls.
14582 (vec_extract<mode>): Renamed to ...
14583 (vec_extract<mode><ssescalarmodelower>): ... this.
14584 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14585 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14586 element mode after mode in gen_vec_init* calls.
14587 (VEC_INIT_HALF_MODE): New mode iterator.
14588 (vec_init<mode>): Renamed to ...
14589 (vec_init<mode><ssescalarmodelower>): ... this.
14590 (vec_init<mode><ssehalfvecmodelower>): New expander.
14591 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14592 (vec_extractv2sfsf): ... this.
14593 (vec_initv2sf): Renamed to ...
14594 (vec_initv2sfsf): ... this.
14595 (vec_extractv2si): Renamed to ...
14596 (vec_extractv2sisi): ... this.
14597 (vec_initv2si): Renamed to ...
14598 (vec_initv2sisi): ... this.
14599 (vec_extractv4hi): Renamed to ...
14600 (vec_extractv4hihi): ... this.
14601 (vec_initv4hi): Renamed to ...
14602 (vec_initv4hihi): ... this.
14603 (vec_extractv8qi): Renamed to ...
14604 (vec_extractv8qiqi): ... this.
14605 (vec_initv8qi): Renamed to ...
14606 (vec_initv8qiqi): ... this.
14607 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14608 (vec_init<mode>): Renamed to ...
14609 (vec_init<mode><VEC_base_l>): ... this.
14610 (vec_extract<mode>): Renamed to ...
14611 (vec_extract<mode><VEC_base_l>): ... this.
14612 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14613 (vec_initv2sfsf): ... this.
14614 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14615 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14616 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14617 element mode after mode in gen_vec_init* calls.
14618 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14619 (vec_init<mode><Vel>): ... this.
14620 (vec_extract<mode>): Renamed to ...
14621 (vec_extract<mode><Vel>): ... this.
14622 * config/aarch64/iterators.md (Vel): New mode attribute.
14623 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14624 Add element mode after mode in gen_vec_extract* calls.
14625 * config/s390/vector.md (non_vec_l): New mode attribute.
14626 (vec_extract<mode>): Renamed to ...
14627 (vec_extract<mode><non_vec_l>): ... this.
14628 (vec_init<mode>): Renamed to ...
14629 (vec_init<mode><non_vec_l>): ... this.
14630 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14631 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14633 * config/arm/iterators.md (V_elem_l): New mode attribute.
14634 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14635 (vec_extract<mode><V_elem_l>): ... this.
14636 (vec_extractv2di): Renamed to ...
14637 (vec_extractv2didi): ... this.
14638 (vec_init<mode>): Renamed to ...
14639 (vec_init<mode><V_elem_l>): ... this.
14640 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14641 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14642 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14643 Add element mode after gen_vec_extract* calls.
14644 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14645 (vec_init<mode><unitmode>): ... this.
14646 (vec_extract<mode>): Renamed to ...
14647 (vec_extract<mode><unitmode>): ... this.
14648 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14649 (vec_init<mode><unitmode>): ... this.
14650 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14651 (vec_initv2sfsf): ... this.
14652 (vec_extractv2sf): Renamed to ...
14653 (vec_extractv2sfsf): ... this.
14654 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14655 Add element mode after gen_vec_extract* calls.
14656 * config/mips/mips.md (unitmode): New mode iterator.
14657 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14658 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14659 * config/spu/spu.md (inner_l): New mode attribute.
14660 (vec_init<mode>): Renamed to ...
14661 (vec_init<mode><inner_l>): ... this.
14662 (vec_extract<mode>): Renamed to ...
14663 (vec_extract<mode><inner_l>): ... this.
14664 * config/sparc/sparc.md (veltmode): New mode iterator.
14665 (vec_init<VMALL:mode>): Renamed to ...
14666 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14667 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14668 (vec_initv2sisi): ... this.
14669 (vec_initv2sf): Renamed to ...
14670 (vec_initv2sfsf): ... this.
14671 (vec_extractv2sf): Renamed to ...
14672 (vec_extractv2sfsf): ... this.
14673 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14674 (vec_init<mode>): Renamed to ...
14675 (vec_init<mode><VEC_base_l>): ... this.
14676 (vec_extract<mode>): Renamed to ...
14677 (vec_extract<mode><VEC_base_l>): ... this.
14678 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14679 (vec_initv2sfsf): ... this.
14680 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14681 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14682 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14683 gen_vec_init* calls.
14685 2017-08-01 Richard Biener <rguenther@suse.de>
14687 PR tree-optimization/81297
14688 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14689 TREE_OVERFLOW from INTEGER_CSTs.
14691 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14693 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14695 2017-07-31 Carl Love <cel@us.ibm.com>
14697 * config/rs6000/rs6000-c: Add support for built-in functions
14698 vector signed char vec_xl_be (signed long long, signed char *);
14699 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14700 vector signed int vec_xl_be (signed long long, signed int *);
14701 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14702 vector signed long long vec_xl_be (signed long long, signed long long *);
14703 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14704 vector signed short vec_xl_be (signed long long, signed short *);
14705 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14706 vector double vec_xl_be (signed long long, double *);
14707 vector float vec_xl_be (signed long long, float *);
14708 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14709 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14710 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14712 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14713 (altivec_expand_builtin): Add switch statement to call
14714 altivec_expand_xl_be for each builtin.
14715 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14716 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14717 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14718 __builtin_vsx_le_be_v16qi.
14719 * doc/extend.texi: Update the built-in documentation file for the
14720 new built-in functions.
14722 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14725 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14727 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14729 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14731 * config.gcc: Add z14.
14732 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14733 CPU model numbers for z13s and z14.
14734 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14736 * config/s390/s390-opts.h (enum processor_type): Rename
14737 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14738 * config/s390/s390.c (processor_table): Add field for CPU name to
14739 be passed to Binutils.
14740 (s390_asm_output_machine_for_arch): Use the new field in
14741 processor_table for Binutils.
14742 (s390_expand_builtin): Replace arch12 with z14.
14743 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14744 (s390_get_sched_attrmask): Likewise.
14745 (s390_get_unit_mask): Likewise.
14746 * config/s390/s390.opt: Add z14 to processor_type enum.
14748 2017-07-31 Martin Jambor <mjambor@suse.cz>
14751 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14752 regardless of optimization level.
14754 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14755 Martin Liska <mliska@suse.cz>
14757 * predict.def: Remove old comment and adjust probability.
14758 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14759 PREDICT statements.
14761 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14764 * config/i386/i386.c (ix86_function_naked): New function.
14765 (ix86_can_use_return_insn_p): Return false for naked functions.
14766 (ix86_expand_prologue): Skip prologue for naked functions.
14767 (ix86_expand_epilogue): Skip epilogue for naked functions
14768 and emit trap instruction.
14769 (ix86_warn_func_return): New function.
14770 (ix86_attribute_table): Add "naked" attribute specification.
14771 (TARGET_WARN_FUNC_RETURN): Define.
14772 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14774 2017-07-31 Martin Liska <mliska@suse.cz>
14776 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14777 (dump_gimple_bb_header): Always dump BB info.
14778 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14780 2017-07-31 Martin Liska <mliska@suse.cz>
14783 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14784 also with current_function_decl non-null equality.
14786 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14789 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14790 change type to the element type, instead add eltype variable and
14791 use it where we are interested in the element type.
14793 PR tree-optimization/81603
14794 * ipa-polymorphic-call.c
14795 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14796 offset arithmetic in offset_int, bail out if the resulting bit offset
14797 doesn't fit into shwi.
14799 2017-07-31 Martin Liska <mliska@suse.cz>
14801 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14802 (gimplify_save_expr): Fix comment.
14804 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14807 * config/i386/i386.c (ix86_function_arg): Update arguments for
14809 (ix86_compute_frame_layout): Set the initial stack offset to
14810 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14811 INCOMING_FRAME_SP_OFFSET.
14812 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14813 stack before exception handler returns.
14814 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14815 the 'ERROR_CODE' for exception handler.
14817 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14819 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14820 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14821 (ASM_OUTPUT_REG_POP): Ditto.
14822 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14823 instead of asm_fprintf to output pure string.
14825 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14827 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14828 to imported_module_or_decl hook.
14829 (debug_nothing_tree_tree_tree_bool): Remove.
14830 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14831 * debug.c (do_nothing_debug_hooks): Use
14832 debug_nothing_tree_tree_tree_bool_bool instead of
14833 debug_nothing_tree_tree_tree_bool.
14834 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14835 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14836 * sdbout.c (sdb_debug_hooks): Likewise.
14837 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14838 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14840 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14841 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14842 attribute, don't add anything.
14844 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14846 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14847 (fold_build2_stat_loc): Likewise.
14848 (fold_build3_stat_loc): Likewise.
14849 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14850 (fold_build1_loc): Remove macro.
14851 (fold_build2_loc): Likewise.
14852 (fold_build3_loc): Likewise.
14854 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14856 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14857 (gimple_build_debug_bind_source_stat): Likewise.
14858 * gimple.h (gimple_build_debug_bind): Remove macro.
14859 (gimple_build_debug_bind_source): Likewise.
14861 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14863 * bitmap.c (bitmap_alloc): Adjust.
14864 (bitmap_gc_alloc): Likewise.
14865 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14867 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14869 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14870 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14871 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14872 (bitmap_gc_alloc_stat): Likewise.
14873 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14875 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14877 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14878 * rtl.h (shallow_copy_rtx): Remove macro.
14880 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14882 * emit-rtl.c (gen_raw_REG): Adjust.
14883 * gengenrtl.c (gendef): Likewise.
14884 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14885 * rtl.h (rtx_alloc): Remove macro.
14887 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14889 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14890 (build_tree_list_stat): Likewise.
14891 * tree.h (build_tree_list): Remove macro.
14892 (build_tree_list_vec): Likewise.
14894 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14896 * tree.c (make_vector_stat): Remove _stat from name.
14897 (build_vector_stat): Likewise.
14898 * tree.h (make_vector_stat): Remove macro.
14899 (build_vector_stat): Likewise.
14901 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14903 * tree.h (build_var_debug_value): Remove prototype.
14905 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14907 * tree.c (tree_cons_stat): Remove _stat from name.
14908 * tree.h (tree_cons): Remove macro.
14910 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14912 * tree.c (build_vl_exp_stat): Remove _stat from name.
14913 * tree.h (build_vl_exp): Remove macro.
14915 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14917 * tree.c (build_decl_stat): Remove _stat from name.
14918 * tree.h (build_decl): Remove macro.
14920 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14922 * gimple.c (gimple_build_with_ops_stat): Adjust.
14923 (gimple_alloc_stat): Remove _stat from name.
14924 * gimple.h (gimple_alloc): Remove macro.
14926 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14928 * tree.c (make_tree_vec_stat): Remove _stat from name.
14929 (grow_tree_vec_stat): Likewise.
14930 * tree.h (make_tree_vec_stat): Adjust prototype.
14931 (grow_tree_vec_stat): Likewise.
14932 (make_tree_vec): Remove macro.
14933 (grow_tree_vec): Likewise.
14935 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14937 * fold-const.c (fold_build1_stat_loc): Adjust.
14938 (fold_build2_stat_loc): Likewise.
14939 (fold_build3_stat_loc): Likewise.
14940 * tree.c (build0_stat): Remove _stat from name.
14941 (build1_stat): Likewise.
14942 (build2_stat): Likewise.
14943 (build3_stat): Likewise.
14944 (build4_stat): Likewise.
14945 (build5_stat): Likewise.
14946 * tree.h (build1_loc): Remove macro, and rename _stat function
14948 (build2_loc): Likewise.
14949 (build3_loc): Likewise.
14950 (build4_loc): Likewise.
14951 (build5_loc): Likewise.
14953 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14955 * tree.c (make_int_cst_stat): Remove _stat from name.
14956 * tree.h (make_int_cst_stat): Adjust prototype.
14957 (make_int_cst): Remove macro.
14959 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14961 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14962 * tree.h (make_tree_binfo_stat): Adjust prototype.
14963 (make_tree_binfo): Remove.
14965 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14967 * tree.c (copy_node_stat): Rename to copy_node.
14968 (build_distinct_type_copy): Adjust.
14969 * tree.h (copy_node_stat): Adjust prototype.
14970 (copy_node): Remove macro.
14972 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14974 * tree.c (make_node_stat): rename to make_node.
14975 (build_tree_list_stat): Adjust.
14976 (build0_stat): Likewise.
14977 (build2_stat): Likewise.
14978 (build3_stat): Likewise.
14979 (build4_stat): Likewise.
14980 (build5_stat): Likewise.
14981 (build_decl_stat): Likewise.
14982 * tree.h (make_node_stat): Adjust prototype.
14983 (make_node): remove macro.
14985 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14987 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14988 (PPC_FEATURE2_SCV): Likewise.
14989 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14991 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14993 * config/aarch64/aarch64.c
14994 (aarch64_internal_mov_immediate): Add new special pattern.
14995 * config/aarch64/aarch64.md (*movdi_aarch64):
14996 Add reg/32bit const mov case.
14998 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14999 Richard Sandiford <richard.sandiford@linaro.org>
15001 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15002 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15003 Add integer and movi cases.
15004 (movi-split-hf-df-sf split, fp16): New.
15005 (enabled): Added TARGET_FP_F16INST.
15006 * config/aarch64/iterators.md (GPF_HF): New.
15007 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15009 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15011 * config/aarch64/aarch64.c
15012 (aarch64_simd_container_mode): Add prototype.
15013 (aarch64_expand_mov_immediate): Add HI support.
15014 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15015 (aarch64_can_const_movi_rtx_p): New.
15016 (aarch64_preferred_reload_class):
15017 Remove restrictions of using FP registers for certain SIMD operations.
15018 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15019 (aarch64_valid_floating_const): Add integer move validation.
15020 (aarch64_simd_imm_scalar_p): Remove.
15021 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15022 (aarch64_legitimate_constant_p): Expand list of supported cases.
15023 * config/aarch64/aarch64-protos.h
15024 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15025 (aarch64_reinterpret_float_as_int): New.
15026 (aarch64_simd_imm_scalar_p): Remove.
15027 * config/aarch64/constraints.md (Uvi): New.
15028 (Dd): Split into Ds and new Dd.
15029 * config/aarch64/aarch64.md (*movsi_aarch64):
15031 (*movdi_aarch64): Add SIMD mov case.
15033 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15035 * tree-predcom.c: (struct chain): Handle store-store chain in which
15036 stores for elimination only store loop invariant values.
15037 (execute_pred_commoning_chain): Ditto.
15038 (prepare_initializers_chain_store_elim): Ditto.
15039 (prepare_finalizers): Ditto.
15040 (is_inv_store_elimination_chain): New function.
15041 (initialize_root_vars_store_elim_1): New function.
15043 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15045 * tree-predcom.c: Revise general description of the pass.
15046 (enum chain_type): New enum type for store elimination.
15047 (struct chain): New field supporting store elimination.
15048 (struct component): Ditto.
15049 (dump_chain): Dump store-stores chain.
15050 (release_chain): Release resources.
15051 (split_data_refs_to_components): Compute and create component
15052 contains only stores for elimination.
15053 (get_chain_last_ref_at): New function.
15054 (make_invariant_chain): Initialization.
15055 (make_rooted_chain): Specify chain type in parameter and record it.
15056 (add_looparound_copies): Skip for store-stores chain.
15057 (determine_roots_comp): Compute type of chain and pass it to
15059 (initialize_root_vars_store_elim_2): New function.
15060 (finalize_eliminated_stores): New function.
15061 (remove_stmt): Handle store for elimination.
15062 (execute_pred_commoning_chain): Execute predictive commoning on
15063 store-store chains.
15064 (determine_unroll_factor): Skip unroll for store-stores chain.
15065 (prepare_initializers_chain_store_elim): New function.
15066 (prepare_initializers_chain): Hanlde store-store chain.
15067 (prepare_finalizers_chain, prepare_finalizers): New function.
15068 (tree_predictive_commoning_loop): Return integer value indicating
15069 if loop is unrolled or lcssa form is corrupted.
15070 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15072 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15074 * tree-predcom.c (initialize_root): Delete.
15075 (execute_pred_commoning_chain): Initialize root vars and replace
15076 reference of non-combined chain directly, rather than call above
15079 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15081 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15082 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15084 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15086 * tree-predcom.c (struct chain): New field init_seq.
15087 (release_chain): Release init_seq.
15088 (prepare_initializers_chain): Record intialization stmts in above
15090 (insert_init_seqs): New function.
15091 (tree_predictive_commoning_loop): Call insert_init_seqs.
15093 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15095 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15097 2017-07-28 Richard Biener <rguenther@suse.de>
15099 * match.pd: Remove superfluous :c.
15100 * genmatch.c (simplify::id): Add member.
15101 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15103 (current_id): New global.
15104 (dt_node::parent): Move from ...
15105 (dt_operand::parent): ... here. Add for_id member.
15106 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15107 (decision_tree::find_node): Relax order requirement when
15108 merging DT_TRUE nodes to ones inbetween the current simplify
15109 and the one we try to merge with. Add diagnostic whenever
15110 we need to enforce pattern order by not merging.
15111 (decision_tree::insert): Set current_id.
15112 (decision_tree::print_node): Dump parent node and for_id.
15113 (parser::last_id): Add member.
15114 (parser::push_simplify): Assign unique id.
15115 (parser::parser): Initialize last_id.
15117 2017-07-28 Martin Liska <mliska@suse.cz>
15120 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15121 gimple_build_debug_bind.
15123 2017-07-28 Richard Biener <rguenther@suse.de>
15125 PR tree-optimization/81502
15126 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15128 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15130 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15131 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15132 for BIT_FIELD_REF args.
15133 * fold-const.c (make_bit_field_ref): Likewise.
15134 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15136 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15139 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15140 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15141 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15142 Or it into SANITIZER_UNDEFINED.
15143 * ubsan.c: Include gimple-fold.h and varasm.h.
15144 (ubsan_expand_ptr_ifn): New function.
15145 (instrument_pointer_overflow): New function.
15146 (maybe_instrument_pointer_overflow): New function.
15147 (instrument_object_size): Formatting fix.
15148 (pass_ubsan::execute): Call instrument_pointer_overflow
15149 and maybe_instrument_pointer_overflow.
15150 * internal-fn.c (expand_UBSAN_PTR): New function.
15151 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15152 * sanitizer.def (__ubsan_handle_pointer_overflow,
15153 __ubsan_handle_pointer_overflow_abort): New builtins.
15154 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15155 * internal-fn.def (UBSAN_PTR): New internal function.
15156 * opts.c (sanitizer_opts): Add pointer-overflow.
15157 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15158 * fold-const.c (build_range_check): Compute pointer range check in
15159 integral type if pointer arithmetics would be needed. Formatting
15162 2017-07-28 Martin Liska <mliska@suse.cz>
15165 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15166 parameters that are of a variable-length.
15168 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15170 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15172 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15173 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15174 (CRT_CALL_STATIC_FUNCTION): Likewise.
15175 (ASM_DEFAULT_SPEC): New define.
15176 (ASM_SPEC32): Likewise.
15177 (ASM_SPEC64): Likewise.
15178 (ASM_SPEC_COMMON): Likewise.
15179 (ASM_SPEC): Likewise.
15180 (INVALID_64BIT): Likewise.
15181 (LINK_OS_DEFAULT_SPEC): Likewise.
15182 (LINK_OS_SPEC32): Likewise.
15183 (LINK_OS_SPEC64): Likewise.
15184 (POWERPC_LINUX): Likewise.
15185 (PTRDIFF_TYPE): Likewise.
15186 (RESTORE_FP_PREFIX): Likewise.
15187 (RESTORE_FP_SUFFIX): Likewise.
15188 (SAVE_FP_PREFIX): Likewise.
15189 (SAVE_FP_SUFFIX): Likewise.
15190 (SIZE_TYPE): Likewise.
15191 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15192 (TARGET_64BIT): Likewise.
15193 (TARGET_64BIT): Likewise.
15194 (TARGET_AIX): Likewise.
15195 (WCHAR_TYPE_SIZE): Likewise.
15196 (WCHAR_TYPE): Undefine.
15197 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15198 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15199 (CPP_OS_RTEMS_SPEC): Delete.
15200 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15201 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15203 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15205 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15207 PR tree-optimization/81578
15208 * tree-parloops.c (build_new_reduction): Bail out if
15209 reduction_code isn't one of the standard OpenMP reductions.
15210 Move the details printing after that decision.
15212 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15214 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15215 related to reload_in_progress.
15216 (splat_input_operand): Likewise.
15217 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15219 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15221 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15222 (TARGET_INSTANTIATE_DECLS): Likewise.
15223 (legitimate_indexed_address_p): Delete reload_in_progress code.
15224 (rs6000_debug_legitimate_address_p): Likewise.
15225 (rs6000_eliminate_indexed_memrefs): Likewise.
15226 (rs6000_emit_le_vsx_store): Likewise.
15227 (rs6000_emit_move_si_sf_subreg): Likewise.
15228 (rs6000_emit_move): Likewise.
15229 (register_to_reg_type): Likewise.
15230 (rs6000_pre_atomic_barrier): Likewise.
15231 (rs6000_machopic_legitimize_pic_address): Likewise.
15232 (rs6000_allocate_stack_temp): Likewise.
15233 (rs6000_address_for_fpconvert): Likewise.
15234 (rs6000_address_for_altivec): Likewise.
15235 (rs6000_secondary_memory_needed_rtx): Delete function.
15236 (rs6000_check_sdmode): Likewise.
15237 (rs6000_alloc_sdmode_stack_slot): Likewise.
15238 (rs6000_instantiate_decls): Likewise.
15239 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15240 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15241 Delete reload_in_progress.
15242 (*vec_reload_and_plus_<mptrsize>): Likewise.
15243 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15244 (vsx_div_v2di): Likewise.
15245 (vsx_udiv_v2di): Likewise.
15247 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15249 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15250 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15251 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15252 (rs6000_debug_reg_global): Delete print of LRA status.
15253 (rs6000_option_override_internal): Delete dead LRA related code.
15254 (rs6000_lra_p): Delete function.
15255 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15257 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15259 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15260 * config/riscv/rtems.h: New file.
15262 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15263 Sudakshina Das <sudi.das@arm.com>
15265 * config/aarch64/aarch64.md
15266 (define_split for and<mode>3nr_compare): Move
15267 non aarch64_logical_operand to a register.
15268 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15269 register immediate operand to a register.
15270 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15272 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15274 PR middle-end/81564
15275 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15277 2017-07-27 Richard Biener <rguenther@suse.de>
15279 PR tree-optimization/81573
15280 PR tree-optimization/81494
15281 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15282 multi defuse cycle case.
15284 2017-07-27 Richard Biener <rguenther@suse.de>
15286 PR tree-optimization/81571
15287 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15290 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15292 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15293 earlier and only if MASK_FPU is set. Adjust formatting.
15295 2017-07-27 Martin Liska <mliska@suse.cz>
15297 * opt-functions.awk: Add validation of value of Init.
15298 * optc-gen.awk: Pass new argument.
15300 2017-07-27 Martin Liska <mliska@suse.cz>
15302 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15303 Fix wrong condition.
15305 2017-07-27 Martin Liska <mliska@suse.cz>
15307 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15308 BBs and edges seen by autoFDO.
15310 2017-07-27 Richard Biener <rguenther@suse.de>
15312 PR tree-optimization/81502
15313 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15314 with incompatible but same sized type.
15315 (execute_update_addresses_taken): Likewise.
15317 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15319 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15320 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15322 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15325 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15326 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15327 Change s_operand to memory_operand.
15329 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15331 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15332 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15333 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15334 Emit instructions rather than returning an expression. Handle TFmode
15335 and KFmode by casting to TImode.
15336 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15337 (rs6000_emit_le_vsx_store): Likewise.
15338 * config/rs6000/vsx.md (VSX_TI): New iterator.
15339 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15340 (*vsx_le_undo_permute_<mode>): Likewise.
15341 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15342 emit the split sequence.
15343 (*vsx_le_perm_store_<mode>): Likewise.
15345 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15347 PR tree-optimization/81555
15348 PR tree-optimization/81556
15349 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15350 if true, force CHANGED for the recursive invocation.
15351 (reassociate_bb): Remember original length of ops array, pass
15352 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15354 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15355 attributes for noipa attribute. For naked attribute use
15356 lookup_attribute first before lookup_attribute_spec.
15357 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15359 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15360 for functions with noipa attribute.
15361 (cgraph_externally_visible_p): Return true for functions with noipa
15363 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15364 for functions with noipa attribute.
15365 * doc/extend.texi: Document noipa function attribute.
15366 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15367 also for functions with noipa attribute.
15368 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15370 2017-07-26 Andrew Pinski <apinski@cavium.com>
15372 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15373 vec_unalign_load_cost and vec_unalign_store_cost.
15375 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15377 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15378 -mvsx-small-integer option.
15379 (ISA_3_0_MASKS_IEEE): Likewise.
15380 (OTHER_VSX_VECTOR_MASKS): Likewise.
15381 (POWERPC_MASKS): Likewise.
15382 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15383 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15384 code, only testing for DImode being allowed in non-VSX floating
15386 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15387 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15389 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15390 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15391 TARGET_P8_VECTOR test.
15392 (rs6000_secondary_reload_simple_move): Likewise.
15393 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15394 since TARGET_P9_VECTOR was already tested.
15395 (rs6000_opt_masks): Remove -mvsx-small-integer.
15396 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15397 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15399 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15400 test for TARGET_VEXTRACTUB was used, and that uses
15402 (p9 extract splitter): Likewise.
15403 (vsx_extract_<mode>_di_p9): Likewise.
15404 (vsx_extract_<mode>_store_p9): Likewise.
15405 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15406 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15407 the elimination of TARGET_VSX_SMALL_INTEGER.
15408 (vsx_extract_<mode>_p8): Likewise.
15409 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15410 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15411 (vsx_set_<mode>_p9): Likewise.
15412 (vsx_set_v4sf_p9): Likewise.
15413 (vsx_set_v4sf_p9_zero): Likewise.
15414 (vsx_insert_extract_v4sf_p9): Likewise.
15415 (vsx_insert_extract_v4sf_p9_2): Likewise.
15416 * config/rs6000/rs6000.md (sign extend splitter): Change
15417 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15418 (floatsi<mode>2_lfiwax_mem): Likewise.
15419 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15420 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15421 since a test for TARGET_P9_VECTOR was used.
15422 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15423 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15424 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15425 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15426 TARGET_P8_VECTOR test.
15427 (fix_trunc<mode>si2_stfiwx): Likewise.
15428 (fix_trunc<mode>si2_internal): Likewise.
15429 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15430 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15432 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15433 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15434 TARGET_P8_VECTOR test.
15435 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15436 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15437 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15439 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15440 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15441 since a test for TARGET_P9_VECTOR was used.
15442 (splitter for loading small constants): Likewise.
15444 2017-07-26 Andrew Pinski <apinski@cavium.com>
15446 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15449 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15452 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15453 (fp_valid_at): Likewise.
15455 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15457 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15458 (qdf24xx_addrcost_table): Likewise.
15459 (cortexa57_tunings): Update to use generic_branch_cost.
15460 (cortexa72_tunings): Likewise.
15461 (cortexa73_tunings): Likewise.
15462 (qdf24xx_tunings): Likewise.
15464 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15466 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15467 (thunderx2t99_branch_cost): Likewise.
15468 (cortexa35_tunings): Update to use generic_branch_cost.
15469 (cortexa53_tunings): Likewise.
15470 (cortexa57_tunings): Likewise.
15471 (cortexa72_tunings): Likewise.
15472 (cortexa73_tunings): Likewise.
15473 (thunderx2t99_tunings): Likewise.
15475 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15477 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15478 (sparc_option_override): Honour MASK_FSMULD.
15479 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15480 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15481 * config/sparc/sparc.opt (mfsmuld): New option.
15482 * doc/invoke.texi (mfsmuld): Document option.
15484 2017-07-26 Marek Polacek <polacek@redhat.com>
15486 PR middle-end/70992
15487 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15489 2017-07-26 Richard Biener <rguenther@suse.de>
15491 * gimple-match-head.c (do_valueize): Return OP if valueize
15493 (get_def): New helper to get at the def stmt of a SSA name
15494 if valueize allows.
15495 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15496 do_valueize to get at the def stmt.
15497 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15499 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15501 PR middle-end/46932
15502 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15504 2017-07-26 Martin Liska <mliska@suse.cz>
15507 * function.c (expand_function_start): Make expansion of
15508 nonlocal_goto_save_area after parm_birth_insn.
15510 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15512 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15513 from all CPU target flags enable members.
15515 2017-07-26 Richard Biener <rguenther@suse.de>
15517 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15518 (decision_tree::gen): Make 'type' const.
15519 (write_predicate): Likewise.
15521 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15523 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15524 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15525 (rs6000_option_override_internal): Likewise.
15526 (rs6000_expand_vector_set): Likewise.
15527 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15528 (TARGET_UPPER_REGS_SF): Likewise.
15529 (TARGET_UPPER_REGS_DI): Likewise.
15530 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15531 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15532 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15533 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15534 (Splitters for DI constants in Altivec registers): Likewise.
15535 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15536 (vsx_set_v4sf_p9): Likewise.
15537 (vsx_set_v4sf_p9_zero): Likewise.
15538 (vsx_insert_extract_v4sf_p9): Likewise.
15539 (vsx_insert_extract_v4sf_p9_2): Likewise.
15541 2017-07-25 Carl Love <cel@us.ibm.com>
15543 * doc/extend.texi: Update the built-in documentation file for the
15544 existing built-in functions
15545 vector signed char vec_cnttz (vector signed char);
15546 vector unsigned char vec_cnttz (vector unsigned char);
15547 vector signed short vec_cnttz (vector signed short);
15548 vector unsigned short vec_cnttz (vector unsigned short);
15549 vector signed int vec_cnttz (vector signed int);
15550 vector unsigned int vec_cnttz (vector unsigned int);
15551 vector signed long long vec_cnttz (vector signed long long);
15552 vector unsigned long long vec_cnttz (vector unsigned long long);
15554 2017-07-25 Andrew Pinski <apinski@cavium.com>
15556 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15557 accesses where the use is for the first operand of a BIT_INSERT.
15559 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15562 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15563 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15565 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15567 * config/i386/gstabs.h: Delete.
15568 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15570 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15572 * config/i386/i386.c (ix86_decompose_address): Do not check for
15573 register RTX when looking at index_reg or base_reg.
15574 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15576 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15578 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15579 to update EH info here.
15581 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15583 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15585 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15587 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15589 2017-07-25 Torsten Duwe <duwe@suse.de>
15591 * common.opt: Introduce -fpatchable-function-entry
15592 command line option, and its variables function_entry_patch_area_size
15593 and function_entry_patch_area_start.
15594 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15595 including a two-value parser.
15596 * target.def (print_patchable_function_entry): New target hook.
15597 * targhooks.h (default_print_patchable_function_entry): New function.
15598 * targhooks.c (default_print_patchable_function_entry): Likewise.
15599 * toplev.c (process_options): Switch off IPA-RA if
15600 patchable function entries are being generated.
15601 * varasm.c (assemble_start_function): Look at the
15602 patchable-function-entry command line switch and current
15603 function attributes and maybe generate NOP instructions by
15604 calling the print_patchable_function_entry hook.
15605 * doc/extend.texi: Document patchable_function_entry attribute.
15606 * doc/invoke.texi: Document -fpatchable_function_entry
15607 command line option.
15608 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15610 * doc/tm.texi: Re-generate.
15612 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15615 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15616 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15618 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15620 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15622 2017-07-25 Richard Biener <rguenther@suse.de>
15624 PR tree-optimization/81455
15625 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15626 not walk in cycles when looking for guards.
15628 2017-07-25 Richard Biener <rguenther@suse.de>
15630 PR tree-optimization/81529
15631 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15632 when optimizing backedge uses.
15634 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15636 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15638 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15639 to dl_section_ref. On AIX, append an expression to subtract
15640 the size of the section length to dl_section_ref.
15642 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15644 * configure.ac: If any of the config.* scripts fail, exit 1.
15645 * configure: Regenerate.
15647 2017-07-25 Richard Biener <rguenther@suse.de>
15649 PR middle-end/81546
15650 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15651 of immediate uses, be more verbose on errors.
15653 2017-07-25 Richard Biener <rguenther@suse.de>
15655 PR tree-optimization/81510
15656 * tree-vect-loop.c (vect_is_simple_reduction): When the
15657 reduction stmt is not inside the loop bail out.
15659 2017-07-25 Richard Biener <rguenther@suse.de>
15661 PR tree-optimization/81303
15662 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15663 profitability check against LOOP_VINFO_NITERSM1.
15665 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15667 * domwalk.c (cmp_bb_postorder): Simplify.
15668 (sort_bbs_postorder): New function. Use it...
15669 (dom_walker::walk): ...here to optimize common cases.
15671 2017-07-25 Martin Liska <mliska@suse.cz>
15674 * ipa-visibility.c (function_and_variable_visibility): Make the
15675 redirection just on target that supports aliasing.
15676 Fix GNU coding style.
15678 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15681 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15683 * config/arm/rtems.h: Likewise.
15684 * config/bfin/rtems.h: Likewise.
15685 * config/i386/rtemself.h: Likewise.
15686 * config/lm32/rtems.h: Likewise.
15687 * config/m32c/rtems.h: Likewise.
15688 * config/m68k/rtemself.h: Likewise.
15689 * config/microblaze/rtems.h: Likewise.
15690 * config/mips/rtems.h: Likewise.
15691 * config/moxie/rtems.h: Likewise.
15692 * config/nios2/rtems.h: Likewise.
15693 * config/powerpcspe/rtems.h: Likewise.
15694 * config/rs6000/rtems.h: Likewise.
15695 * config/rtems.h: Likewise.
15696 * config/sh/rtems.h: Likewise.
15697 * config/sh/rtemself.h: Likewise.
15698 * config/sparc/rtemself.h: Likewise.
15700 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15703 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15704 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15705 * tree-ssa-structalias.c (alias_get_name): Same.
15707 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15710 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15711 instructions if no du chain is found.
15713 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15715 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15717 2017-07-25 Richard Biener <rguenther@suse.de>
15719 PR middle-end/81505
15720 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15723 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15725 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15726 upper-regs options.
15727 (ISA_2_7_MASKS_SERVER): Likewise.
15728 (ISA_3_0_MASKS_IEEE): Likewise.
15729 (OTHER_P8_VECTOR_MASKS): Likewise.
15730 (OTHER_VSX_VECTOR_MASKS): Likewise.
15731 (POWERPC_MASKS): Likewise.
15732 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15733 duplicate list of options.
15734 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15735 explicit -mupper-regs options.
15736 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15737 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15738 alias for -mupper-regs-df.
15739 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15740 (rs6000_init_hard_regno_mode_ok): Likewise.
15741 (rs6000_option_override_internal): Likewise.
15742 (rs6000_opt_masks): Likewise.
15743 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15744 options in terms of whether -mvsx or -mpower8-vector was used.
15745 (TARGET_UPPER_REGS_DI): Likewise.
15746 (TARGET_UPPER_REGS_SF): Likewise.
15747 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15748 -mupper-regs-* options.
15750 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15752 * passes.c (emergency_dump_function): Print some empty lines and a
15753 header before the RTL dump.
15755 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15757 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15759 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15762 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15763 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15765 2017-07-24 Carl Love <cel@us.ibm.com>
15767 * config/rs6000/rs6000-c.c: Add support for built-in functions
15768 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15769 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15770 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15771 vec_extract_fp_from_shortl): Add defines for the two builtins.
15772 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15773 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15775 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15776 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15777 * doc/extend.texi: Update the built-in documentation file for the
15778 new built-in function.
15780 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15783 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15785 * doc/generic.texi: Likewise.
15786 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15787 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15789 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15791 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15792 (aarch64_mls_elt_merge<mode>): Likewise.
15794 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15796 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15797 having __cxa_atexit.
15799 2017-07-23 Michael Collison <michael.collison@arm.com>
15801 * config/arm/arm.c (arm_option_override): Deprecate
15802 use of -mstructure-size-boundary.
15803 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15804 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15806 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15809 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15810 Reduce cost estimate for direct moves.
15812 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15815 * config/i386/i386.c (ix86_option_override_internal): Disable
15816 BMI, BMI2 and TBM instructions for -m16.
15818 2017-07-21 Carl Love <cel@us.ibm.com>
15820 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15821 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15822 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15823 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15824 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15825 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15826 VMULOSW): New enum "unspec" values.
15827 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15828 altivec_vmulosw): New patterns.
15829 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15830 VMULOSW): Add definitions.
15832 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15834 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15835 (qdf24xx): Likewise.
15836 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15837 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15838 (AARCH64_FL_V8_1): Renumber.
15839 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15840 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15841 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15842 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15843 rdma to feature modifiers list.
15845 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15847 PR middle-end/56727
15848 * ipa-visibility (function_and_variable_visibility): Convert
15849 recursive PLT call to direct call if appropriate.
15851 2017-07-21 Andrew Pinski <apinski@cavium.com>
15853 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15854 operand 1 to see if the types precision matches.
15855 * fold-const.c (operand_equal_p): Likewise.
15857 2017-07-21 Richard Biener <rguenther@suse.de>
15859 PR tree-optimization/81303
15860 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15861 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15862 (vect_peeling_hash_get_lowest_cost): Adjust.
15863 (vect_enhance_data_refs_alignment): Likewise. Use
15864 vect_get_peeling_costs_all_drs to compute the penalty for no
15865 peeling to match up costs.
15867 2017-07-21 Richard Biener <rguenther@suse.de>
15869 PR tree-optimization/81500
15870 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15871 we didn't identify a reduction path.
15873 2017-07-21 Tom de Vries <tom@codesourcery.com>
15874 Cesar Philippidis <cesar@codesourcery.com>
15876 PR gcov-profile/81442
15877 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15880 2017-07-21 Tom de Vries <tom@codesourcery.com>
15883 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15885 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15886 nvptx_override_options_after_change.
15888 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15890 * dwarf2out.c (output_file_names): Avoid double testing for
15891 dwarf_version >= 5.
15893 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15895 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15897 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15899 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15901 (try_crossjump_to_edge): Do not punt on partitioned functions.
15903 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15905 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15906 Put all BBs reachable only via paths crossing cold region to cold
15908 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15910 2016-07-21 Richard Biener <rguenther@suse.de>
15912 PR tree-optimization/81303
15913 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15914 into account prologue and epilogue iterations when raising
15915 min_profitable_iters to sth at least covering one vector iteration.
15917 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15919 * config/arm/arm.c (arm_test_cpu_arch_dat):
15922 2017-07-20 Nathan Sidwell <nathan@acm.org>
15924 Remove TYPE_METHODS.
15925 * tree.h (TYPE_METHODS): Delete.
15926 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15927 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15928 (dbxout_type_methods): Scan TYPE_FIELDS.
15929 (dbxout_type): Don't check TYPE_METHODS here.
15930 * function.c (use_register_for_decl): Always ignore register for
15931 class types when not optimizing.
15932 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15933 * tree.c (free_lang_data_in_type): Stitch out member functions and
15934 templates from TYPE_FIELDS.
15935 (build_distinct_type_copy, verify_type_variant,
15936 verify_type): Member fns are on TYPE_FIELDS.
15937 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15938 * tree-pretty-print.c (dump_generic_node): Likewise.
15940 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15943 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15944 V2TImode and V4TImode.
15945 (ix86_expand_vector_extract): Likewise.
15946 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15947 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15948 (ssescalarmode): Handle V4TImode and V2TImode.
15949 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15950 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15951 (VEXTRACTI128_MODE): New mode iterator.
15952 (splitter for *vec_extractv?ti first element): New.
15953 (VEC_INIT_MODE): New mode iterator.
15954 (vec_init<mode>): Consolidate 3 expanders into one using
15955 VEC_INIT_MODE mode iterator.
15957 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15959 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15960 non_spilled_static_chain_regno_p.
15962 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15964 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15966 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15968 * bb-reorder.c (connect_traces): Allow copying of blocks within
15971 2017-07-20 Richard Biener <rguenther@suse.de>
15973 * gimple.h (gimple_phi_result): Add gphi * overload.
15974 (gimple_phi_result_ptr): Likewise.
15975 (gimple_phi_arg): Likewise. Adjust index assert to only
15976 allow actual argument accesses rather than all slots available
15978 (gimple_phi_arg_def): Add gphi * overload.
15979 * tree-phinodes.c (make_phi_node): Initialize only actual
15981 (resize_phi_node): Clear memory not covered by old node,
15982 do not initialize excess argument slots.
15983 (reserve_phi_args_for_new_edge): Initialize new argument slot
15986 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15988 PR tree-optimization/81388
15990 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15992 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15993 by removing computation of may_be_zero.
15995 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15996 Tom de Vries <tom@codesourcery.com>
15998 PR middle-end/81030
15999 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16000 when gimple level profile disagrees with what RTL expander did.
16002 2017-07-20 Richard Biener <rguenther@suse.de>
16004 PR tree-optimization/61171
16005 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16006 (vect_analyze_stmt): Add slp instance parameter.
16007 (vectorizable_reduction): Likewise.
16008 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16009 (vect_is_simple_reduction): Deal with chains not detected
16010 as SLP reduction chain, specifically not properly associated
16011 chains containing a mix of plus/minus.
16012 (get_reduction_op): Remove.
16013 (get_initial_defs_for_reduction): Simplify, pass in whether
16014 this is a reduction chain, pass in the SLP node for the PHIs.
16015 (vect_create_epilog_for_reduction): Get the SLP instance as
16017 (vectorizable_reduction): Get the SLP instance as arg.
16018 During analysis remember the SLP node with the PHIs in the
16019 instance. Simplify getting at the vectorized reduction PHIs.
16020 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16021 through SLP instance.
16022 (vect_slp_analyze_operations): Likewise.
16023 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16024 (vect_transform_stmt): Likewise.
16026 2017-07-20 Tom de Vries <tom@codesourcery.com>
16028 PR tree-optimization/81489
16029 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16030 read of phi arg location to before loop that modifies phi.
16032 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16034 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16037 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16039 PR middle-end/81331
16040 * except.c (execute): Fix ordering issue.
16042 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16044 PR rtl-optimization/81423
16045 * combine.c (make_compound_operation_int): Don't try to optimize
16046 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16048 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16050 PR rtl-optimization/81423
16051 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16052 with a constant that is -1 in the truncated to mode.
16054 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16056 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16057 (determine_unlikely_bbs): ... here.
16058 * predict.h (propagate_unlikely_bbs_forward): Declare.
16059 * cfgexpand.c (pass_expand::execute): Use it.
16060 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16062 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16063 propagate_unlikely_bbs_forward.
16065 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16067 PR middle-end/81331
16068 * except.c (maybe_add_nop_after_section_switch): New function.
16071 2017-07-19 Tom de Vries <tom@codesourcery.com>
16073 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16075 2017-07-19 Tom de Vries <tom@codesourcery.com>
16077 * gimple.h (gimple_phi_arg): Make assert more strict.
16079 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16081 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16082 * config/rs6000/mmintrin.h: New file.
16083 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16085 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16087 PR tree-optimization/81346
16088 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16090 2017-07-19 Tom de Vries <tom@codesourcery.com>
16092 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16094 2017-07-19 Tom de Vries <tom@codesourcery.com>
16096 * config/nvptx/nvptx-modes.def: Add V2DImode.
16097 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16098 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16099 (nvptx_output_mov_insn): Handle lack of mov.b128.
16100 (nvptx_print_operand): Handle 'H' and 'L' codes.
16101 (nvptx_vector_mode_supported): Allow V2DImode.
16102 (nvptx_preferred_simd_mode): New function.
16103 (nvptx_data_alignment): New function.
16104 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16105 nvptx_preferred_simd_mode.
16106 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16108 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16110 2017-07-19 Tom de Vries <tom@codesourcery.com>
16112 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16113 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16114 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16115 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16116 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16117 (mov<VECIM>_insn): New define_insn.
16118 (define_expand "mov<VECIM>): New define_expand.
16120 2017-07-19 Tom de Vries <tom@codesourcery.com>
16122 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16124 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16126 PR tree-optimization/81346
16127 * fold-const.h (fold_div_compare, range_check_type): Declare.
16128 * fold-const.c (range_check_type): New function.
16129 (build_range_check): Use range_check_type.
16130 (fold_div_compare): No longer static, rewritten into
16131 a match.pd helper function.
16132 (fold_comparison): Don't call fold_div_compare here.
16133 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16134 as helper function.
16136 2017-07-19 Nathan Sidwell <nathan@acm.org>
16138 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16139 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16140 * tree.c (find_decls_types_r, verify_type): Use
16141 TYPE_{MIN,MAX}_VALUE_RAW.
16142 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16143 (hash_tree): Likewise.
16144 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16146 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16149 2017-07-18 Tom de Vries <tom@codesourcery.com>
16151 PR middle-end/81464
16152 * omp-expand.c (expand_omp_for_static_chunk): Handle
16153 equal-argument loop exit phi.
16155 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16158 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16159 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16160 operand 2 predicate.
16161 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16162 operand 2 predicate.
16163 (ror,rol -> rorx splitters): Use const_int_operand as
16164 operand 2 predicate.
16166 2017-06-18 Richard Biener <rguenther@suse.de>
16168 PR tree-optimization/81410
16169 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16170 the gap in the ! slp_perm SLP case after each group.
16172 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16174 PR middle-end/81463
16175 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16178 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16180 PR middle-end/81462
16181 * predict.c (set_even_probabilities): Cleanup; do not affect
16182 probabilities that are already known.
16183 (combine_predictions_for_bb): Call even when count is set.
16185 2017-07-18 Nathan Sidwell <nathan@acm.org>
16187 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16190 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16193 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16194 optimization for loop niter analysis.
16196 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16199 * config/avr/avr.c (avr_optimize_casesi): Don't use
16200 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16202 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16204 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16205 body_cost_vec from _vect_peel_extended_info.
16206 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16207 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16210 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16212 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16214 2017-07-18 Richard Biener <rguenther@suse.de>
16216 PR tree-optimization/80620
16217 PR tree-optimization/81403
16218 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16219 info when re-using a VN table entry.
16221 2017-07-18 Richard Biener <rguenther@suse.de>
16223 PR tree-optimization/81418
16224 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16225 vectype_in. Verify that with lane-reducing reduction operations
16226 we have a single def-use cycle.
16228 2017-07-17 Carl Love <cel@us.ibm.com>
16230 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16232 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16233 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16234 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16235 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16236 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16237 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16238 VMULOSW): New enum "unspec" values.
16239 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16240 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16241 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16242 altivec_vmulosw): New patterns.
16243 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16244 VMULOSW): Add definitions.
16245 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16247 * config/alpha/alpha.c: Include predict.h.
16249 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16251 * tree-vrp.c (compare_assert_loc): Fix comparison function
16252 to return predictable results.
16254 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16256 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16258 (subdi3): Likewise.
16259 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16260 * doc/invoke.texi (mexpand-adddi): Update text.
16262 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16264 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16265 that also clobbers the CC register. The old expand code is moved
16267 (*arc_clzsi2): ... here.
16268 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16269 the CC register. The old expand code is moved to ...
16270 (arc_ctzsi2): ... here.
16272 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16274 * config/arc/arc.opt (mindexed-loads): Use initial value
16275 TARGET_INDEXED_LOADS_DEFAULT.
16276 (mauto-modify-reg): Use initial value
16277 TARGET_AUTO_MODIFY_REG_DEFAULT.
16278 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16279 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16280 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16281 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16283 2017-07-17 Martin Liska <mliska@suse.cz>
16286 * opts.c (finish_options): Do not allow -fgnu-tm
16287 w/ -fsanitize={kernel-,}address. Say sorry.
16289 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16292 * tree-loop-distribution.c (classify_partition): Only assert on
16293 numer of iterations.
16294 (merge_dep_scc_partitions): Delete prameter. Update function call.
16295 (distribute_loop): Remove code handling loop with unknown niters.
16296 (pass_loop_distribution::execute): Skip loop with unknown niters.
16298 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16301 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16302 function sort_partitions_by_post_order.
16304 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16306 PR tree-optimization/81374
16307 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16308 the max index of basic blocks, rather than number of basic blocks.
16310 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16312 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16314 (arc_legitimate_pic_operand_p): Likewise.
16315 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16317 (arc_needs_pcl_p): Likewise.
16318 (arc_legitimate_pc_offset_p): Likewise.
16319 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16320 function is also used in constrains.md.
16321 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16322 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16323 PLUS. Only return true/false in known cases, otherwise assert.
16324 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16325 is already called in arc_legitimate_constant_p.
16326 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16328 (LEGITIMATE_PIC_OPERAND_P): Use
16329 arc_raw_symbolic_reference_mentioned_p function.
16330 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16335 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16336 Andrew Burgess <andrew.burgess@embecosm.com>
16338 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16339 (arc_return_address_register): New function.
16340 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16341 (arc_handle_fndecl_attribute): Add naked attribute.
16342 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16343 (TARGET_WARN_FUNC_RETURN): Likewise.
16344 (arc_allocate_stack_slots_for_args): New function.
16345 (arc_warn_func_return): Likewise.
16346 (machine_function): Change type fn_type.
16347 (arc_compute_function_type): Consider new naked function type,
16348 change function return type.
16349 (arc_must_save_register): Adapt to handle new
16350 arc_compute_function_type's return type.
16351 (arc_expand_prologue): Likewise.
16352 (arc_expand_epilogue): Likewise.
16353 (arc_return_address_regs): Delete.
16354 (arc_return_address_register): New function.
16355 (arc_epilogue_uses): Use above function.
16356 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16357 (arc_function_type): Change encoding, add naked type.
16358 (ARC_INTERRUPT_P): Change to handle the new encoding.
16359 (ARC_FAST_INTERRUPT_P): Likewise.
16360 (ARC_NORMAL_P): Define.
16361 (ARC_NAKED_P): Likewise.
16362 (arc_compute_function_type): Delete prototype.
16363 * config/arc/arc.md (in_ret_delay_slot): Use
16364 arc_return_address_register function.
16365 (simple_return): Likewise.
16366 (p_return_i): Likewise.
16368 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16370 PR tree-optimization/81428
16371 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16372 can't be built for those types.
16374 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16376 Remove stuff dead since r239246.
16378 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16379 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16380 (avr_inform_devices): Remove dead stuff.
16382 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16384 * config/arm/arm_neon.h: Fix softp typo.
16386 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16388 PR tree-optimization/81365
16389 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16390 aggregate moves onto bb predecessor edges, make sure there are no
16391 loads that could alias the lhs in between the start of bb and the
16394 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16397 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16398 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16399 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16401 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16403 PR tree-optimization/81396
16404 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16405 (init_symbolic_number): Initialize it to 1.
16406 (perform_symbolic_merge): Add n_ops from both operands into the new
16408 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16409 without base_addr as useless if they need more than one operation.
16410 (bswap_replace): Handle !bswap case for NULL base_addr.
16412 2017-07-17 Tom de Vries <tom@codesourcery.com>
16415 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16418 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16420 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16421 conditional builtin define __FIX_LEON3FT_B2BST.
16423 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16425 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16426 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16429 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16431 PR rtl-optimization/81424
16432 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16433 to remove potential trapping from operands if -fnon-call-exceptions.
16435 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16437 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16438 profile_proability for scalling.
16439 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16441 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16443 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16445 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16447 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16448 fixpoint arithmetics.
16450 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16452 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16453 fixpoint arithmetics.
16455 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16457 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16458 fixpoint arithmetics.
16460 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16462 * profile-count.h (profile_probability::from_reg_br_prob_note,
16463 profile_probability::to_reg_br_prob_note): New functions.
16464 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16465 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16466 * predict.c (probability_reliable_p): Update.
16467 (edge_probability_reliable_p): Update.
16468 (br_prob_note_reliable_p): Update.
16469 (invert_br_probabilities): Update.
16470 (add_reg_br_prob_note): New function.
16471 (combine_predictions_for_insn): Update.
16472 * asan.c (asan_clear_shadow): Update.
16473 * cfgbuild.c (compute_outgoing_frequencies): Update.
16474 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16475 (update_br_prob_note): Update.
16476 (rtl_verify_edges): Update.
16477 (purge_dead_edges): Update.
16478 (fixup_reorder_chain): Update.
16479 * emit-rtl.c (try_split): Update.
16480 * ifcvt.c (cond_exec_process_insns): Update.
16481 (cond_exec_process_if_block): Update.
16482 (dead_or_predicable): Update.
16483 * internal-fn.c (expand_addsub_overflow): Update.
16484 (expand_neg_overflow): Update.
16485 (expand_mul_overflow): Update.
16486 * loop-doloop.c (doloop_modify): Update.
16487 * loop-unroll.c (compare_and_jump_seq): Update.
16488 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16489 * predict.h: Update.
16490 * reorg.c (mostly_true_jump): Update.
16492 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16493 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16494 * config/arc/arc.c: (emit_unlikely_jump): Update.
16495 * config/arm/arm.c: (emit_unlikely_jump): Update.
16496 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16497 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16498 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16499 (ix86_print_operand): Update.
16500 (ix86_split_fp_branch): Update.
16501 (predict_jump): Update.
16502 * config/ia64/ia64.c (ia64_print_operand): Update.
16503 * config/mmix/mmix.c (mmix_print_operand): Update.
16504 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16505 (rs6000_expand_split_stack_prologue): Update.
16506 * config/rs6000/rs6000.c: Update.
16507 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16508 (s390_expand_vec_movstr): Update.
16509 (s390_expand_cs_tdsi): Update.
16510 (s390_expand_split_stack_prologue): Update.
16511 * config/sh/sh.c (sh_print_operand): Update.
16512 (expand_cbranchsi4): Update.
16513 (expand_cbranchdi4): Update.
16514 * config/sparc/sparc.c (output_v9branch): Update.
16515 * config/spu/spu.c (get_branch_target): Update.
16516 (ea_load_store_inline): Update.
16517 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16518 * config/tilepro/tilepro.c: Update.
16520 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16522 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16523 (gimplify_save_expr): Likewise.
16525 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16527 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16529 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16531 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16533 * timevar.def (TV_IPA_FNSUMMARY): Define.
16535 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16537 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16538 to back store errata sensitive sequence from being generated.
16539 (sqrtdf2_fix): Likewise.
16541 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16543 * tree-ssa-threadupdate.c (compute_path_counts,
16544 update_joiner_offpath_counts): Use profile_probability.
16546 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16549 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16551 * config/arm/arm-c.c (arm_cpu_builtins): Define
16552 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16554 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16556 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16557 array entries to represent __ieee128 versions of the
16558 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16559 scalar_extract_sig, and scalar_insert_exp built-in functions.
16560 (altivec_resolve_overloaded_builtin): Add special case handling
16561 for the __builtin_scalar_insert_exp function, as represented by
16562 the P9V_BUILTIN_VEC_VSIEDP constant.
16563 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16564 exponent support for __ieee128 argument.
16565 (VSESQP): Add scalar extract signature support for __ieee128
16567 (VSTDCNQP): Add scalar test negative support for __ieee128
16569 (VSIEQP): Add scalar insert exponent support for __int128 argument
16570 with __ieee128 result.
16571 (VSIEQPF): Add scalar insert exponent support for __ieee128
16572 argument with __ieee128 result.
16573 (VSTDCQP): Add scalar test data class support for __ieee128
16575 (VSTDCNQP): Add overload support for scalar test negative with
16576 __ieee128 argument.
16577 (VSTDCQP): Add overload support for scalar test data class
16578 __ieee128 argument.
16579 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16580 UNSPEC_VSX_SXSIGDP.
16581 (UNSPEC_VSX_SIEXPQP): New constant.
16582 (xsxexpqp): New insn for VSX scalar extract exponent quad
16584 (xsxsigqp): New insn for VSX scalar extract significand quad
16586 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16587 precision with floating point argument.
16588 (xststdcqp): New expand for VSX scalar test data class quad
16590 (xststdcnegqp): New expand for VSX scalar test negative quad
16592 (xststdcqp): New insn to match expansions for VSX scalar test data
16593 class quad precision and VSX scalar test negative quad precision.
16594 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16595 special case operand checking to enforce that second operand of
16596 VSX scalar test data class with quad precision argument is a 7-bit
16598 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16599 prototypes and descriptions of __ieee128 versions of
16600 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16601 scalar_test_data_class, and scalar_test_neg built-in functions.
16603 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16605 PR tree-optimization/81162
16606 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16607 replace a negate with an add.
16609 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16611 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16613 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16615 * config/arm/arm-c.c (arm_cpu_builtins): Define
16616 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16618 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16620 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16621 (armv8-r): Set ARM Cortex-R52 as default CPU.
16622 * config/arm/arm-tables.opt: Regenerate.
16623 * config/arm/arm-tune.md: Regenerate.
16624 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16626 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16627 extension for -mcpu=cortex-r52.
16629 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16631 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16632 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16633 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16634 (fp-armv8): Define it as FP_ARMv8 only.
16635 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16636 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16638 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16639 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16640 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16641 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16642 than TARGET_FPU_ARMV8.
16643 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16644 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16645 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16647 * config/arm/neon.md (neon_vrint): Likewise.
16648 (neon_vcvt): Likewise.
16649 (neon_<fmaxmin_op><mode>): Likewise.
16650 (<fmaxmin><mode>3): Likewise.
16651 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16652 * config/arm/predicates.md (arm_cond_move_operator): Check against
16653 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16655 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16657 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16658 to top of function.
16660 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16662 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16663 loop in comment with memset.
16665 2017-07-14 Martin Liska <mliska@suse.cz>
16667 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16668 * dwarf2out.c (is_java): Remove the function.
16669 (output_pubname): Remove usage of the function.
16670 (lower_bound_default): Remove usage of DW_LANG_Java.
16671 (gen_compile_unit_die): Likewise.
16672 * gcc.c: Remove compiler defaults for .java and .zip files.
16673 * gimple-expr.c (remove_suffix): Change as there's no longer
16674 extension than 4-letter one.
16675 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16676 (gimplify_save_expr): Likewise.
16677 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16678 as it's possible even for other languages than Java.
16679 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16680 * lto-opts.c (lto_write_options): Remove reference to Java.
16681 * opts.c (strip_off_ending): Update file extension handling.
16682 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16683 * tree-eh.c (lower_resx): Likewise.
16684 * tree.c (free_lang_data_in_type): Remove dead code.
16685 (find_decls_types_r): Likewise.
16686 (build_common_builtin_nodes): Remove Java from a comment.
16687 (verify_type): Remove dead code.
16688 * varasm.c (assemble_external): Remove Java from a comment.
16690 2017-07-14 Martin Liska <mliska@suse.cz>
16692 * opts.c (finish_options): Add quotes.
16693 (common_handle_option): Likewise.
16695 2017-07-14 Martin Liska <mliska@suse.cz>
16697 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16698 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16699 Remove N_SO_PASCAL.
16700 * dwarf2out.c (lower_bound_default): Do not handle
16702 (gen_compile_unit_die): Likewise.
16703 * gcc.c: Remove default extension binding for GNU Pascal.
16704 * stmt.c: Remove Pascal language from a comment.
16705 * xcoffout.c: Likewise.
16707 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16710 * diagnostic-show-locus.c (fixit_cmp): New function.
16711 (layout::layout): Sort m_fixit_hints.
16712 (column_range::column_range): Assert that the values are valid.
16713 (struct char_span): New struct.
16714 (correction::overwrite): New method.
16715 (struct source_line): New struct.
16716 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16717 calls in terms of classes source_line and char_span, and
16718 correction::overwrite.
16719 (selftest::test_overlapped_fixit_printing_2): New function.
16720 (selftest::diagnostic_show_locus_c_tests): Call it.
16722 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16724 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16725 early if there is no lhs.
16727 2017-07-13 Martin Liska <mliska@suse.cz>
16729 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16730 (gen_reference_type_die): Likewise.
16731 * stor-layout.c: Remove Pascal-related comment.
16733 2017-07-13 Martin Liska <mliska@suse.cz>
16735 * opts.c (finish_options): Add quotes to error messages.
16736 (parse_sanitizer_options): Likewise.
16738 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16740 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16742 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16744 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16746 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16748 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16750 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16752 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16755 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16756 provides the hardware capability bits, define the macro
16757 __BUILTIN_CPU_SUPPORTS__.
16758 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16759 if GLIBC does not provide the hardware capability bits. Add a
16760 gcc_unreachable call if the built-in cpu function is neither
16761 __builtin_cpu_is nor __builtin_cpu_supports.
16762 (rs6000_get_function_versions_dispatcher): Change the warning
16763 that an old GLIBC is used which does not export the capability
16764 bits to be an error.
16765 * doc/extend.texi (target_clones attribute): Document the
16766 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16767 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16768 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16769 the macros defined by GCC if the newer GLIBC is available.
16771 2017-07-12 Jeff Law <law@redhat.com>
16773 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16774 remaining includes slightly.
16775 * config/riscv/riscv-builtins.c: Include profile-count.h.
16777 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16780 * config/avr/avr.c (avr_set_current_function): In diagnostic
16781 messages: Quote keywords and (parts of) identifiers.
16782 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16785 2017-07-12 Carl Love <cel@us.ibm.com>
16787 * config/rs6000/rs6000-c.c: Add support for built-in functions
16788 vector bool char vec_revb (vector bool char);
16789 vector bool short vec_revb (vector short char);
16790 vector bool int vec_revb (vector bool int);
16791 vector bool long long vec_revb (vector bool long long);
16792 * doc/extend.texi: Update the built-in documentation file for the
16793 new built-in functions.
16795 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16797 * config/s390/s390.md: Remove movcc splitter.
16799 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16801 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16802 load/store on condition.
16804 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16807 * config/avr/avr.c (avr_encode_section_info)
16808 [progmem && !TREE_READONLY]: Error if progmem object needs
16811 2017-07-11 Michael Collison <michael.collison@arm.com>
16813 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16816 2017-07-11 Carl Love <cel@us.ibm.com>
16818 * config/rs6000/rs6000-c.c: Add support for builtins
16819 vector unsigned int vec_parity_lsbb (vector signed int);
16820 vector unsigned int vec_parity_lsbb (vector unsigned int);
16821 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16822 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16823 vector unsigned long long vec_parity_lsbb (vector signed long long);
16824 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16825 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16826 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16827 * doc/extend.texi: Update the built-in documentation file for the
16828 new built-in functions.
16830 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16832 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16833 (layout::m_primary_loc): New field.
16834 (layout::layout): Initialize new field. Move location filtering
16835 logic from here to...
16836 (layout::maybe_add_location_range): ...this new method. Add
16837 support for filtering to just the lines already specified by other
16839 (layout::will_show_line_p): New method.
16840 (gcc_rich_location::add_location_if_nearby): New method.
16841 (selftest::test_add_location_if_nearby): New test function.
16842 (selftest::diagnostic_show_locus_c_tests): Call it.
16843 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16846 2017-07-11 Tom de Vries <tom@codesourcery.com>
16848 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16849 (bb_first_real_insn): New function.
16850 (nvptx_single): Add extra initialization of broadcasted condition
16853 2017-07-11 Nathan Sidwell <nathan@acm.org>
16855 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16857 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16859 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16860 Binutils doc as TEXI will mess them up.
16861 * doc/invoke.texi (AVR Options): Same here.
16863 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16865 * config/sparc/sparc.opt (mfix-ut700): New option.
16866 (mfix-gr712rc): Likewise.
16867 (sparc_fix_b2bst): New variable.
16868 * doc/invoke.texi (SPARC options): Document them.
16869 (ARM options): Fix warnings.
16870 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16871 instructions to prevent sequences that can trigger the store-store
16872 errata for certain LEON3FT processors.
16873 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16874 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16875 * config/sparc/sparc.md (fix_b2bst): New attribute.
16876 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16878 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16881 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16883 (*rsqrtsf2_sse): Ditto.
16885 (div<mode>3): Macroize insn from divdf3 and divsf3
16886 using MODEF mode iterator.
16888 2017-07-10 Martin Sebor <msebor@redhat.com>
16890 PR tree-optimization/80397
16891 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16892 instead of testing for equality to INTEGER_TYPE.
16894 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16896 * config.gcc: Remove uclibc from arc target spec.
16898 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16900 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16902 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16905 * lto-wrapper.c (remove_option): New function.
16906 (merge_and_complain): Merge PIC/PIE options more realistically.
16908 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16910 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16914 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16915 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16916 * config.in: Regenerate.
16917 * configure: Regenerate.
16918 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16919 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16920 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16921 (TARGET_GASISR_PROLOGUES): ...target mask.
16922 * common/config/avr/avr-common.c
16923 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16924 Set -mgas-isr-prologues.
16925 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16926 INSERT_PASS_BEFORE for it.
16927 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16928 * config/avr/avr.c (avr_option_override)
16929 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16930 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16931 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16932 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16933 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16935 (make_avr_pass_pre_proep): New function.
16936 (emit_push_sfr) <treg>: Add argument to function and use it
16937 instead of TMP_REG.
16938 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16939 and set machine->gasisr.yes.
16940 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16941 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16942 __gcc_isr.n_pushed to .L__stack_usage.
16943 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16944 (avr_asm_final_postscan_insn): ...this new static function.
16945 * config/avr/avr.h (machine_function)
16946 <is_no_gccisr, use_L__stack_usage>: New fields.
16947 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16948 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16949 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16950 (gasisr, *gasisr): New expander and insn.
16951 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16952 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16953 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16955 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16957 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16960 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16962 Move jump-tables out of .text again.
16965 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16966 (ASM_OUTPUT_ADDR_VEC): New function.
16967 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16968 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16969 INSN_ADDRESSes as asm comment.
16970 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16971 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16972 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16973 * config/avr/avr.md (*tablejump): Adjust comment.
16974 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16975 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16977 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16978 (avr_output_addr_vec): New proto.
16979 (avr_log_t) <insn_addresses>: New field.
16981 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16984 * config/i386/i386.c (ix86_function_arg_advance): Set
16985 outgoing_args_on_stack to true if there are outgoing arguments
16987 (ix86_function_arg): Likewise.
16988 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16989 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16990 * config/i386/i386.h (machine_function): Add
16991 outgoing_args_on_stack.
16993 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16995 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16996 supporting pthreds.
16997 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16999 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17001 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17002 (REAL_H): Remove $(MACHMODE_H).
17003 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17005 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17006 $(MACHMODE_H) and double-int.h.
17007 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17009 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17012 2017-07-07 Andrew Pinski <apinski@cavium.com>
17014 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17015 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17017 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17019 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17020 Add warning if GCC was not configured to link against a GLIBC that
17021 exports the hardware capability bits.
17022 (make_resolver_func): Make resolver function private and not a
17023 COMDAT function. Create the name with clone_function_name instead
17024 of make_unique_name.
17027 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17028 correct operand in doing the split.
17030 2017-07-07 Carl Love <cel@us.ibm.com>
17032 * config/rs6000/rs6000-c: Add support for built-in function
17033 vector unsigned short vec_pack_to_short_fp32 (vector float,
17035 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17036 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17037 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17038 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17039 (convert_4f32_8i16): Add define_expand.
17040 * doc/extend.texi: Update the built-in documentation file for the
17041 new built-in function.
17043 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17045 * config/sparc/m8.md: New file.
17046 * config/sparc/sparc.md: Include m8.md.
17048 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17050 * config/sparc/sparc.opt: New option -mvis4b.
17051 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17052 (sparc_option_override): Handle VIS4B.
17053 (enum sparc_builtins): Define
17054 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17055 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17056 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17057 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17058 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17059 (check_constant_argument): New function.
17060 (sparc_vis_init_builtins): Define builtins
17061 __builtin_vis_dictunpack{8,16,32},
17062 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17063 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17064 __builtin_vis_fpcmpde{8,16,32}shl and
17065 __builtin_vis_fpcmpur{8,16,32}shl.
17066 (sparc_expand_builtin): Check that the constant operands to
17067 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17068 constant and in range.
17069 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17071 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17072 (SPARC_IMM5_P): Likewise.
17073 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17074 (enabled): Handle vis4b.
17075 (UNSPEC_DICTUNPACK): New unspec.
17076 (UNSPEC_FPCMPSHL): Likewise.
17077 (UNSPEC_FPUCMPSHL): Likewise.
17078 (UNSPEC_FPCMPDESHL): Likewise.
17079 (UNSPEC_FPCMPURSHL): Likewise.
17080 (cpu_feature): New CPU feature `vis4b'.
17081 (dictunpack{8,16,32}): New insns.
17082 (FPCSMODE): New mode iterator.
17083 (fpcscond): New code iterator.
17084 (fpcsucond): Likewise.
17085 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17086 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17087 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17088 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17089 * config/sparc/constraints.md: Define constraints `q' for unsigned
17090 2-bit integer constants and `t' for unsigned 5-bit integer
17092 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17094 (imm5_operand_dictunpack16): Likewise.
17095 (imm5_operand_dictunpack32): Likewise.
17096 (imm2_operand): Likewise.
17097 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17098 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17099 ditunpack* and fpcmp*shl builtins.
17101 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17103 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17104 * config.in: Add HAVE_AS_SPARC6 define.
17105 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17107 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17109 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17110 (CPP_CPU_SPEC): Handle m8.
17111 (ASM_CPU_SPEC): Likewise.
17112 * config/sparc/sparc-opts.h (enum processor_type): Add
17114 * config/sparc/sparc.c (m8_costs): New struct.
17115 (sparc_option_override): Handle TARGET_CPU_m8.
17116 (sparc32_initialize_trampoline): Likewise.
17117 (sparc64_initialize_trampoline): Likewise.
17118 (sparc_issue_rate): Likewise.
17119 (sparc_register_move_cost): Likewise.
17120 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17121 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17122 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17123 (CPP_CPU_SPEC): Handle M8.
17124 (ASM_CPU_SPEC): Likewise.
17125 (AS_M8_FLAG): Define.
17126 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17127 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17128 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17130 * configure: Regenerate.
17131 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17134 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17136 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17138 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17139 ("*movdi_insn_sp32"): Do not set v3pipe.
17140 ("*movsi_insn"): Likewise.
17141 ("*movdi_insn_sp64"): Likewise.
17142 ("*movsf_insn"): Likewise.
17143 ("*movdf_insn_sp32"): Likewise.
17144 ("*movdf_insn_sp64"): Likewise.
17145 ("*zero_extendsidi2_insn_sp64"): Likewise.
17146 ("*sign_extendsidi2_insn"): Likewise.
17147 ("*mov<VM32:mode>_insn"): Likewise.
17148 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17149 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17150 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17151 ("<vlop:code><VL:mode>3"): Likewise.
17152 ("*not_<vlop:code><VL:mode>3"): Likewise.
17153 ("*nand<VL:mode>_vis"): Likewise.
17154 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17155 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17156 ("one_cmpl<VL:mode>2"): Likewise.
17157 ("faligndata<VM64:mode>_vis"): Likewise.
17158 ("alignaddrsi_vis"): Likewise.
17159 ("alignaddrdi_vis"): Likweise.
17160 ("alignaddrlsi_vis"): Likewise.
17161 ("alignaddrldi_vis"): Likewise.
17162 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17163 ("bmaskdi_vis"): Likewise.
17164 ("bmasksi_vis"): Likewise.
17165 ("bshuffle<VM64:mode>_vis"): Likewise.
17166 ("cmask8<P:mode>_vis"): Likewise.
17167 ("cmask16<P:mode>_vis"): Likewise.
17168 ("cmask32<P:mode>_vis"): Likewise.
17169 ("pdistn<P:mode>_vis"): Likewise.
17170 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17172 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17174 * config/sparc/sparc.md ("subtype"): New insn attribute.
17175 ("*wrgsr_sp64"): Set insn subtype.
17176 ("*rdgsr_sp64"): Likewise.
17177 ("alignaddrsi_vis"): Likewise.
17178 ("alignaddrdi_vis"): Likewise.
17179 ("alignaddrlsi_vis"): Likewise.
17180 ("alignaddrldi_vis"): Likewise.
17181 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17182 ("fexpand_vis"): Likewise.
17183 ("fpmerge_vis"): Likewise.
17184 ("faligndata<VM64:mode>_vis"): Likewise.
17185 ("bshuffle<VM64:mode>_vis"): Likewise.
17186 ("cmask8<P:mode>_vis"): Likewise.
17187 ("cmask16<P:mode>_vis"): Likewise.
17188 ("cmask32<P:mode>_vis"): Likewise.
17189 ("fchksm16_vis"): Likewise.
17190 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17191 ("fmean16_vis"): Likewise.
17192 ("fp<plusminus_insn>64_vis"): Likewise.
17193 ("<plusminus_insn>v8qi3"): Likewise.
17194 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17195 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17196 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17197 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17198 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17199 ("*movqi_insn"): Likewise.
17200 ("*movhi_insn"): Likewise.
17201 ("*movsi_insn"): Likewise.
17202 ("movsi_pic_gotdata_op"): Likewise.
17203 ("*movdi_insn_sp32"): Likewise.
17204 ("*movdi_insn_sp64"): Likewise.
17205 ("movdi_pic_gotdata_op"): Likewise.
17206 ("*movsf_insn"): Likewise.
17207 ("*movdf_insn_sp32"): Likewise.
17208 ("*movdf_insn_sp64"): Likewise.
17209 ("*zero_extendhisi2_insn"): Likewise.
17210 ("*zero_extendqihi2_insn"): Likewise.
17211 ("*zero_extendqisi2_insn"): Likewise.
17212 ("*zero_extendqidi2_insn"): Likewise.
17213 ("*zero_extendhidi2_insn"): Likewise.
17214 ("*zero_extendsidi2_insn_sp64"): Likewise.
17215 ("ldfsr"): Likewise.
17216 ("prefetch_64"): Likewise.
17217 ("prefetch_32"): Likewise.
17218 ("tie_ld32"): Likewise.
17219 ("tie_ld64"): Likewise.
17220 ("*tldo_ldub_sp32"): Likewise.
17221 ("*tldo_ldub1_sp32"): Likewise.
17222 ("*tldo_ldub2_sp32"): Likewise.
17223 ("*tldo_ldub_sp64"): Likewise.
17224 ("*tldo_ldub1_sp64"): Likewise.
17225 ("*tldo_ldub2_sp64"): Likewise.
17226 ("*tldo_ldub3_sp64"): Likewise.
17227 ("*tldo_lduh_sp32"): Likewise.
17228 ("*tldo_lduh1_sp32"): Likewise.
17229 ("*tldo_lduh_sp64"): Likewise.
17230 ("*tldo_lduh1_sp64"): Likewise.
17231 ("*tldo_lduh2_sp64"): Likewise.
17232 ("*tldo_lduw_sp32"): Likewise.
17233 ("*tldo_lduw_sp64"): Likewise.
17234 ("*tldo_lduw1_sp64"): Likewise.
17235 ("*tldo_ldx_sp64"): Likewise.
17236 ("*mov<VM32:mode>_insn"): Likewise.
17237 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17238 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17240 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17242 * config/sparc/sparc.md ("type"): New insn type viscmp.
17243 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17245 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17246 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17247 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17248 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17250 ("n7_vis_logical_11cycle"): Likewise.
17251 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17252 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17253 * config/sparc/niagara.md ("niag_vis"): Likewise.
17254 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17255 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17257 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17259 * config/sparc/sparc.md: New instruction type `bmask'.
17260 (bmaskdi_vis): Use the `bmask' type.
17261 (bmasksi_vis): Likewise.
17262 * config/sparc/ultra3.md (us3_array): Likewise.
17263 * config/sparc/niagara7.md (n7_array): Likewise.
17264 * config/sparc/niagara4.md (n4_array): Likewise.
17265 * config/sparc/niagara2.md (niag2_vis): Likewise.
17266 (niag3_vis): Likewise.
17267 * config/sparc/niagara.md (niag_vis): Likewise.
17269 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17271 * ipa-comdats.c: Remove optimize check from gate.
17272 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17273 for functions not optimized.
17274 (ipa_fn_summary_read): Skip optimize check.
17275 (ipa_fn_summary_write): Likewise.
17276 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17278 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17280 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17281 (check_callers): Give up on caller which is not optimized.
17282 (inline_small_functions): Likewise.
17283 (ipa_inline): Do not give up when not optimizing.
17284 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17285 away unoptimizes cdtors.
17286 (whole_program_function_and_variable_visibility): Do
17287 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17288 * ipa.c (process_references): Do not check optimize.
17289 (symbol_table::remove_unreachable_nodes): Update optimize check.
17290 (set_writeonly_bit): Update optimize check.
17291 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17292 (pass_ipa_single_use::gate): Remove.
17294 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17296 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17297 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17298 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17299 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17300 permute_load, permute_store, adjust_extract, adjust_splat,
17301 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17302 replace_swap_with_copy, dump_swap_insn_table,
17303 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17304 recombine_lvx_pattern, recombine_stvx_pattern,
17305 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17306 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17307 to file rs6000-p8swap.c.
17308 * config/rs6000/rs6000-p8swap.c: New file.
17309 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17310 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17311 and rs6000*-*-* targets.
17313 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17315 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17317 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17319 * lto-wrapper.c (merge_and_complain): Do not merge
17320 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17321 fsigned_zeros, ftrapping_math, fwrapv.
17322 (append_compiler_options): Do not track these options.
17323 (append_linker_options): Likewie
17325 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17327 * cgraphunit.c (cgraph_node::finalize_function): When
17328 !flag_toplevel_reorde set no_reorder flag.
17329 (varpool_node::finalize_decl): Likewise.
17330 (symbol_table::compile): Drop no toplevel reorder path.
17332 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17334 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17335 edges; zero probability is not better than uninitialized.
17337 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17339 * asan.h (asan_sanitize_allocas_p): Declare.
17340 * asan.c (asan_sanitize_allocas_p): New function.
17341 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17342 (handle_builtin_alloca): Likewise.
17343 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17344 if !asan_sanitize_allocas_p.
17345 * params.def (asan-instrument-allocas): Add new option.
17346 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17347 * opts.c (common_handle_option): Disable allocas sanitization for
17350 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17352 * asan.c: Include gimple-fold.h.
17353 (get_last_alloca_addr): New function.
17354 (handle_builtin_stackrestore): Likewise.
17355 (handle_builtin_alloca): Likewise.
17356 (asan_emit_allocas_unpoison): Likewise.
17357 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17358 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17359 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17360 (instrument_builtin_call): Pass gimple iterator to
17361 get_mem_refs_of_builtin_call.
17362 (last_alloca_addr): New global.
17363 * asan.h (asan_emit_allocas_unpoison): Declare.
17364 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17365 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17366 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17367 if function calls alloca.
17368 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17369 * gimple-fold.h (replace_call_with_value): Declare.
17370 * internal-fn.c: Include asan.h.
17371 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17372 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17374 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17376 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17377 (C_SELFTEST_FLAGS): New.
17378 (CPP_SELFTEST_FLAGS): New.
17379 (SELFTEST_DEPS): New, from deps of s-selftest.
17380 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17381 (CPP_SELFTEST_DEPS): New.
17382 (selftest): Add dependency on s-selftest-c++.
17383 (s-selftest): Rename to...
17384 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17385 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17386 than SELFTEST_FLAGS.
17387 (selftest-gdb): Rename to...
17388 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17390 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17391 (selftest-valgrind): Rename to...
17392 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17394 (selftest-valgrind): Reintroduce as an alias for
17395 selftest-c-valgrind.
17396 (s-selftest-c++): New.
17397 (selftest-c++-gdb): New.
17398 (selftest-c++-valgrind): New.
17400 2017-07-06 Olivier Hainque <hainque@adacore.com>
17402 * gcc.c (process_command): When deciding if undefined variables
17403 should be ignored when processing specs, accept "gcc -v" as well.
17405 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17407 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17408 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17410 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17412 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17413 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17414 * config/arm/arm-tables.opt: Regenerate.
17415 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17417 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17419 2017-07-06 Carl Love <cel@us.ibm.com>
17421 * ChangeLog: Clean up from mid air collision
17423 2017-07-06 Carl Love <cel@us.ibm.com>
17425 * config/rs6000/rs6000-c.c: Add support for built-in functions
17426 vector signed int vec_subc (vector signed int, vector signed int);
17427 vector signed __int128 vec_subc (vector signed __int128,
17428 vector signed __int128);
17429 vector unsigned __int128 vec_subc (vector unsigned __int128,
17430 vector unsigned __int128);
17431 vector signed int vec_sube (vector signed int, vector signed int,
17432 vector signed int);
17433 vector unsigned int vec_sube (vector unsigned int,
17434 vector unsigned int,
17435 vector unsigned int);
17436 vector signed __int128 vec_sube (vector signed __int128,
17437 vector signed __int128,
17438 vector signed__int128);
17439 vector unsigned __int128 vec_sube (vector unsigned __int128,
17440 vector unsigned __int128,
17441 vector unsigned __int128);
17442 vector signed int vec_subec (vector signed int, vector signed int,
17443 vector signed int);
17444 vector unsigned int vec_subec (vector unsigned int,
17445 vector unsigned int,
17446 vector unsigned int);
17447 vector signed __int128 vec_subec (vector signed __int128,
17448 vector signed __int128,
17449 vector signed__int128);
17450 vector unsigned __int128 vec_subec (vector unsigned __int128,
17451 vector unsigned __int128,
17452 vector unsigned __int128);
17453 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17454 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17455 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17456 BU_ALTIVEC_OVERLOAD_X definitions.
17457 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17458 * doc/extend.texi: Update the built-in documentation file for the new
17459 built-in functions.
17461 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17464 * diagnostic-show-locus.c (layout::layout): Use start and finish
17465 spelling location for the start and finish of each range.
17466 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17467 Add unused aspect param.
17468 * input.c (expand_location_1): Add "aspect" param, and use it
17469 to access the correct part of the location.
17470 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17472 (expand_location_to_spelling_point): Likewise.
17473 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17474 param, and pass it to expand_location_1.
17476 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17478 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17479 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17480 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17481 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17482 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17483 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17484 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17485 _mm_maskz_getmant_ss): New intrinsics.
17486 (__builtin_ia32_getexpss128_mask): Changed to ...
17487 __builtin_ia32_getexpss128_round ... this.
17488 (__builtin_ia32_getexpsd128_mask): Changed to ...
17489 __builtin_ia32_getexpsd128_round ... this.
17490 * config/i386/i386-builtin-types.def
17491 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17492 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17493 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17494 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17495 __builtin_ia32_getmantss_mask_round): New builtins.
17496 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17497 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17498 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17499 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17500 * config/i386/sse.md
17501 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17502 avx512f_sgetexp<mode><mask_scalar_name>
17503 <round_saeonly_scalar_name> ... this.
17504 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17505 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17506 vgetexp<ssescalarmodesuffix>
17507 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17508 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17509 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17510 avx512f_vgetmant<mode><mask_scalar_name>
17511 <round_saeonly_scalar_name> ... this.
17512 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17513 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17514 vgetmant<ssescalarmodesuffix>
17515 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17516 %0<mask_scalar_operand4>, %1, %2
17517 <round_saeonly_scalar_mask_op4>, %3} ... this.
17518 * config/i386/subst.md (mask_scalar_operand4,
17519 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17520 round_saeonly_scalar_nimm_predicate): New subst attributes.
17522 2017-07-06 Julia Koval <julia.koval@intel.com>
17524 * config/i386/i386.c (ix86_erase_embedded_rounding):
17525 Remove code for old rounding pattern.
17527 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17529 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17531 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17533 * doc/sourcebuild.texi (Test Directives, Variants of
17534 dg-require-support): Add documentation for dg-require-stack-check.
17536 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17538 * config/i386/subst.md (mask_scalar, round_scalar,
17539 round_saeonly_scalar): New meta-templates.
17540 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17541 round_scalar_mask_operand3, round_scalar_mask_op3,
17542 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17543 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17544 round_saeonly_scalar_constraint,
17545 round_saeonly_scalar_prefix): New subst attribute.
17546 * config/i386/sse.md
17547 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17548 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17549 <round_scalar_name> ... this.
17550 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17551 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17552 <round_scalar_name> ... this.
17553 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17554 <sse>_vm<code><mode>3<mask_scalar_name>
17555 <round_saeonly_scalar_name> ... this.
17556 (v<plusminus_mnemonic><ssescalarmodesuffix>
17557 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17558 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17559 v<plusminus_mnemonic><ssescalarmodesuffix>
17560 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17561 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17562 (v<multdiv_mnemonic><ssescalarmodesuffix>
17563 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17564 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17565 v<multdiv_mnemonic><ssescalarmodesuffix>
17566 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17567 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17568 (v<maxmin_float><ssescalarmodesuffix>
17569 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17570 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17571 v<maxmin_float><ssescalarmodesuffix>
17572 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17573 %0<mask_scalar_operand3>, %1, %<iptr>2
17574 <round_saeonly_scalar_mask_op3>} ... this.
17576 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17578 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17579 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17581 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17582 Alan Hayward <alan.hayward@arm.com>
17583 David Sherwood <david.sherwood@arm.com>
17585 * combine.c (simplify_if_then_else): Remove "enum" before
17587 * compare-elim.c (can_eliminate_compare): Likewise.
17588 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17590 (aarch64_lookup_simd_builtin_type): Likewise.
17591 (aarch64_simd_builtin_type): Likewise.
17592 (aarch64_init_simd_builtin_types): Likewise.
17593 (aarch64_simd_expand_args): Likewise.
17594 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17596 (aarch64_reverse_mask): Likewise.
17597 (aarch64_simd_emit_reg_reg_move): Likewise.
17598 (aarch64_gen_adjusted_ldpstp): Likewise.
17599 (aarch64_ccmp_mode_to_code): Likewise.
17600 (aarch64_operands_ok_for_ldpstp): Likewise.
17601 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17602 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17604 (aarch64_min_divisions_for_recip_mul): Likewise.
17605 (aarch64_reassociation_width): Likewise.
17606 (aarch64_get_condition_code_1): Likewise.
17607 (aarch64_simd_emit_reg_reg_move): Likewise.
17608 (aarch64_simd_attr_length_rglist): Likewise.
17609 (aarch64_reverse_mask): Likewise.
17610 (aarch64_operands_ok_for_ldpstp): Likewise.
17611 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17612 (aarch64_gen_adjusted_ldpstp): Likewise.
17613 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17615 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17616 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17617 (arm_lookup_simd_builtin_type): Likewise.
17618 (arm_simd_builtin_type): Likewise.
17619 (arm_init_simd_builtin_types): Likewise.
17620 (arm_expand_builtin_args): Likewise.
17621 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17622 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17623 (ft32_setup_incoming_varargs): Likewise.
17624 (ft32_function_arg): Likewise.
17625 (ft32_function_arg_advance): Likewise.
17626 (ft32_pass_by_reference): Likewise.
17627 (ft32_arg_partial_bytes): Likewise.
17628 (ft32_valid_pointer_mode): Likewise.
17629 (ft32_addr_space_pointer_mode): Likewise.
17630 (ft32_addr_space_legitimate_address_p): Likewise.
17631 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17633 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17634 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17635 (iamcu_alignment): Likewise.
17636 (canonicalize_vector_int_perm): Likewise.
17637 (ix86_noce_conversion_profitable_p): Likewise.
17638 (ix86_mpx_bound_mode): Likewise.
17639 (ix86_operands_ok_for_move_multiple): Likewise.
17640 * config/microblaze/microblaze-protos.h
17641 (microblaze_expand_conditional_branch_reg): Likewise.
17642 * config/microblaze/microblaze.c
17643 (microblaze_expand_conditional_branch_reg): Likewise.
17644 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17646 (rs6000_reassociation_width): Likewise.
17647 (rs6000_invalid_binary_op): Likewise.
17648 (fusion_p9_p): Likewise.
17649 (emit_fusion_p9_load): Likewise.
17650 (emit_fusion_p9_store): Likewise.
17651 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17653 (riscv_hard_regno_mode_ok_p): Likewise.
17654 (riscv_address_insns): Likewise.
17655 (riscv_split_symbol): Likewise.
17656 (riscv_legitimize_move): Likewise.
17657 (riscv_function_value): Likewise.
17658 (riscv_hard_regno_nregs): Likewise.
17659 (riscv_expand_builtin): Likewise.
17660 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17661 (riscv_build_integer): Likewise.
17662 (riscv_split_integer): Likewise.
17663 (riscv_legitimate_constant_p): Likewise.
17664 (riscv_cannot_force_const_mem): Likewise.
17665 (riscv_regno_mode_ok_for_base_p): Likewise.
17666 (riscv_valid_base_register_p): Likewise.
17667 (riscv_valid_offset_p): Likewise.
17668 (riscv_valid_lo_sum_p): Likewise.
17669 (riscv_classify_address): Likewise.
17670 (riscv_legitimate_address_p): Likewise.
17671 (riscv_address_insns): Likewise.
17672 (riscv_load_store_insns): Likewise.
17673 (riscv_force_binary): Likewise.
17674 (riscv_split_symbol): Likewise.
17675 (riscv_force_address): Likewise.
17676 (riscv_legitimize_address): Likewise.
17677 (riscv_move_integer): Likewise.
17678 (riscv_legitimize_const_move): Likewise.
17679 (riscv_legitimize_move): Likewise.
17680 (riscv_address_cost): Likewise.
17681 (riscv_subword): Likewise.
17682 (riscv_output_move): Likewise.
17683 (riscv_canonicalize_int_order_test): Likewise.
17684 (riscv_emit_int_order_test): Likewise.
17685 (riscv_function_arg_boundary): Likewise.
17686 (riscv_pass_mode_in_fpr_p): Likewise.
17687 (riscv_pass_fpr_single): Likewise.
17688 (riscv_pass_fpr_pair): Likewise.
17689 (riscv_get_arg_info): Likewise.
17690 (riscv_function_arg): Likewise.
17691 (riscv_function_arg_advance): Likewise.
17692 (riscv_arg_partial_bytes): Likewise.
17693 (riscv_function_value): Likewise.
17694 (riscv_pass_by_reference): Likewise.
17695 (riscv_setup_incoming_varargs): Likewise.
17696 (riscv_print_operand): Likewise.
17697 (riscv_elf_select_rtx_section): Likewise.
17698 (riscv_save_restore_reg): Likewise.
17699 (riscv_for_each_saved_reg): Likewise.
17700 (riscv_register_move_cost): Likewise.
17701 (riscv_hard_regno_mode_ok_p): Likewise.
17702 (riscv_hard_regno_nregs): Likewise.
17703 (riscv_class_max_nregs): Likewise.
17704 (riscv_memory_move_cost): Likewise.
17705 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17706 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17707 (rl78_addr_space_address_mode): Likewise.
17708 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17710 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17711 (rs6000_reassociation_width): Likewise.
17712 (rs6000_invalid_binary_op): Likewise.
17713 (fusion_p9_p): Likewise.
17714 (emit_fusion_p9_load): Likewise.
17715 (emit_fusion_p9_store): Likewise.
17716 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17717 (ok_for_simple_move_operands): Likewise.
17718 (ok_for_simple_move_strict_operands): Likewise.
17719 (ok_for_simple_arith_logic_operands): Likewise.
17720 (visium_legitimize_reload_address): Likewise.
17721 (visium_select_cc_mode): Likewise.
17722 (output_cbranch): Likewise.
17723 (visium_split_double_move): Likewise.
17724 (visium_expand_copysign): Likewise.
17725 (visium_expand_int_cstore): Likewise.
17726 (visium_expand_fp_cstore): Likewise.
17727 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17728 (visium_function_arg): Likewise.
17729 (visium_function_arg_advance): Likewise.
17730 (visium_libcall_value): Likewise.
17731 (visium_setup_incoming_varargs): Likewise.
17732 (visium_legitimate_constant_p): Likewise.
17733 (visium_legitimate_address_p): Likewise.
17734 (visium_legitimize_address): Likewise.
17735 (visium_secondary_reload): Likewise.
17736 (visium_register_move_cost): Likewise.
17737 (visium_memory_move_cost): Likewise.
17738 (prepare_move_operands): Likewise.
17739 (ok_for_simple_move_operands): Likewise.
17740 (ok_for_simple_move_strict_operands): Likewise.
17741 (ok_for_simple_arith_logic_operands): Likewise.
17742 (visium_function_value_1): Likewise.
17743 (rtx_ok_for_offset_p): Likewise.
17744 (visium_legitimize_reload_address): Likewise.
17745 (visium_split_double_move): Likewise.
17746 (visium_expand_copysign): Likewise.
17747 (visium_expand_int_cstore): Likewise.
17748 (visium_expand_fp_cstore): Likewise.
17749 (visium_split_cstore): Likewise.
17750 (visium_select_cc_mode): Likewise.
17751 (visium_split_cbranch): Likewise.
17752 (output_cbranch): Likewise.
17753 (visium_print_operand_address): Likewise.
17754 * expmed.c (flip_storage_order): Likewise.
17755 * expmed.h (emit_cstore): Likewise.
17756 (flip_storage_order): Likewise.
17757 * genrecog.c (validate_pattern): Likewise.
17758 * hsa-gen.c (gen_hsa_addr): Likewise.
17759 * internal-fn.c (expand_arith_overflow): Likewise.
17760 * ira-color.c (allocno_copy_cost_saving): Likewise.
17761 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17762 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17763 (process_invariant_for_inheritance): Likewise.
17764 * lra-eliminations.c (move_plus_up): Likewise.
17765 * omp-low.c (lower_oacc_reductions): Likewise.
17766 * simplify-rtx.c (simplify_subreg): Likewise.
17767 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17768 (TARGET_CHKP_BOUND_MODE): Likewise..
17769 * targhooks.c (default_chkp_bound_mode): Likewise.
17770 (default_setup_incoming_vararg_bounds): Likewise.
17771 * targhooks.h (default_chkp_bound_mode): Likewise.
17772 (default_setup_incoming_vararg_bounds): Likewise.
17773 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17774 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17775 (have_whole_vector_shift): Likewise.
17776 * tree-vect-stmts.c (vectorizable_load): Likewise.
17777 * doc/tm.texi: Regenerate.
17779 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17781 Graceful degrade if Binutils PR21472 is not available.
17784 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17785 .rodata in flash test fails.
17786 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17787 * confgure: Regenerate.
17788 * config.in: Regenerate.
17789 * config/avr/avr.c (avr_asm_named_section)
17790 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17791 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17792 (avr_asm_init_sections): Same.
17794 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17796 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17797 (fma<VH:mode>4_intrinsic): Likewise.
17798 (*fmsub<VCVTF:mode>4): Likewise.
17799 (*fmsub<VH:mode>4_intrinsic): Likewise.
17801 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17804 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17805 Don't depend on "optimize > 0".
17806 (out_movhi_r_mr, out_movqi_mr_r): Same.
17807 (out_movhi_mr_r, out_movqi_r_mr): Same.
17808 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17809 io_address_operand on "optimize > 0".
17811 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17813 * tree-loop-distribution.c: Add general explanantion on the pass.
17814 (generate_loops_for_partition): Mark distributed loop.
17815 (pg_add_dependence_edges): New parameter. Handle alias data
17816 dependence specially and record it in the parameter if asked.
17817 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17818 (init_partition_graph_vertices, add_partition_graph_edge): New.
17819 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17820 (free_partition_graph_vdata, build_partition_graph): New.
17821 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17822 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17823 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17824 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17825 (version_for_distribution_p, finalize_partitions): New.
17826 (distribute_loop): Handle alias data dependence specially. Factor
17827 out loop fusion code as functions and call these functions.
17829 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17831 * tree-loop-distribution.c (classify_partition): New parameter and
17832 better handle reduction statement.
17833 (rdg_build_partitions): Revise comment.
17834 (distribute_loop): Compute statements in all partitions and pass it
17835 to classify_partition.
17837 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17839 * tree-loop-distribution.c (enum partition_type): New.
17840 (struct partition): New field type.
17841 (partition_merge_into): Add parameter. Update partition type.
17842 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17843 (build_rdg_partition_for_vertex): Compute partition type.
17844 (rdg_build_partitions): Dump partition type.
17845 (distribute_loop): Update calls to partition_merge_into.
17847 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17849 * tree-loop-distribution.c (struct ddr_hasher): New.
17850 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17852 (classify_partition): Call get_data_dependence.
17853 (pg_add_dependence_edges): Ditto.
17854 (distribute_loop): Release data dependence hash table.
17856 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17858 * tree-loop-distribution.c (ref_base_address): Delete.
17859 (similar_memory_accesses): Rename ...
17860 (share_memory_accesses): ... to this. Check if partitions access
17861 the same memory reference.
17862 (distribute_loop): Call share_memory_accesses.
17864 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17866 * tree-loop-distribution.c (struct partition): New field recording
17867 its data reference.
17868 (partition_alloc, partition_free): Init and release data refs.
17869 (partition_merge_into): Merge data refs.
17870 (build_rdg_partition_for_vertex): Collect data refs for partition.
17871 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17873 (distribute_loop): Remve data refs from vertice data of partition
17876 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17878 * tree-loop-distribution.c (params.h): Include header file.
17879 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17880 (datarefs_vec): New global var.
17881 (create_rdg_vertices): Use datarefs_vec directly.
17882 (free_rdg): Don't free data references.
17883 (build_rdg): Update use. Don't free data references.
17884 (distribute_loop): Compute global variable for data references.
17885 Bail out if there are too many data references.
17887 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17889 * tree-loop-distribution.c (loop_nest): New global var.
17890 (build_rdg): Use loop directly, rather than loop nest.
17891 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17893 (distribute_loop): Compute global variable loop nest. Update use.
17895 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17897 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17898 (partition_merge_into): New parameter. Dump reason for fusion.
17899 (distribute_loop): Update use of partition_merge_into.
17901 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17903 * tree-loop-distribution.c (bb_top_order_index): New.
17904 (bb_top_order_index_size, bb_top_order_cmp): New.
17905 (stmts_from_loop): Use topological order.
17906 (pass_loop_distribution::execute): Compute and release topological
17907 order for basic blocks.
17909 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17911 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17914 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17916 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17917 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17918 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17919 * internal-fn.def (LOOP_DIST_ALIAS): New.
17920 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17921 (fold_loop_internal_call): ... this.
17922 (vect_loop_dist_alias_call): New function.
17923 (set_uid_loop_bbs): Call fold_loop_internal_call.
17924 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17927 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17930 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17931 Require dead FLAGS_REG at the beginning of a peephole.
17933 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17936 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17937 arguments in the call to __builtin_ia32_sbb_u32.
17938 (_subborrow_u64): Swap _X and _Y arguments in the call to
17939 __builtin_ia32_sbb_u64.
17941 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17944 * tree-vrp.c (compare_assert_loc): Turn into a function template
17945 with stable template parameter. Only test if a->e is NULL,
17946 !a->e == !b->e has been verified already. Use e == NULL or
17947 e != NULL instead of e or ! e tests. If stable is true, don't use
17948 iterative_hash_expr, on the other side allow a or b or both NULL
17949 and sort the NULLs last.
17950 (process_assert_insertions): Sort using compare_assert_loc<false>
17951 instead of compare_assert_loc, later sort using
17952 compare_assert_loc<true> before calling process_assert_insertions_for
17953 in a loop. Use break instead of continue once seen NULL pointer.
17955 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17957 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17958 Cortex-R7 and Cortex-R8 processors.
17960 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17962 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17963 uninitialized while src is not.
17965 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17967 * common/config/arm/arm-common.c: Adjust include path for
17969 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17970 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17971 (arm-cpu-data.h): Likewise.
17972 (arm-cpu-cdata.h): Likewise.
17973 * config/arm/arm-cpu.h: Delete.
17974 * config/arm/arm-cpu-cdata.h: Delete.
17975 * config/arm/arm-cpu-data.h: Delete.
17977 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17979 * config/arm/arm-cpus.in (cortex-a55): New.
17980 (cortex-a75): Likewise.
17981 (cortex-a75.cortex-a55): Likewise.
17982 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17984 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17985 * config/arm/arm-cpu-cdata.h: Regenerate.
17986 * config/arm/arm-cpu-data.h: Regenerate.
17987 * config/arm/arm-cpu.h: Regenerate.
17988 * config/arm/arm-tables.opt: Regenerate.
17989 * config/arm/arm-tune.md: Regenerate.
17991 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17993 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17995 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17997 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18000 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18002 PR tree-optimization/81292
18003 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18004 full_string_p, also call adjust_related_strinfos if the adjustment
18005 is simple, otherwise invalidate related strinfos.
18007 2017-07-04 Martin Liska <mliska@suse.cz>
18010 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18011 newly created variable as DECL_IGNORED_P.
18013 2017-07-04 Martin Liska <mliska@suse.cz>
18016 * ipa-inline.c (inline_small_functions):
18017 Use xstrdup_for_dump.
18019 2017-07-04 Tom de Vries <tom@codesourcery.com>
18021 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18023 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18026 * config/darwin.c (darwin_function_switched_text_sections):
18029 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18031 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18033 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18035 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18037 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18039 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18040 min_profitable_iters, and th as inclusive lower bounds.
18041 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18042 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18043 for min_profitable_iters and min_profitable_estimate.
18044 (vect_transform_loop): Treat th as an inclusive lower bound.
18045 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18047 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18050 * config/darwin.c (darwin_function_switched_text_sections):
18051 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18052 in two pieces, and suppress the use of buf.
18054 2017-07-03 Nathan Sidwell <nathan@acm.org>
18056 * hash-table.h (hash_table_mod1): Fix indentation.
18058 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18060 PR middle-end/81290
18061 * predict.c (force_edge_cold): Be more careful about propagation
18063 * profile-count.h (profile_probability::guessed,
18064 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18066 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18068 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18070 * doc/invoke.texi (rcpc architecture extension): Document it.
18072 2017-07-03 Richard Biener <rguenther@suse.de>
18074 PR tree-optimization/60510
18075 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18076 the scalar reduction PHI and use it.
18077 (vectorizable_reduction): Properly guard the single_defuse_cycle
18078 path for non-SLP reduction chains where we cannot use it.
18079 Rework reduc_def/index and vector type deduction. Rework
18080 vector operand gathering during reduction op code-gen.
18081 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18082 chains dissolve the chain and leave it to non-SLP reduction
18085 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18087 * tree-data-ref.h (dr_alignment): Declare.
18088 * tree-data-ref.c (dr_alignment): New function.
18089 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18090 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18092 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18094 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18096 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18097 and base_misalignment fields.
18098 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18099 * tree-data-ref.c: Include builtins.h.
18100 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18101 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18102 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18103 * tree-vect-data-refs.c: Include tree-cfg.h.
18104 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18105 fields instead of calculating an alignment here.
18106 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18107 innermost_loop_behavior fields.
18109 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18111 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18113 (DR_STEP_ALIGNMENT): New macro.
18114 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18115 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18116 (create_data_ref): Print it.
18117 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18118 to tell whether the step preserves vector (mis)alignment.
18119 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18120 Move the check for an integer step and generalise to all INTEGER_CST.
18121 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18122 Print the outer step alignment.
18124 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18126 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18127 with offset_alignment.
18128 (DR_ALIGNED_TO): Delete.
18129 (DR_OFFSET_ALIGNMENT): New macro.
18130 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18131 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18132 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18133 (create_data_ref): Likewise.
18134 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18135 (vect_analyze_data_refs): Likewise.
18136 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18137 creating dummy innermost behavior.
18139 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18141 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18142 with a "innermost_loop_behavior *" and refeence tree.
18143 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18144 (create_data_ref): Update call accordingly.
18145 * tree-predcom.c (find_looparound_phi): Likewise.
18147 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18149 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18150 fields with dr_wrt_vec_loop.
18151 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18152 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18153 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18154 (vect_dr_behavior): New function.
18155 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18156 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18157 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18158 track whether the step preserves the misalignment.
18159 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18160 Use vect_dr_behavior.
18161 (vect_setup_realignment): Update call accordingly.
18162 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18163 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18164 call to vect_create_addr_base_for_vector_ref.
18165 (vect_create_cond_for_align_checks): Likewise.
18166 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18167 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18168 (vect_recog_mask_conversion_pattern): Likewise.
18169 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18170 (new_stmt_vec_info): Remove redundant zeroing.
18172 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18174 * common/config/arm/arm-common.c (arm_be8_option): New function.
18175 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18176 (ISA_ARMv6): Add isa_bit_be8.
18177 * config/arm/arm.h (arm_be8_option): Add prototype.
18178 (BE8_SPEC_FUNCTION): New define.
18179 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18180 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18181 (mlittle-endian): Similarly.
18182 (mbe8, mbe32): New options.
18183 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18184 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18186 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18188 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18190 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18192 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18193 (cleanup_tree_cfg_bb): Use it.
18194 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18196 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18198 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18201 * loop-doloop.c (add_test): Update profile.
18203 2017-07-03 Martin Liska <mliska@suse.cz>
18206 * sanopt.c (rewrite_usage_of_param): New function.
18207 (sanitize_rewrite_addressable_params): Likewise.
18208 (pass_sanopt::execute): Call rewrite_usage_of_param.
18210 2017-07-03 Richard Biener <rguenther@suse.de>
18212 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18213 back to using VIEW_CONVERT_EXPR.
18215 2017-07-03 Martin Liska <mliska@suse.cz>
18218 * doc/extend.texi: Document when a resolver function is
18219 generated for target_clones.
18221 2017-07-03 Martin Liska <mliska@suse.cz>
18223 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18224 and shadow memory of auto variables which are subject of
18225 use-after-scope sanitization.
18226 (asan_expand_mark_ifn): Add do set only when is_poison.
18228 2016-07-03 Richard Biener <rguenther@suse.de>
18230 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18232 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18233 (vectorizable_reduction): Perform reduction PHI creation when
18234 visiting a reduction PHI and adjust and simplify code generation
18235 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18236 (vect_transform_loop): Visit reduction PHIs.
18237 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18238 defs into the SLP tree.
18239 (vect_build_slp_tree): Reduction defs terminate the recursion.
18240 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18242 (vect_get_vec_defs_for_stmt_copy): Export.
18243 (vect_get_vec_defs): Likewise.
18244 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18246 (vect_get_vec_defs_for_stmt_copy): Declare.
18247 (vect_get_vec_defs): Likewise.
18249 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18251 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18252 parameter with a "loop" parameter and use it instead of the
18253 loop containing DR_STMT. Don't check simple_iv when doing
18254 BB analysis. Describe the two analysis modes in the comment.
18256 2017-07-03 Tom de Vries <tom@codesourcery.com>
18258 PR tree-optimization/69468
18259 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18260 (find_same_succ_bb): Handle ignore_edge_flags.
18262 2017-07-03 Tom de Vries <tom@codesourcery.com>
18264 PR tree-optimization/81192
18265 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18267 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18269 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18271 2017-07-03 Tom de Vries <tom@codesourcery.com>
18273 PR tree-optimization/81192
18274 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18275 BB_SAME_SUCC (bb) == NULL.
18277 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18279 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18282 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18284 * dumpfile.c: Include profile-count.h
18285 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18287 (insert_cond_bb): Update profile.
18288 * tree-cfg.h (insert_cond_bb): Update prototype.
18289 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18290 * tree-dump.c: Do not include tree-cfg.
18292 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18294 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18296 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18298 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18301 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18303 * tree-complex.c (expand_complex_div_wide): update profile.
18305 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18306 Alan Hayward <alan.hayward@arm.com>
18307 David Sherwood <david.sherwood@arm.com>
18309 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18310 (CORETYPES_H): New define.
18311 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18312 (insn-modes-inline.h, s-modes-inline-h): New rules.
18313 (generated_files): Add insn-modes-inline.h.
18314 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18315 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18316 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18317 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18318 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18319 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18320 (build/gencodes.o, build/genconditions.o): Likewise.
18321 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18322 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18323 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18324 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18325 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18326 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18327 * coretypes.h: Include everything up to real.h for generators.
18328 Include insn-modes.h first. Include wide-int-print.h after
18329 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18330 * machmode.h: Don't include insn-modes.h here.
18331 * function-tests.c: Remove includes of signop.h, machmode.h,
18332 double-int.h and wide-int.h.
18334 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18336 * optc-save-gen.awk: Likewise.
18337 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18338 * godump.c: Remove include of wide-int-print.h.
18339 * pretty-print.h: Likewise.
18340 * wide-int-print.cc: Likewise.
18341 * wide-int.cc: Likewise.
18342 * hash-map-tests.c: Remove include of signop.h.
18343 * hash-set-tests.c: Likewise.
18344 * rtl-tests.c: Likewise.
18345 * mkconfig.sh: Remove include of machmode.h.
18346 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18348 (emit_insn_modes_inline_h): ...this new function. Emit the code
18349 into an insn-modes-inline.h header file, adding appropriate
18350 include guards and end comments.
18351 (emit_insn_modes_c_header): Remove include of machmode.h.
18352 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18354 (main): Handle -i flag and call emit_insn_modes_inline_h when
18357 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18359 * tree-ssa-strlen.c (strinfo): Rename the length field to
18360 nonzero_chars. Add a full_string_p field.
18361 (compare_nonzero_chars, zero_length_string_p): New functions.
18362 (get_addr_stridx): Add an offset_out parameter.
18363 Use compare_nonzero_chars.
18364 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18365 (new_strinfo): Update after above changes to strinfo.
18366 (set_endptr_and_length): Set full_string_p.
18367 (get_string_length): Update after above changes to strinfo.
18368 (unshare_strinfo): Update call to new_strinfo.
18369 (maybe_invalidate): Likewise.
18370 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18371 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18372 as a uhwi instead of an shwi. Update after above changes to
18373 strinfo and new_strinfo.
18374 (zero_length_string): Assert that chainsi contains full strings.
18375 Use zero_length_string_p. Update call to new_strinfo.
18376 (adjust_related_strinfos): Update after above changes to strinfo.
18377 Copy full_string_p from origsi.
18378 (adjust_last_stmt): Use zero_length_string_p.
18379 (handle_builtin_strlen): Update after above changes to strinfo and
18380 new_strinfo. Install the lhs as the string length if the previous
18381 entry didn't describe a full string.
18382 (handle_builtin_strchr): Update after above changes to strinfo
18384 (handle_builtin_strcpy): Likewise.
18385 (handle_builtin_strcat): Likewise.
18386 (handle_builtin_malloc): Likewise.
18387 (handle_pointer_plus): Likewise.
18388 (handle_builtin_memcpy): Likewise. Track nonzero characters
18389 that aren't necessarily followed by a nul terminator.
18390 (handle_char_store): Likewise.
18392 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18394 PR tree-optimization/80769
18395 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18396 for malloc and calloc. Document the new invariant that all related
18397 strinfos have delayed lengths or none do.
18398 (verify_related_strinfos): Move earlier in file.
18399 (set_endptr_and_length): New function, split out from...
18400 (get_string_length): ...here. Also set the lengths of related
18402 (zero_length_string): Assert that chainsi has known (rather than
18404 (adjust_related_strinfos): Likewise.
18406 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18408 PR tree-optimization/81136
18409 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18410 assert that two references with the same misalignment have the same
18411 compile-time misalignment if those compile-time misalignments
18414 2017-07-01 Andi Kleen <ak@linux.intel.com>
18416 * print-tree.c (print_node): Print all attributes.
18418 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18420 * cfg.c (scale_bbs_frequencies): New function.
18421 * cfg.h (scale_bbs_frequencies): Declare it.
18422 * cfgloopanal.c (single_likely_exit): Cleanup.
18423 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18425 (scale_loop_profile): Likewise.
18426 (loop_version): Likewise.
18427 (create_empty_loop_on_edge): Update.
18428 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18429 scale_loop_frequencies, scale_loop_profile, loopify,
18430 loop_version): Update prototypes.
18431 * modulo-sched.c (sms_schedule): Update.
18432 * predict.c (unlikely_executed_edge_p): Also check probability.
18433 (probably_never_executed_edge_p): Fix typo.
18434 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18435 * tree-parloops.c (gen_parallel_loop): Update.
18436 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18437 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18438 * tree-ssa-loop-split.c (split_loop): Update.
18439 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18440 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18441 (vect_loop_versioning): Update.
18442 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18444 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18446 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18448 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18450 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18451 to keep profile consistent.
18453 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18455 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18456 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18457 * profile-count.h (max_safe_multiplier): Make unsigned.
18458 (profile_count::guessed_zero): New.
18460 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18462 * bb-reorder.c (fix_up_crossing_landing_pad,
18463 fix_crossing_conditional_branches): Use make_single_succ_edge
18464 to keep profile consistent.
18466 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18468 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18471 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18474 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18475 the right scopes, make sure cond_jump isn't preserved between multiple
18476 iterations. Search for fallthru edge whenever there are 3+ edges and
18477 use find_fallthru_edge for it.
18479 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18481 Patch by Alexander Monakov <amonakov@ispras.ru>
18482 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18483 probabilities consistently.
18485 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18487 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18488 emit_cmp_and_jump_insns.
18490 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18493 * tree-inline.c (expand_call_inline): Combine profile statuses.
18495 2017-06-30 Andrew Pinski <apinski@cavium.com>
18497 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18498 fold_stmt returned true.
18500 2017-06-30 Nathan Sidwell <nathan@acm.org>
18502 * ggc.h (empty_string): Delete.
18503 * cfgexpand.c (expand_asm_stmt): Use plain "".
18504 * optabs.c (expand_asm_memory_barrier): Likewise.
18505 * stringpool.c (empty_string): Delete.
18506 (digit_vector, digit_string): Delete.
18507 (ggc_alloc_string): Use plain "", don't optimize single digit
18508 strings. Use ggc_alloc_atomic.
18510 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18512 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18513 comparison set and one other set, use the cost of the non-comparison
18516 2017-06-30 Nathan Sidwell <nathan@acm.org>
18518 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18521 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18523 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18524 loops. Remove now unneeded calls to gimple_switch_set_label() that
18525 just set removed labels to NULL_TREE.
18527 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18529 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18530 (set_range_info): ...here. Only call set_range_info_raw if domain
18532 (set_nonzero_bits): Call set_range_info_raw.
18533 * tree-ssanames.h (set_range_info_raw): New.
18535 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18538 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18539 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18540 of nonimmediate_operand and <store_mask_constraint> instead of m
18541 for the input operand. For V8FI iterator, always split if input
18542 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18543 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18544 <store_mask_predicate> instead of register_operand and
18545 <store_mask_constraint> instead of v for the input operand. Make
18546 sure both operands aren't MEMs for if not <mask_applied>.
18548 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18550 * lto-wrapper.c (copy_file) Close both file descriptors before
18553 2017-06-30 Martin Liska <mliska@suse.cz>
18556 * multiple_target.c (create_dispatcher_calls): Make ifunc
18557 also for function that don't have calls or are not referenced.
18559 2017-06-30 Richard Biener <rguenther@suse.de>
18561 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18562 analyze the first scalar stmt. Move vector type computation
18563 for the BB case here from ...
18564 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18565 live operation processing in the SLP case properly.
18567 2017-06-30 Richard Biener <rguenther@suse.de>
18569 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18571 2017-06-30 Martin Liska <mliska@suse.cz>
18574 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18575 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18577 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18579 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18582 2017-06-30 Richard Biener <rguenther@suse.de>
18584 PR tree-optimization/81249
18585 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18586 condition reduction result to original scalar type.
18588 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18590 * profile-count.h (enum profile_quality): Fix typos and whitespace
18593 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18595 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18596 type for branch probabilities.
18598 2017-06-29 Julian Brown <julian@codesourcery.com>
18599 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18601 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18602 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18603 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18604 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18606 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18608 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18609 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18610 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18611 CC usage from generic code to here.
18612 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18613 CC usage into the target macros.
18615 2017-06-29 Maya Rashish <coypu@sdf.org>
18617 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18620 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18622 * arm/arm-builtins.c: Include profile-count.h
18623 * except.c (sjlj_emit_function_enter): Use
18624 profile_probability::unlikely.
18626 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18628 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18629 and tocrel_offset be pointer args rather than implicitly using
18631 (legitimate_constant_pool_address_p, rs6000_emit_move,
18632 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18633 tocrel_offset and use in toc_relative_expr_p call.
18634 (print_operand, print_operand_address): Use static tocrel_base_oac
18635 and tocrel_offset_oac.
18636 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18639 2017-06-29 Maya Rashish <coypu@sdf.org>
18641 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18643 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18645 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18646 objects, take into account only the alignment of 'op0' and 'mode1' if
18649 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18651 * ccmp.c (ccmp_tree_comparison_p): New function.
18652 (ccmp_candidate_p): Update to use above function.
18653 (get_compare_parts): New function.
18654 (expand_ccmp_next): Update to use new functions.
18655 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18657 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18658 take mode as argument.
18659 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18660 * expr.c (expand_expr_real_1): Pass mode as argument.
18662 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18664 * combine.c (combine_instructions): Print insns to dump_file, together
18667 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18669 * asan.c (asan_emit_stack_protection): Update.
18670 (create_cond_insert_point): Update.
18671 * auto-profile.c (afdo_propagate_circuit): Update.
18672 * basic-block.h (struct edge_def): Turn probability to
18673 profile_probability.
18674 (EDGE_FREQUENCY): Update.
18675 * bb-reorder.c (find_traces_1_round): Update.
18676 (better_edge_p): Update.
18677 (sanitize_hot_paths): Update.
18678 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18679 (make_single_succ_edge): Update.
18680 (check_bb_profile): Update.
18681 (dump_edge_info): Update.
18682 (update_bb_profile_for_threading): Update.
18683 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18684 probabilitycount to 0.
18685 * cfgbuild.c (compute_outgoing_frequencies): Update.
18686 * cfgcleanup.c (try_forward_edges): Update.
18687 (outgoing_edges_match): Update.
18688 (try_crossjump_to_edge): Update.
18689 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18690 (expand_gimple_tailcall): Update.
18691 (construct_init_block): Use make_single_succ_edge.
18692 (construct_exit_block): Use make_single_succ_edge.
18693 * cfghooks.c (verify_flow_info): Update.
18694 (redirect_edge_succ_nodup): Update.
18695 (split_edge): Update.
18696 (account_profile_record): Update.
18697 * cfgloopanal.c (single_likely_exit): Update.
18698 * cfgloopmanip.c (scale_loop_profile): Update.
18699 (set_zero_probability): Remove.
18700 (duplicate_loop_to_header_edge): Update.
18701 * cfgloopmanip.h (loop_version): Update prototype.
18702 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18703 (force_nonfallthru_and_redirect): Update.
18704 (update_br_prob_note): Update.
18705 (rtl_verify_edges): Update.
18706 (purge_dead_edges): Update.
18707 (rtl_lv_add_condition_to_bb): Update.
18708 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18709 * cgraphunit.c (init_lowered_empty_function): Update.
18710 (cgraph_node::expand_thunk): Update.
18711 * cilk-common.c: Include profile-count.h
18712 * dojump.c (inv): Remove.
18713 (jumpifnot): Update.
18714 (jumpifnot_1): Update.
18715 (do_jump_1): Update.
18717 (do_jump_by_parts_greater_rtx): Update.
18718 (do_compare_rtx_and_jump): Update.
18719 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18720 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18721 * dwarf2cfi.c: Include profile-count.h
18722 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18723 (sjlj_emit_dispatch_table): Likewise.
18724 * explow.c: Include profile-count.h
18725 * expmed.c (emit_store_flag_force): Update.
18726 (do_cmp_and_jump): Update.
18727 * expr.c (compare_by_pieces_d::generate): Update.
18728 (compare_by_pieces_d::finish_mode): Update.
18729 (emit_block_move_via_loop): Update.
18730 (store_expr_with_bounds): Update.
18731 (store_constructor): Update.
18732 (expand_expr_real_2): Update.
18733 (expand_expr_real_1): Update.
18734 * expr.h (try_casesi, try_tablejump): Update prototypes.
18735 * gimple-pretty-print.c (dump_probability): Update.
18736 (dump_profile): New.
18737 (dump_gimple_label): Update.
18738 (dump_gimple_bb_header): Update.
18739 * graph.c (draw_cfg_node_succ_edges): Update.
18740 * hsa-gen.c (convert_switch_statements): Update.
18741 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18742 (find_if_case_1): Update.
18743 (find_if_case_2): Update.
18744 * internal-fn.c (expand_arith_overflow_result_store): Update.
18745 (expand_addsub_overflow): Update.
18746 (expand_neg_overflow): Update.
18747 (expand_mul_overflow): Update.
18748 (expand_vector_ubsan_overflow): Update.
18749 * ipa-cp.c (good_cloning_opportunity_p): Update.
18750 * ipa-split.c (split_function): Use make_single_succ_edge.
18751 * ipa-utils.c (ipa_merge_profiles): Update.
18752 * loop-doloop.c (add_test): Update.
18753 (doloop_modify): Update.
18754 * loop-unroll.c (compare_and_jump_seq): Update.
18755 (unroll_loop_runtime_iterations): Update.
18756 * lra-constraints.c (lra_inheritance): Update.
18757 * lto-streamer-in.c (input_cfg): Update.
18758 * lto-streamer-out.c (output_cfg): Update.
18759 * mcf.c (adjust_cfg_counts): Update.
18760 * modulo-sched.c (sms_schedule): Update.
18761 * omp-expand.c (expand_omp_for_init_counts): Update.
18762 (extract_omp_for_update_vars): Update.
18763 (expand_omp_ordered_sink): Update.
18764 (expand_omp_for_ordered_loops): Update.
18765 (expand_omp_for_generic): Update.
18766 (expand_omp_for_static_nochunk): Update.
18767 (expand_omp_for_static_chunk): Update.
18768 (expand_cilk_for): Update.
18769 (expand_omp_simd): Update.
18770 (expand_omp_taskloop_for_outer): Update.
18771 (expand_omp_taskloop_for_inner): Update.
18772 * omp-simd-clone.c (simd_clone_adjust): Update.
18773 * optabs.c (expand_doubleword_shift): Update.
18774 (expand_abs): Update.
18775 (emit_cmp_and_jump_insn_1): Update.
18776 (expand_compare_and_swap_loop): Update.
18777 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18778 * predict.c (predictable_edge_p): Update.
18779 (edge_probability_reliable_p): Update.
18780 (set_even_probabilities): Update.
18781 (combine_predictions_for_insn): Update.
18782 (combine_predictions_for_bb): Update.
18783 (propagate_freq): Update.
18784 (estimate_bb_frequencies): Update.
18785 (force_edge_cold): Update.
18786 * profile-count.c (profile_count::dump): Add missing space into dump.
18787 (profile_count::debug): Add newline.
18788 (profile_count::differs_from_p): Explicitly convert to unsigned.
18789 (profile_count::stream_in): Update.
18790 (profile_probability::dump): New member function.
18791 (profile_probability::debug): New member function.
18792 (profile_probability::differs_from_p): New member function.
18793 (profile_probability::differs_lot_from_p): New member function.
18794 (profile_probability::stream_in): New member function.
18795 (profile_probability::stream_out): New member function.
18796 * profile-count.h (profile_count_quality): Rename to ...
18797 (profile_quality): ... this one.
18798 (profile_probability): New.
18799 (profile_count): Update.
18800 * profile.c (compute_branch_probabilities): Update.
18801 * recog.c (peep2_attempt): Update.
18802 * sched-ebb.c (schedule_ebbs): Update.
18803 * sched-rgn.c (find_single_block_region): Update.
18804 (compute_dom_prob_ps): Update.
18805 (schedule_region): Update.
18806 * sel-sched-ir.c (compute_succs_info): Update.
18807 * stmt.c (struct case_node): Update.
18808 (do_jump_if_equal): Update.
18809 (get_outgoing_edge_probs): Update.
18810 (conditional_probability): Update.
18811 (emit_case_dispatch_table): Update.
18812 (expand_case): Update.
18813 (expand_sjlj_dispatch_table): Update.
18814 (emit_case_nodes): Update.
18815 * targhooks.c: Update.
18816 * tracer.c (better_p): Update.
18817 (find_best_successor): Update.
18818 * trans-mem.c (expand_transaction): Update.
18819 * tree-call-cdce.c: Update.
18820 * tree-cfg.c (gimple_split_edge): Upate.
18821 (move_sese_region_to_fn): Upate.
18822 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18823 * tree-eh.c (lower_resx): Upate.
18824 (cleanup_empty_eh_move_lp): Upate.
18825 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18826 * tree-inline.c (copy_edges_for_bb): Update.
18827 (copy_cfg_body): Update.
18828 * tree-parloops.c (gen_parallel_loop): Update.
18829 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18830 (gimple_gen_time_profiler): Update.
18831 * tree-ssa-dce.c (remove_dead_stmt): Update.
18832 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18833 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18834 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18835 (unloop_loops): Update.
18836 (try_peel_loop): Update.
18837 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18838 * tree-ssa-loop-split.c (connect_loops): Update.
18839 (split_loop): Update.
18840 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18841 (hoist_guard): Update.
18842 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18843 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18844 (value_replacement): Update.
18845 * tree-ssa-reassoc.c (branch_fixup): Update.
18846 * tree-ssa-tail-merge.c (replace_block_by): Update.
18847 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18848 (create_edge_and_update_destination_phis): Update.
18849 (compute_path_counts): Update.
18850 (recompute_probabilities): Update.
18851 (update_joiner_offpath_counts): Update.
18852 (freqs_to_counts_path): Update.
18853 (duplicate_thread_path): Update.
18854 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18855 (struct switch_conv_info): Update.
18856 (gen_inbound_check): Update.
18857 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18858 (vect_do_peeling): Update.
18859 (vect_loop_versioning): Update.
18860 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18861 (optimize_mask_stores): Update.
18862 * ubsan.c (ubsan_expand_null_ifn): Update.
18863 * value-prof.c (gimple_divmod_fixed_value): Update.
18864 (gimple_divmod_fixed_value_transform): Update.
18865 (gimple_mod_pow2): Update.
18866 (gimple_mod_pow2_value_transform): Update.
18867 (gimple_mod_subtract): Update.
18868 (gimple_mod_subtract_transform): Update.
18869 (gimple_ic): Update.
18870 (gimple_stringop_fixed_value): Update.
18871 (gimple_stringops_transform): Update.
18872 * value-prof.h: Update.
18874 2017-06-29 Carl Love <cel@us.ibm.com>
18876 * config/rs6000/rs6000-c.c: Add support for built-in functions
18877 vector signed int vec_signed (vector float);
18878 vector signed long long vec_signed (vector double);
18879 vector signed int vec_signed2 (vector double, vector double);
18880 vector signed int vec_signede (vector double);
18881 vector signed int vec_signedo (vector double);
18882 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18883 instruction generator.
18884 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18885 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18886 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18888 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18889 vunsignede_v2df): Add define_expands.
18890 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18891 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18892 VEC_UNSIGNEDO): Add definitions.
18893 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18894 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18895 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18896 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18897 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18898 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18899 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18900 vec_unsignede, vec_unsignedo): Add builtin defines.
18901 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18903 * doc/extend.texi: Update the built-in documentation file for the
18904 new built-in functions.
18906 2017-06-29 Richard Biener <rguenther@suse.de>
18908 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18909 reduction chains to LOOP_VINFO_REDUCTIONS.
18910 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18911 SLP reductions after processing reduction chains.
18913 2017-06-29 Nathan Sidwell <nathan@acm.org>
18915 * builtins.c (fold_builtin_FUNCTION): Use
18916 lang_hooks.decl_printable_name.
18918 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18920 PR middle-end/81194
18921 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18922 with only one label.
18923 * stmt.c (expand_case): Assert NCASES is greater than one.
18925 2017-06-29 Richard Biener <rguenther@suse.de>
18927 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18929 (group_case_labels): Likewise.
18930 (find_taken_edge): Push sanity checking on val to workers...
18931 (find_taken_edge_cond_expr): ... here
18932 (find_taken_edge_switch_expr): ... and here, handle cases
18933 with just a default label.
18934 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18935 (group_case_labels): Likewise.
18936 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18937 group_case_labels does anything cleanup the CFG again.
18939 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18941 PR tree-optimization/81196
18942 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18943 exit condition comparing two IVs.
18945 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18947 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18948 profile to the dummy entry at the end of the list of architectures.
18949 * config/arm/arm-cpu-cdata.h: Regenerated.
18951 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18952 Michael Collison <michael.collison@arm.com>
18955 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18957 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18958 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18959 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18960 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18961 of shift when the shift amount is masked with constant equal to
18962 the size of the mode.
18963 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18966 2017-06-29 Martin Liska <mliska@suse.cz>
18968 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18970 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18973 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18976 (edge_growth_cache_entry::edge_growth_cache_entry): New
18978 (reset_edge_growth_cache): Update to use constructor.
18980 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18982 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18983 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18984 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18986 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18988 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18989 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18991 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18993 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18994 (*-linux-uclibc*): Add t-uclibc tmake_file.
18995 * config/t-musl: New.
18996 * config/t-uclibc: New.
18998 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19000 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19002 (gen_comm_data): Emit architectural setting of arch_prof.
19003 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19005 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19006 (armv8-m.base, armv8-m.main): Likewise.
19007 * arm-protos.h (arm_build_target): Add profile field.
19008 (arch_option): Likewise.
19009 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19011 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19012 arm_active_target.profile.
19014 2017-06-28 Richard Biener <rguenther@suse.de>
19016 PR middle-end/81227
19017 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19018 TYPE_OVERFLOW_WRAPS.
19019 * match.pd (negate_expr_p): Likewise.
19020 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19021 fold_build2, not fold_binary.
19023 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19025 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19026 Convert memory address to Pmode.
19027 (aarch64_print_operand): Assert MEM operands are always Pmode.
19029 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19032 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19033 Remove redundant if.
19034 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19035 * config/arm/aarch-common-protos.h
19036 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19037 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19039 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19042 * multiple_target.c (create_dispatcher_calls): Set the default
19043 clone to be static, not public.
19045 2017-06-28 Richard Biener <rguenther@suse.de>
19047 * tree-vect-loop.c (vectorizable_reduction): Move special
19048 cond reduction IV var creation ...
19049 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19050 parameter. Use STMT_VINFO_VECTYPE.
19051 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19054 2017-06-28 Martin Liska <mliska@suse.cz>
19057 * ipa-visibility.c (non_local_p): Handle visibility.
19059 2017-06-28 Martin Liska <mliska@suse.cz>
19062 * common.opt: Add IntegerRange to various options.
19063 * opt-functions.awk (integer_range_info): New function.
19064 * optc-gen.awk: Add integer_range_info to cl_options struct.
19065 * opts-common.c (decode_cmdline_option): Handle
19066 CL_ERR_INT_RANGE_ARG.
19067 (cmdline_handle_error): Likewise.
19068 * opts.c (print_filtered_help): Show valid interval in
19069 when --help is provided.
19070 * opts.h (struct cl_option): Add range_min and range_max fields.
19071 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19073 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19075 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19076 (x * C EQ/NE y * C): New transformation.
19078 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19080 * genmultilib (combination_space): Accept '+' in option names.
19082 2017-06-28 Martin Liska <mliska@suse.cz>
19085 * asan.c (instrument_derefs): Bail out inner references
19086 that are hard register variables.
19088 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19091 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19092 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19094 2017-06-28 Richard Biener <rguenther@suse.de>
19096 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19097 (vect_get_slp_defs): Adjust.
19098 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19100 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19102 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19103 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19104 (vectorizable_reduction): Adjust.
19105 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19107 (vect_get_slp_defs): Likewise.
19108 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19109 (vectorizable_bswap): Adjust.
19110 (vectorizable_call): Likewise.
19111 (vectorizable_conversion): Likewise.
19112 (vectorizable_assignment): Likewise.
19113 (vectorizable_shift): Likewise.
19114 (vectorizable_operation): Likewise.
19115 (vectorizable_store): Likewise.
19116 (vectorizable_condition): Likewise.
19117 (vectorizable_comparison): Likewise.
19119 2017-06-28 Michael Collison <michael.collison@arm.com>
19122 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19124 (arm_gen_movmemqi): Removed unused variable 'i'.
19125 Convert 'for' loop into 'while' loop.
19126 (arm_expand_prologue): Remove last unnecessary set of insn.
19127 (thumb_pop): Remove unused variable 'pushed_words'.
19128 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19130 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19132 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19133 * config/s390/s390-protos.h: Add prototype of
19134 s390_rel_address_ok_p.
19135 * config/s390/s390.c (s390_got_symbol): New function.
19136 (s390_rel_address_ok_p): New function.
19137 (legitimize_pic_address): Use s390_rel_address_ok_p.
19138 (s390_load_got): Use s390_got_symbol.
19139 (s390_option_override): Issue error if
19140 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19141 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19143 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19145 2017-06-27 Andrew Pinski <apinski@cavium.com>
19147 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19148 (X * copysign (1.0, X)): New pattern.
19149 (X * copysign (1.0, -X)): New pattern.
19150 (copysign (-1.0, CST)): New pattern.
19152 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19154 * genmultilib (combination_space): Remove variable.
19155 Validate reuse rules against regular expression for any sequence
19156 of multilib options in any order.
19158 2017-06-27 Michael Collison <michael.collison@arm.com>
19160 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19161 call aarch64_split_simd_combine.
19162 * (aarch64_combine_internal<mode>): Delete pattern.
19163 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19164 Allow register and subreg operands.
19166 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19168 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19169 specific need, just fallback on defaults.
19170 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19172 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19173 Olivier Hainque <hainque@adacore.com>
19175 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19177 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19178 targets. Pick a default if no particular attempt applied.
19179 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19182 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19184 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19185 (x86_64-wrs-vxworks7): Likewise.
19187 2017-06-27 Marek Polacek <polacek@redhat.com>
19190 * ubsan.c (instrument_null): Check get_base_address's result for null.
19192 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19194 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19196 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19198 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19199 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19200 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19201 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19202 New function types.
19203 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19204 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19205 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19206 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19207 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19208 BUILT_IN_FEUPDATEENV): New builtins.
19209 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19210 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19211 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19212 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19214 (builtin_structptr_types): Adjust size.
19215 * tree.c (builtin_structptr_types): Add four entries.
19217 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19218 Olivier Hainque <hainque@adacore.com>
19220 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19221 (TLS_SYM): New local macro, forcing reference to __tls__ on
19222 link command lines for VxWorks 7 RTPs, triggering initialization
19224 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19225 OS features TLS support, true for RTPs on VxWorks 7.
19226 * config/vxworks.c (vxworks_override_options): Setup emutls
19229 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19231 * predict.c (test_prediction_value_range): Use -1U instead of -1
19232 to avoid narrowing conversion warning.
19233 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19234 to avoid narrowing conversion warning.
19235 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19237 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19239 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19241 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19242 64bit configurations.
19243 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19244 (SIZE_TYPE): Likewise.
19245 * config/vxworks.c (vxworks_emutls_var_fields): Use
19246 long_unsigned_type_node instead of unsigned_type_node as the offset
19247 field type, which is "pointer" mode in emutls.c.
19249 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19252 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19254 PR middle-end/81207
19255 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19256 gimple_vuse copying separately from gimple_vdef copying.
19258 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19260 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19263 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19264 Olivier Hainque <hainque@adacore.com>
19266 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19267 for all vxworks7 targets.
19268 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19269 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19270 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19271 variations for VX6/VX7 and 32/64bits later on in ...
19272 (VXWORKS_LIB_SPEC): Leverage new macros.
19273 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19274 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19276 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19278 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19279 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19281 2017-06-26 Carl Love <cel@us.ibm.com>
19283 * config/rs6000/rs6000-c.c: Add support for built-in functions
19284 vector bool char vec_reve (vector bool char);
19285 vector signed char vec_reve (vector signed char);
19286 vector unsigned char vec_reve (vector unsigned char);
19287 vector bool int vec_reve (vector bool int);
19288 vector signed int vec_reve (vector signed int);
19289 vector unsigned int vec_reve (vector unsigned int);
19290 vector bool long long vec_reve (vector bool long long);
19291 vector signed long long vec_reve (vector signed long long);
19292 vector unsigned long long vec_reve (vector unsigned long long);
19293 vector bool short vec_reve (vector bool short);
19294 vector signed short vec_reve (vector signed short);
19295 vector double vec_reve (vector double);
19296 vector float vec_reve (vector float);
19297 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19298 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19299 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19300 (altivec_vreve): New pattern.
19301 * config/rs6000/altivec.h (vec_reve): New define.
19302 * doc/extend.texi (vec_rev): Update the built-in documentation file
19303 for the new built-in functions.
19305 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19307 PR tree-optimization/71815
19308 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19310 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19312 (slsr_process_phi): Likewise.
19313 (replace_uncond_cands_and_profitable_phis): Don't replace a
19314 multiply candidate with a stride of 1 (copy or cast).
19315 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19317 (lowest_cost_path): Likewise.
19318 (total_savings): Likewise.
19320 2017-06-26 Richard Biener <rguenther@suse.de>
19323 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19324 Use def_builtin_pure for all gather builtins.
19326 2017-06-26 Richard Biener <rguenther@suse.de>
19328 PR tree-optimization/81203
19329 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19330 non-dominating BBs.
19332 2017-06-26 Marek Polacek <polacek@redhat.com>
19335 * doc/invoke.texi: Document -Wmultistatement-macros.
19337 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19339 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19340 arm_neon_ok_no_float_abi effective target.
19342 2017-06-26 Richard Biener <rguenther@suse.de>
19344 PR tree-optimization/80928
19345 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19346 (copy_bbs): Set BB_DUPLICATED flag early.
19347 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19349 (execute_on_shrinking_pred): Likewise.
19350 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19351 BB_DUPLICATED blocks.
19352 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19353 iterate over all PHIs considering removal of *gsi.
19355 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19357 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19360 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19362 * config/rs6000/rs6000-string.c: (expand_block_clear,
19363 do_load_for_compare, select_block_compare_mode,
19364 compute_current_alignment, expand_block_compare,
19365 expand_strncmp_align_check, expand_strn_compare,
19366 expand_block_move, rs6000_output_load_multiple)
19367 Move functions related to string/block move/compare
19368 to a separate file.
19369 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19370 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19371 for this function which is now used in two files.
19372 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19373 * config.gcc: Add rs6000-string.o to extra_objs for
19374 targets powerpc*-*-* and rs6000*-*-*.
19376 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19379 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19380 32-bit, since indexed is not valid for DImode.
19381 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19382 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19383 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19384 (define_peephole2 for Altivec d-form store): Likewise.
19387 * multiple_target.c (create_dispatcher_calls): Only create the
19388 dispatcher call if the function is the default clone of a
19389 versioned function.
19391 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19393 PR middle-end/80902
19394 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19395 a call, force the call to not be a tail call.
19397 2017-06-23 Jeff Law <law@redhat.com>
19399 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19402 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19404 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19405 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19406 handling for early expansion of vector shifts (sl,sr,sra,rl).
19407 (builtin_function_type): Add vector shift right instructions
19408 to the unsigned argument list.
19410 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19412 rtl-optimizatoin/79286
19413 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19414 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19415 trap. PIC register plus a const unspec without offset can never trap.
19417 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19419 * tree.h (builtin_structptr_type): New type.
19420 (builtin_structptr_types): Declare new array.
19421 * tree.c (builtin_structptr_types): New array.
19422 (free_lang_data, build_common_tree_nodes): Use it.
19424 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19427 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19430 2017-06-22 Matt Turner <mattst88@gmail.com>
19432 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19433 Lake models to skylake case. Assume skylake for unknown
19434 models with clflushopt.
19436 2017-06-22 Jeff Law <law@redhat.com>
19438 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19439 frame sizes that do not satisfy aarch64_uimm12_shift.
19441 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19443 * profile-count.h (apply_probability,
19444 apply_scale, probability_in): Fix checks for zero.
19446 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19448 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19449 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19451 2016-06-22 Richard Biener <rguenther@suse.de>
19453 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19454 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19455 REDUC_MAX_EXPR support.
19456 (vectorizable_reduction): Likewise.
19457 (vect_create_epilog_for_reduction): Likewise.
19459 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19461 * match.pd (A / (1 << B) -> A >> B): New.
19462 * generic-match-head.c: Include optabs-tree.h.
19463 * gimple-match-head.c: Likewise.
19464 * optabs-tree.h (target_supports_op_p): New.
19465 * optabs-tree.c (target_supports_op_p): New.
19467 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19469 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19470 $gcc_cv_ld --help output.
19471 (gcc_cv_ld_demangle): Likewise.
19472 (gcc_cv_ld_eh_frame_hdr): Likewise.
19473 (gcc_cv_ld_pie): Likewise.
19474 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19475 (gcc_cv_ld_buildid): Likewise.
19476 (gcc_cv_ld_sysroot): Likewise.
19477 (ld_bndplt_support): Likewise.
19478 (ld_pushpopstate_support): Likewise.
19479 * configure: Regenerate.
19480 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19482 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19485 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19486 operands indexes to avoid gap between operands and match_dups.
19488 2017-06-21 Andrew Pinski <apinski@cavium.com>
19490 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19491 Increment Arith_shift and Arith_shift_reg by 1.
19492 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19494 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19495 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19496 (aarch64_strip_extend): Add new argument and test for it.
19497 (aarch64_cheap_mult_shift_p): New function.
19498 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19499 add a cost if it is true.
19500 Update calls to aarch64_strip_extend.
19501 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19503 2017-06-21 Andrew Pinski <apinski@cavium.com>
19505 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19507 (thunderxt88): Likewise.
19508 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19509 (thunderx_prefetch_tune): New variable.
19510 (thunderx2t99_prefetch_tune): Update for the correct values.
19511 (thunderxt88_tunings): New variable.
19512 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19513 generic_prefetch_tune.
19514 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19516 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19518 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19519 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19520 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19521 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19522 (aarch64_atomic_cas<mode>, GPI): Likewise.
19524 2017-06-21 Martin Liska <mliska@suse.cz>
19526 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19527 statements on cold and hot labels.
19528 * predict.c (tree_estimate_probability_bb): Remove the
19529 prediction from this place.
19531 2017-06-21 Martin Liska <mliska@suse.cz>
19533 PR tree-optimization/79489
19534 * gimplify.c (maybe_add_early_return_predict_stmt): New
19536 (gimplify_return_expr): Call the function.
19537 * predict.c (tree_estimate_probability_bb): Remove handling
19539 * predict.def: Update comment about early return predictor.
19540 * gimple-predict.h (is_gimple_predict): New function.
19541 * predict.def: Change default value of early return to 66.
19542 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19544 * passes.def: Put pass_strip_predict_hints to the beginning of
19547 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19549 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19550 FUNCTION_DECL declarations.
19551 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19553 (dwaf2out_decl): Likewise.
19554 * godump.c (go_early_global_decl): Skip call to the real debug hook
19555 for FUNCTION_DECL declarations.
19556 * passes.c (rest_of_decl_compilation): Skip call to the
19557 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19558 -fdump-go-spec is passed.
19560 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19562 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19563 Reorder for compactness.
19564 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19565 (def_builtin_pure, def_builtin_pure2): New functions.
19566 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19568 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19570 * match.pd (nop_convert): New predicate.
19571 ((A +- CST1) +- CST2): Allow some NOP conversions.
19573 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19576 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19577 with ctors/dtors if GOVD_SHARED is set.
19579 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19581 * config/aarch64/aarch64.md (movti_aarch64):
19582 Emit mov rather than orr.
19583 (movtf_aarch64): Likewise.
19584 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19585 Emit mov rather than orr.
19587 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19589 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19590 Swap alternatives, make integer dup more expensive.
19592 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19594 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19595 Return true for non-tls symbols.
19597 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19599 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19600 (cortex-a75): Likewise.
19601 (cortex-a75.cortex-a55): Likewise.
19602 * config/aarch64/aarch64-tune.md: Regenerate.
19603 * doc/invoke.texi (-mtune): Document new values for -mtune.
19605 2017-06-21 Tom de Vries <tom@codesourcery.com>
19607 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19608 stack_size feature.
19609 (Effective-Target Keywords, Other attributes): Suggest using
19610 dg-add-options stack_size feature to get stack limit in stack_size
19611 effective target documentation.
19613 2017-06-21 Julian Brown <julian@codesourcery.com>
19614 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19616 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19617 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19618 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19620 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19621 attribute type list for neon_multiply.
19622 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19623 attribute type list for neon_multiply.
19624 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19625 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19626 attribute type list for neon_multiply.
19627 * config/arm/types.md (crypto_pmull): Add.
19628 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19629 attribute type list.
19631 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19633 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19636 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19638 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19639 to make sure not to dereference a NULL cost_classes_ptr pointer.
19641 2017-06-20 Carl Love <cel@us.ibm.com>
19643 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19644 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19645 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19646 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19647 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19648 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19649 VMULOSW): New enum "unspec" values.
19650 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19651 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19652 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19653 altivec_vmulosw): New patterns.
19654 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19655 VMULOSW): Add definitions.
19657 2017-06-20 Julia Koval <julia.koval@intel.com>
19659 * config/i386/i386.c: Fix rounding expand for new pattern.
19660 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19662 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19664 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19665 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19667 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19669 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19672 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19674 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19675 architecture, then by implementer ID.
19676 * config/aarch64/aarch64-tune.md: Regenerate.
19678 2017-06-20 Richard Biener <rguenther@suse.de>
19680 PR middle-end/81097
19681 * fold-const.c (split_tree): Fold to type before negating.
19683 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19685 * diagnostic-show-locus.c
19686 (selftest::test_fixit_deletion_affecting_newline): New function.
19687 (selftest::diagnostic_show_locus_c_tests): Call it.
19689 2017-06-20 Andreas Schwab <schwab@suse.de>
19692 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19695 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19697 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19698 __ARM_FEATURE_COPROC according to support.
19700 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19702 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19703 Rewritten to avoid overflow for > 32-bit pointers.
19706 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19707 by removing enum keyword.
19708 (ubsan_type_descriptor): Likewise. Formatting fix.
19711 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19712 splitter): Require TARGET_SSE2 in the condition.
19714 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19717 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19718 for doing vector set of SFmode on ISA 3.0.
19719 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19720 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19722 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19723 SFmode value into a V4SF variable that was extracted from another
19724 V4SF variable without converting the element to double precision
19725 and back to single precision vector format.
19726 (vsx_insert_extract_v4sf_p9_2): Likewise.
19728 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19730 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19731 in UWHI to avoid undefined overflow.
19734 * ubsan.h (enum ubsan_encode_value_phase): New.
19735 (ubsan_encode_value): Change second argument to
19736 enum ubsan_encode_value_phase with default value of
19737 UBSAN_ENCODE_VALUE_GENERIC.
19738 * ubsan.c (ubsan_encode_value): Change second argument to
19739 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19740 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19741 create_tmp_var_raw instead of create_tmp_var and use a
19743 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19744 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19745 ubsan_encode_value callers.
19748 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19749 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19750 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19752 2017-06-19 Richard Biener <rguenther@suse.de>
19754 PR middle-end/81118
19755 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19756 estimates if we changed anything.
19758 2017-06-19 Richard Biener <rguenther@suse.de>
19760 PR tree-optimization/80887
19761 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19762 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19763 simplified lookups, then reset mprts_hook.
19764 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19766 (try_to_simplify): Likewise.
19768 2017-06-19 Martin Liska <mliska@suse.cz>
19771 * gimplify.c (gimplify_switch_expr):
19772 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19774 2017-06-19 Martin Liska <mliska@suse.cz>
19776 * doc/install.texi: Document that PGO runs in 4 stages.
19778 2017-06-19 Martin Liska <mliska@suse.cz>
19781 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19782 to dispatcher function name.
19783 * multiple_target.c (replace_function_decl): New function.
19784 (create_dispatcher_calls): Redirect both edges and references.
19786 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19788 * profile-count.c (profile_count::dump): Dump quality.
19789 (profile_count::differs_from_p): Update for unsigned val.
19790 * profile-count.h (profile_count_quality): New enum.
19791 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19793 2017-06-19 Richard Biener <rguenther@suse.de>
19795 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19796 struct function as arg.
19797 (estimate_numbers_of_iterations): Export overload with loop arg.
19798 (free_numbers_of_iterations_estimates_loop): Use an overload of
19799 free_numbers_of_iterations_estimates instead.
19800 * tree-cfg.c (remove_bb): Adjust.
19801 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19802 * tree-parloops.c (gen_parallel_loop): Likewise.
19803 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19805 (tree_unroll_loops_completely): Likewise.
19806 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19807 Use an overload instead and export.
19808 (estimated_loop_iterations): Adjust.
19809 (max_loop_iterations): Likewise.
19810 (likely_max_loop_iterations): Likewise.
19811 (estimate_numbers_of_iterations): Take struct function as arg
19813 (loop_exits_before_overflow): Adjust.
19814 (free_numbers_of_iterations_estimates_loop): Use an overload.
19815 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19816 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19818 2017-06-19 Richard Biener <rguenther@suse.de>
19821 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19822 RANGE_EXPR conservatively.
19824 2017-06-16 Carl Love <cel@us.ibm.com>
19826 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19827 definitions for vec_float, vec_float2, vec_floato,
19828 vec_floate built-ins.
19829 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19830 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19832 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19833 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19834 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19835 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19836 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19837 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19838 vec_floato): Add builtin defines.
19839 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19840 Update the built-in documentation file for the new built-in
19843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19845 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19846 (mthumb): Mark as the negative of -marm.
19848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19850 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19852 (ARM Options, -mtune): Document that this accepts the same
19853 extension options as -mcpu.
19854 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19856 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19858 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19859 permitted extensions.
19861 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19863 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19864 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19865 (armv8-m.main): Add option +nodsp.
19866 * config/arm/arm-cpu-cdata.h: Regenerated.
19868 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19870 * config/arm/t-fuchsia: New file.
19871 * config.gcc (arm*-*-fuchsia*): Use it.
19873 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19875 * config/arm/t-symbian: Rewrite for new option infrastructure.
19877 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19879 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19880 (MULTILIB_REQUIRED): Likewise.
19882 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19884 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19885 (MULTILIB_RESUE): Likewise.
19886 (MULTILIB_MATCHES): Likewise.
19887 (MULTLIB_REQUIRED): Likewise.
19889 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19891 * config/arm/t-rtems: Rewrite for new option framework.
19893 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19895 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19896 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19897 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19898 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19899 * config/arm/t-multilib: ... here.
19900 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19901 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19902 armv7-a and armv8*-a when A-profile libraries have not been built.
19903 * config/arm/t-rmprofile: Rewrite.
19905 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19907 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19908 with a backslash. Remove the backslash after substituting unescaped
19910 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19912 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19914 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19915 the driver to pass through the default setting of -mfloat-abi.
19916 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19918 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19919 (all_feat_combs): New rule.
19920 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19922 * config/arm/t-aprofile: Rewrite.
19924 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19926 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19927 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19928 fpu is not specified by the user/command-line.
19929 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19930 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19931 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19932 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19933 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19934 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19936 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19938 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19939 * config/arm/t-arm-elf: Rewritten.
19941 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19943 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19944 have some floating-point instructions.
19945 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19946 (TARGET_MAYBE_HARD_FLOAT): New macro.
19947 * config/arm/arm-builtins.c (arm_init_builtins): Use
19948 TARGET_MAYBE_HARD_FLOAT.
19949 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19951 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19953 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19954 (configargs.h): Include it.
19955 (arm_print_hint_for_fpu_option): New function.
19956 (arm_parse_fpu_option): New function.
19957 (candidate_extension): New class.
19958 (arm_canon_for_multilib): New function.
19959 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19960 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19961 (ARCH_CANONICAL_SPECS): New macro.
19962 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19964 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19966 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19967 are set after handling multilib fragments. Set target_cpu_default2
19970 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19972 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19974 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19975 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19977 (check_arch): Likewise.
19978 * config/arm/arm.c (arm_configure_build_target): Handle
19979 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19980 options in the default.
19982 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19984 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19985 when an option is an alias of another.
19986 * config/arm/parsecpu.awk (optalias): New parser token.
19987 (gen_comm_data): Mark non-alias options as such. Emit entries
19988 for extension aliases.
19989 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19990 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19991 (armv6kz, armv6zk, armv6t2): Likewise.
19992 (armv7): Make vfpv3-d16 an alias.
19993 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19995 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19996 Sort in canonical order.
19997 (armv8-a): Sort in canonical order.
19998 (armv8.1-a, armv8.2-a): Likewise.
19999 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20001 (cortex-a9): Sort in canonical order.
20002 * config/arm/arm.c (selftests.h): Include it.
20003 (arm_test_cpu_arch_data): New function.
20004 (arm_run_self_tests): New function.
20005 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20006 (targetm): Move declaration to the end of the file.
20007 * arm-cpu-cdata.h: Regenerated.
20009 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20011 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20012 call to target_mode_check describing the type of option passed.
20013 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20014 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20015 arm_parse_cpu_option_name to match parameters against list of
20017 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20018 arm_arch_core_flags data structure.
20019 * config/arm/arm-cpu_cdata.h: Regenerated.
20021 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20023 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20025 (arm_print_hint_for_cpu_option): Likewise.
20026 (arm_print_hint_for_arch_option): Likewise.
20027 (arm_parse_cpu_option_name): Likewise.
20028 (arm_parse_arch_option_name): Likewise.
20029 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20030 of entries in the all_fpus list.
20031 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20032 (arm_parse_cpu_option_name): Declare.
20033 (arm_parse_arch_option_name): Declare.
20034 (arm_parse_option_features): Declare.
20035 (arm_intialize_isa): Declare.
20036 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20038 (gen_comm_data): ... here. Make definitions non-static.
20039 * config/arm/arm-cpu-data.h: Regenerated.
20040 * config/arm/arm-cpu-cdata.h: Regenerated.
20042 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20044 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20045 (cpu_arch_extension): New structure.
20046 (cpu_arch_option, arch_option, cpu_option): New structures.
20047 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20048 architecture types.
20049 (gen_data): Generate new format data tables.
20050 * config/arm/arm.c (cpu_tune): New structure.
20051 (cpu_option, processors): Delete.
20052 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20053 (arm_print_hint_for_cpu_option): ... this and ...
20054 (arm_print_hint_for_arch_option): ... this.
20055 (arm_parse_arch_cpu_name): Delete. Replace with ...
20056 (arm_parse_cpu_option_name): ... this and ...
20057 (arm_parse_arch_option_name): ... this.
20058 (arm_unrecognized_feature): Change type of target parameter to
20060 (arm_parse_arch_cpu_features): Delete. Replace with ...
20061 (arm_parse_option_features): ... this.
20062 (arm_configure_build_target): Rework to use new configuration data
20064 (arm_print_tune_info): Rework for new configuration data tables.
20065 * config/arm/arm-cpu-data.h: Regenerated.
20066 * config/arm/arm-cpu.h: Regenerated.
20068 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20070 * Makefile.in (OBJS): Move sbitmap.o from here ...
20071 (OBJS-libcommon): ... to here.
20073 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20075 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20076 (ISA_ALL_CRYPTO): New macro.
20077 (ISA_ALL_SIMD): New macro
20078 (ISA_ALL_FP): New macro.
20079 * config/arm/arm.c (fpu_bitlist): Update initializer.
20080 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20082 (arm9e): Add fpu. Add option for nofp
20083 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20084 (arm926ej-s, arm1026ej-s): Likewise.
20085 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20086 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20087 neon-fp16, neon-vfpv4, nofp and nosimd.
20088 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20089 (cortex-a8): Add fpu. Add option for nofp.
20090 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20091 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20092 (cortex-r4f): Add fpu.
20093 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20094 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20096 (cortex-r8): Likewise.
20097 (cortex-m4): Add fpu. Add option for nofp.
20098 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20099 (cortex-a17.cortex-a7): Likewise.
20100 (cortex-a32): Add fpu. Add options for crypto and nofp.
20101 (cortex-a35, cortex-a53): Likewise.
20102 (cortex-a57): Add fpu. Add option for crypto.
20103 (cortex-a72, cortex-a73): Likewise.
20104 (exynos-m1): Likewise.
20105 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20106 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20107 (cortex-m33): Add fpu. Add option for nofp.
20108 * config/arm/arm-cpu-cdata.h: Regenerated
20109 * config/arm/arm-cpu-data.h: Regenerated.
20111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20113 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20114 (armv5te, armv5tej): Likewise.
20115 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20116 (armv7): Add options fp and vfpv3-d16.
20117 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20118 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20120 (armv7ve): Likewise.
20121 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20122 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20123 (armv8-a): Add nocrypto option.
20124 (armv8.1-a, armv8.2-a): Likewise.
20125 (armv8-m.main): add options fp, fp.dp and nofp.
20127 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20129 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20131 (armv8-a+crc): Delete.
20132 (armv8.1-a): Add options simd, crypto and nofp.
20133 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20134 (armv8.2-a+fp16): Delete.
20135 (armv8-m.main): Add option dsp.
20136 (armv8-m.main+dsp): Delete.
20137 (cortex-a8): Add fpu. Add nofp option.
20138 (cortex-a9): Add fpu. Add nofp and nosimd options.
20139 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20140 link to main cpu and architecture data structures.
20141 (gen_comm_data): Only put isa attributes from the main architecture
20143 (option): New statement for architecture and CPU entries.
20144 * arm.c (struct cpu_option): New structure.
20145 (struct processors): Add entry for options.
20146 (arm_unrecognized_feature): New function.
20147 (arm_parse_arch_cpu_name): Ignore any characters after the first
20149 (arm_parse_arch_cpu_feature): New function.
20150 (arm_configure_build_target): Separate out any CPU and architecture
20151 features and parse separately. Don't error out if -mfpu=auto is
20152 used with only an architecture string.
20153 (arm_print_asm_arch_directives): New function.
20154 (arm_file_start): Call it.
20155 * config/arm/arm-cpu-cdata.h: Regenerated.
20156 * config/arm/arm-cpu-data.h: Likewise.
20157 * config/arm/arm-tables.opt: Likewise.
20159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20161 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20162 assembler when it is not -mfpu=auto.
20164 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20166 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20167 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20168 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20169 (ASM_CPU_SPEC): Rewrite.
20170 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20171 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20172 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20173 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20174 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20175 copied string is NUL-terminated. Also strip any characters prefixed
20177 (arm_rewrite_selected_arch): New function.
20178 (arm_rewrite_march): New function.
20180 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20182 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20183 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20184 (march, mcpu, mtune): Convert to string-based options.
20185 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20186 (arm_parse_arch_cpu_name): New function.
20187 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20188 identify selected architecture or CPU.
20189 (arm_option_save): New function.
20190 (TARGET_OPTION_SAVE): Redefine.
20191 (arm_option_restore): Restore string options.
20192 (arm_option_print): Print string options.
20194 2017-06-16 Martin Sebor <msebor@redhat.com>
20196 PR tree-optimization/80933
20197 PR tree-optimization/80934
20198 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20199 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20200 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20201 (gimple_fold_builtin): Call them.
20203 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20205 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20206 as unlikely; update profile.
20208 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20210 * predict.c (force_edge_cold): Handle declaring edges impossible
20213 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20215 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20217 (try_unroll_loop_completely): Fix reporting.
20219 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20221 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20223 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20226 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20227 if given a non-constant argument for an intrinsic which requires a
20230 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20232 * profile.c (compare_freqs): New function.
20233 (branch_prob): Sort edge list.
20234 (find_spanning_tree): Assume that the list is priority sorted.
20236 2017-06-16 Richard Biener <rguenther@suse.de>
20238 PR tree-optimization/81090
20239 * passes.def (pass_record_bounds): Remove.
20240 * tree-pass.h (make_pass_record_bounds): Likewise.
20241 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20242 make_pass_record_bounds): Likewise.
20243 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20244 not free niter estimates at the beginning but at the end.
20245 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20247 2017-06-16 Richard Biener <rguenther@suse.de>
20249 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20250 initializer to workaround ICE in host GCC 4.8.
20252 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20254 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20256 (clone_inlined_nodes): Update.
20258 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20260 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20261 prefetch settings, and enable prefetching by default at -O3.
20263 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20265 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20266 Set flag_prefetch_loop_arrays according to tuning data.
20268 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20270 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20271 New tune structure.
20272 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20273 [Unrelated to main purpose of the patch] Place the pointer field last
20274 to enable type checking errors when tune structure are wrongly merged.
20275 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20276 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20277 (thunderx2t99_prefetch_tune): New tune constants.
20278 (tune_params *_tunings): Update all tunings (no functional change).
20279 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20280 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20281 from tunings structures.
20283 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20286 * ubsan.c (instrument_null): Add T argument, use it instead
20287 of computing it based on IS_LHS.
20288 (instrument_object_size): Likewise.
20289 (pass_ubsan::execute): Adjust instrument_null and
20290 instrument_object_size callers to pass gimple_get_lhs or
20291 gimple_assign_rhs1 result to it. Use instrument_null instead of
20292 calling get_base_address and instrument_mem_ref. Handle
20293 aggregate call arguments for object-size sanitization.
20295 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20297 PR tree-optimization/81089
20298 * tree-vrp.c (is_masked_range_test): Validate operands of
20301 2017-06-15 Martin Sebor <msebor@redhat.com>
20304 * dumpfile.c (dump_register): Avoid calling memset to initialize
20305 a class with a default ctor.
20306 * gcc.c (struct compiler): Remove const qualification.
20307 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20308 * hash-table.h: Ditto.
20309 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20311 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20312 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20314 * params.h (struct param_info): Make struct members non-const.
20315 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20316 with default initialization.
20317 * vec.h (vec_copy_construct, vec_default_construct): New helper
20319 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20320 with vec_copy_construct.
20321 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20322 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20323 * doc/invoke.texi (-Wclass-memaccess): Document.
20325 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20327 * emit-rtl.h (is_leaf): Update comment about local
20328 register allocator.
20330 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20333 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20334 for a variable to have a section before checking if the section has a
20336 Set section to.persistent if persistent attribute is set.
20337 Warn if .persistent attribute is used on an automatic variable.
20339 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20341 PR rtl-optimization/80474
20342 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20344 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20346 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20349 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20351 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20352 (casesi): Emit bounds checking as RTL.
20353 (casesi_internal_mips16_<mode>): Remove bounds checking.
20355 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20357 * config/xtensa/xtensa.c (xtensa_option_override): Append
20358 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20359 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20360 xtensa_doloop_hooks): Define unconditionally.
20361 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20363 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20364 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20365 for it in xtensa_option_override.
20366 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20367 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20369 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20371 * doc/cppopts.texi: Document '-' special value to -MF.
20373 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20375 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20376 (cortex_a53_fconst): Likewise.
20377 (cortex_a53_fpmul): Likewise.
20378 (cortex_a53_f_load_64): Likewise.
20379 (cortex_a53_f_load_many): Likewise.
20380 (cortex_a53_advsimd_alu): Likewise.
20381 (cortex_a53_advsimd_alu_q): Likewise.
20382 (cortex_a53_advsimd_mul): Likewise.
20383 (cortex_a53_advsimd_mul_q): Likewise.
20384 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20385 Add missing fmul, r2f_cvt and fconst cases.
20387 2017-06-14 Richard Biener <rguenther@suse.de>
20389 PR middle-end/81088
20390 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20392 (fold_binary_loc): When associating do not treat pre-existing
20393 TREE_OVERFLOW on literal constants as a reason to allow
20394 TREE_OVERFLOW on associated literal constants.
20396 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20398 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20399 (MASK_FEATURES): New macro.
20400 * config/sparc/sparc.c (sparc_option_override): Remove the special
20401 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20403 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20405 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20406 a division of 0 if non-call exceptions are enabled.
20408 2017-06-14 Andrew Pinski <apinski@cavium.com>
20409 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20412 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20413 Improve vector initialization code gen for only variable case.
20415 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20417 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20419 2017-06-14 Richard Biener <rguenther@suse.de>
20421 PR tree-optimization/81083
20422 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20425 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20427 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20428 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20429 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20430 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20431 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20432 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20434 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20436 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20437 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20439 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20441 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20443 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20445 * config/rs6000/t-rtems: Don't handle SPE.
20447 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20449 * config/rs6000/t-linux: Don't handle SPE.
20451 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20453 * config/rs6000/eabispe.h: Delete file.
20455 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20457 * config/rs6000/t-spe: Delete file.
20459 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20461 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20462 (rs6000_legitimate_offset_address_p): Return false for anything in
20463 V2SImode or V2SFmode.
20465 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20467 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20468 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20469 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20471 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20472 (rs6000_legitimate_offset_address_p): Ditto.
20473 (rs6000_emit_move): Ditto.
20474 (rs6000_init_builtins): Remove V4HI_type_node.
20476 2017-06-13 Martin Liska <mliska@suse.cz>
20479 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20480 (gate_asan): Likewise.
20481 * asan.h (asan_no_sanitize_address_p): Remove the function.
20482 (sanitize_flags_p): New function.
20483 * builtins.def: Fix coding style.
20484 * common.opt: Use renamed enum value.
20485 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20486 * doc/extend.texi: Document no_sanitize attribute.
20487 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20488 to SANITIZE_UNDEFINED_NONDEFAULT.
20489 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20490 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20491 Use sanitize_flags_p.
20492 * gimplify.c (gimplify_function_tree): Likewise.
20493 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20494 * opts.c (parse_no_sanitize_attribute): New function.
20495 (common_handle_option): Use renamed enum value.
20496 * opts.h (parse_no_sanitize_attribute): Declare.
20497 * tree.c (sanitize_flags_p): New function.
20498 * tree.h: Declared here.
20499 * tsan.c: Use sanitize_flags_p.
20500 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20501 (instrument_mem_ref): Likewise.
20502 (instrument_bool_enum_load): Likewise.
20503 (do_ubsan_in_current_function): Remove the function.
20504 (pass_ubsan::execute): Use sanitize_flags_p.
20505 * ubsan.h: Remove do_ubsan_in_current_function
20506 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20507 (dump_function_to_file): Use it here.
20509 2017-06-13 Martin Jambor <mjambor@suse.cz>
20511 PR tree-optimization/80803
20512 PR tree-optimization/81063
20513 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20514 (propagate_subaccesses_across_link): Enqueue subtree whenever
20515 necessary instead of relying on the caller.
20517 2017-06-13 Martin Jambor <mjambor@suse.cz>
20519 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20520 that have a first_link.
20521 (sort_and_splice_var_accesses): Do not check first_link before
20523 (subtree_mark_written_and_enqueue): Likewise.
20524 (propagate_all_subaccesses): Likewise and do not stop at first
20525 parent with a first_link.
20527 2017-06-13 Martin Jambor <mjambor@suse.cz>
20529 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20532 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20534 * match.pd: New pattern.
20536 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20538 * tree-vrp.c (is_masked_range_test): New function.
20539 (register_edge_assert_for): Determine ranges for
20542 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20544 PR tree-optimization/67328
20545 * fold-const.c (maskable_range_p): New function.
20546 (build_range_check): Generate bittests if possible.
20548 2017-06-13 Martin Liska <mliska@suse.cz>
20550 * gimple-pretty-print.c (dump_probability): Add new argument.
20551 (dump_edge_probability): Dump both probability and count.
20552 (dump_gimple_label): Likewise.
20553 (dump_gimple_bb_header): Likewise.
20555 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20558 * config/avr/avr-devices.c: Fix indentation.
20559 * config/avr/gen-avr-mmcu-specs.c: Dito.
20561 2017-06-13 Richard Biener <rguenther@suse.de>
20563 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20564 instead get vector type from stmt_info.
20565 (vectorizable_reduction): Adjust. Remove dead code.
20567 2017-06-13 Richard Biener <rguenther@suse.de>
20569 PR middle-end/81065
20570 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20571 case of C * (x * C2 + C3).
20572 (fold_addr_of_array_ref_difference): Properly fold index difference.
20574 2017-06-12 David S. Miller <davem@davemloft.net>
20577 * config/sparc/sparc.md (return expander): Emit frame blockage if
20578 function uses alloca.
20580 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20582 * combine.c (make_field_assignment): Check len rather than the mode
20583 precision when calling force_to_mode.
20585 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20587 Support multilibs and devices that see flash in RAM address range.
20590 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20591 (avr_mcu_t) <flash_pm_offset>: New field.
20592 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20593 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20594 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20595 (AVR_TINY_PM_OFFSET): Remove macro.
20596 * config/avr/avr.opt (-mshort-calls): New option.
20597 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20598 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20599 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20600 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20601 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20602 instead of avr_arch->have_jmp_call.
20603 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20604 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20605 avr_arch->flash_pm_offset to define.
20606 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20607 new field flash_pm_offset. Add entry for avrxmega3.
20608 (avr_texinfo): Add entry for avrxmega3.
20609 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20610 attiny212, attiny214,
20611 attiny412, attiny414, attiny416, attiny417,
20612 attiny814, attiny816, attiny817,
20613 attiny1614, attiny1616, attiny1617,
20614 attiny3214, attiny3216, attiny3217.
20615 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20616 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20617 (avr_print_operand_address) [AVR_TINY]: Same.
20618 (avr_asm_init_sections) <readonly_data_section>: Only patch
20619 callback if avr_arch->flash_pm_offset = 0.
20620 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20621 for rodata if avr_arch->flash_pm_offset != 0.
20622 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20623 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20624 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20625 (m_options): Append opt_rcall.
20626 (m_dirnames): Append dir_rcall.
20627 * config/avr/t-multilib: Regenerate.
20629 * configure.ac [target=avr]: Check whether avrxmega3 default
20630 linker description file works as needed.
20631 * configure: Regenerate.
20632 * doc/avr-mmcu.texi: Regenerate.
20633 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20634 <__AVR_ARCH__>: Document avrxmega3 and 103.
20635 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20636 <__AVR_SHORT_CALLS__>: Document it.
20637 <__AVR_PM_BASE_ADDRESS__>: Document it.
20638 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20639 (AVR Variable Attributes) <progmem>: Document this is
20640 not needed for avrxmega3.
20641 (AVR Named Address Spaces) <__flash>: Dito.
20643 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20645 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20647 2017-06-12 Doug Rupp <rupp@adacore.com>
20649 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20650 Append vxworks-stdint.h to the tm_file list.
20651 * config/vxworks-stdint.h: New file.
20653 2017-06-12 Martin Liska <mliska@suse.cz>
20655 PR tree-optimization/81041
20656 * tree-profile.c (gimple_gen_ic_func_profiler):
20657 Create an extra BB in profile-generate
20658 (gimple_gen_time_profiler): Likewise.
20660 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20662 PR tree-optimization/81003
20663 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20664 (update_range_test): Use it instead of force_gimple_operand_gsi.
20666 2017-06-12 Richard Biener <rguenther@suse.de>
20668 PR tree-optimization/81053
20669 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20670 with backedge value not defined in loop. Simplify def stmt
20673 2017-06-11 Tom de Vries <tom@codesourcery.com>
20676 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20678 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20679 nvptx_cannot_force_const_mem.
20681 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20683 * opts.c (finish_options): Move test for flag_split_stack after
20684 it has been initialized.
20686 2017-06-11 Jason Merrill <jason@redhat.com>
20688 * tree.h (id_equal): New.
20689 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20690 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20691 instead of strcmp of IDENTIFIER_POINTER.
20693 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20695 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20696 (mark_all_inlined_calls_cdtor): Fix formating.
20697 (inline_transform): Rescale profile before inlining.
20699 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20701 * cgraph.h (cgraph_edge::clone): Update prototype.
20702 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20703 (cgraph_node::create_clone): Update.
20704 (cgraph_node::create_version_clone): Update.
20705 * tree-inline.c (copy_bb): Update.
20706 (expand_call_inline): Update.
20708 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20710 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20711 factored out from ...
20712 (rs6000_emit_prologue): ... here.
20714 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20716 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20717 factored out from ...
20718 (rs6000_emit_prologue): ... here.
20720 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20722 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20724 (handle_missing_profiles): Fix computation of tp_first_run.
20725 (counts_to_freqs): Do not touch freqs when count is 0.
20727 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20729 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20732 2017-06-10 Tom de Vries <tom@codesourcery.com>
20734 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20735 attributes): Document signal effective target.
20737 2017-06-10 Tom de Vries <tom@codesourcery.com>
20739 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20740 Document effective target stack_size.
20742 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20744 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20745 to the edit_context if they can be auto-applied.
20747 2017-06-9 Ian Lance Taylor <iant@golang.org>
20749 * opts.c (finish_options): If -fsplit-stack, disable implicit
20750 -forder-blocks-and-partition.
20751 * doc/invoke.texi (Optimize Options): Document that when using
20752 -fsplit-stack -forder-blocks-and-partition is not implicitly
20755 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20757 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20758 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20759 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20760 * builtins.def (abort, trap, unreachable): Declare cold.
20761 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20762 * tree-core.h (ECF_COLD): New.
20763 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20764 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20766 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20768 * predict.c (unlikely_executed_stmt_p): Cleanup.
20770 2017-06-09 Richard Biener <rguenther@suse.de>
20772 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20773 model if the ref is always written to.
20775 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20777 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20779 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20781 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20784 2017-06-09 Tom de Vries <tom@codesourcery.com>
20787 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20788 "target cannot support label values" when encountering LABEL_REF.
20790 2017-06-09 Martin Liska <mliska@suse.cz>
20792 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20793 (gimple_gen_ic_func_profiler): Emit direct comparison
20794 of __gcov_indirect_call_callee with NULL.
20795 (gimple_gen_time_profiler): Change probability from
20796 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20798 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20800 * profile.c (edge_gcov_counts): Turn to pointer.
20801 (compute_branch_probabilities, compute_branch_probabilities): Update.
20802 (branch_prob): Do not clear edge_gcov_count.
20803 * profile.h (edge_gcov_counts): Turn to pointer.
20804 (edge_gcov_count): Update.
20806 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20808 * gimple.h (gimple_check_failed): Mark cold.
20810 2017-06-09 Richard Biener <rguenther@suse.de>
20812 PR tree-optimization/66623
20813 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20814 refactor check_reduction into two parts, properly computing
20815 whether we have to check reduction validity for outer loop
20818 2017-06-09 Richard Biener <rguenther@suse.de>
20820 PR tree-optimization/79483
20821 * graphite-scop-detection.c (order): New global.
20822 (get_order): Compute bb to order mapping that satisfies code
20823 generation constraints.
20824 (cmp_pbbs): New helper.
20825 (build_scops): Start domwalk at entry block, sort generated
20828 2017-06-09 Richard Biener <rguenther@suse.de>
20830 PR middle-end/81007
20831 * ipa-polymorphic-call.c
20832 (ipa_polymorphic_call_context::restrict_to_inner_class):
20833 Skip FIELD_DECLs with error_mark_node type.
20834 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20837 2017-06-09 Martin Liska <mliska@suse.cz>
20839 * predict.c (struct branch_predictor): New struct.
20840 (test_prediction_value_range): New test.
20841 (predict_c_tests): New function.
20842 * selftest-run-tests.c (selftest::run_tests): Run the function.
20843 * selftest.h: Declare new tests.
20845 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20848 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20849 gen_add3_insn did not fail.
20850 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20851 r0, construct that number in a temporary reg and add that reg to r0.
20852 If asked to put the result in r0 as well, fail.
20854 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20856 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20857 for early expansion of vec_eqv.
20859 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20861 PR middle-end/81005
20862 * ubsan.c (instrument_null): Avoid pointless code temporary.
20863 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20865 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20869 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20872 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20875 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20877 * predict.c (unlikely_executed_edge_p): Move ahead.
20878 (probably_never_executed_edge_p): Use it.
20880 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20882 PR middle-end/79988
20883 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20884 gimple_call_builtin_p call.
20886 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20888 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20889 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20890 rtl_check_failed_type2, rtl_check_failed_code1,
20891 rtl_check_failed_code2, rtl_check_failed_code_mode,
20892 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20893 rtvec_check_failed_bounds, rtl_check_failed_flag,
20894 _fatal_insn_not_found, _fatal_insn): Likewise.
20895 * tree.h (tree_contains_struct_check_failed,
20896 tree_check_failed, tree_not_check_failed,
20897 tree_class_check_failed, tree_range_check_failed,
20898 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20899 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20900 tree_operand_check_failed, omp_clause_check_failed,
20901 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20904 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20906 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20907 flag_branch_probabilities.
20908 * ipa-inline.c (edge_badness): Likewise.
20909 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20910 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20911 * predict.c (maybe_hot_frequency_p): Likewise.
20912 (probably_never_executed): Likewise.
20913 * sched-ebb.c (schedule_ebbs): Likewise.
20914 * sched-rgn.c (find_single_block_region): Likewise.
20915 * tracer.c (tail_duplicate): Likewise.
20917 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20919 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20920 longer requires x_flag_profile_use.
20922 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20924 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20925 instead of flag_reorder_blocks_and_partition.
20926 * dbxout.c (dbxout_function_end): Likewise.
20927 * dwarf2out.c (gen_subprogram_die): Likewise.
20928 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20929 * hw-doloop.c (reorg_loops): Likewise.
20930 * varasm.c (assemble_start_function,
20931 assemble_end_function): Likewise.
20932 (decide_function_section): Do not check for
20933 flag_reorder_blocks_and_partition.
20935 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20937 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20939 (chkp_get_hard_register_fake_addr_expr): Ditto.
20940 (chkp_build_addr_expr): Add check for hard reg case.
20941 (chkp_parse_array_and_component_ref): Ditto.
20942 (chkp_find_bounds_1): Ditto.
20943 (chkp_process_stmt): Don't generate bounds store for
20946 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20948 * predict.c (maybe_hot_bb_p): Do not check profile status.
20949 (maybe_hot_edge_p): Likewise.
20950 (probably_never_executed): Check for zero counts even if profile
20952 (unlikely_executed_edge_p): New function.
20953 (unlikely_executed_stmt_p): New function.
20954 (unlikely_executed_bb_p): New function.
20955 (set_even_probabilities): Use unlikely predicates.
20956 (combine_predictions_for_bb): Likewise.
20957 (predict_paths_for_bb): Likewise.
20958 (predict_paths_leading_to_edge): Likewise.
20959 (determine_unlikely_bbs): New function.
20960 (estimate_bb_frequencies): Use it.
20961 (compute_function_frequency): Use zero counts even if profile is
20963 * profile-count.h: Fix typo.
20965 2017-08-08 Julia Koval <julia.koval@intel.com>
20967 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20968 _mm512_mask_cvtsepi16_storeu_epi8,
20969 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20970 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20971 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20972 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20973 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20974 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20975 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20976 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20977 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20978 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20979 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20980 __builtin_ia32_pmovuswb256mem_mask,
20981 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20982 __builtin_ia32_pmovwb512mem_mask): New builtins.
20984 2017-08-08 Julia Koval <julia.koval@intel.com>
20986 PR target/73350,80862
20987 * config/i386/subst.md (round): Fix round pattern.
20988 * config/i386/i386.c (ix86_erase_embedded_rounding):
20989 Fix erasing rounding for the fixed pattern.
20991 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20993 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20995 2017-06-08 Martin Liska <mliska@suse.cz>
20997 PR gcov-profile/80911
20998 * gcov.c (block_info::block_info): New constructor.
21000 2017-06-07 Carl Love <cel@us.ibm.com>
21002 * config/rs6000/rs6000-c: The return type of the following
21003 built-in functions was implemented as int not long long. Fix sign
21004 of return value for the unsigned version of vec_mulo and vec_mule.
21005 vector unsigned long long vec_bperm (vector unsigned long long,
21006 vector unsigned char)
21007 vector signed long long vec_mule (vector signed int,
21009 vector unsigned long long vec_mule (vector unsigned int,
21010 vector unsigned int)
21011 vector signed long long vec_mulo (vector signed int,
21013 vector unsigned long long vec_mulo (vector unsigned int,
21014 vector unsigned int)
21015 * doc/extend.texi: Fix the documentation for the built-in
21018 2017-06-07 Carl Love <cel@us.ibm.com>
21021 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21024 2017-06-07 Carl Love <cel@us.ibm.com>
21026 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21027 support, Generate doublehv for signed int/float for BE case only.
21029 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21031 * doc/invoke.texi (mcx16): Rewrite.
21033 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21035 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21036 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21037 *mov<mode>_softfloat, and an anonymous splitter): Use
21038 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21040 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21042 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21044 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21045 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21046 (rs6000_debug_reg_global): Adjust.
21047 (rs6000_init_hard_regno_mode_ok): Adjust.
21048 (rs6000_dbx_register_number): Adjust.
21049 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21050 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21051 Remove SPE_ACC and SPEFSCR.
21052 (REG_ALLOC_ORDER): Ditto.
21053 (FRAME_POINTER_REGNUM): Change to 111.
21054 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21055 (REG_CLASS_NAMES): Ditto.
21056 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21057 (REGISTER_NAMES): Ditto.
21058 (ADDITIONAL_REG_NAMES): Ditto.
21059 (rs6000_reg_names): Ditto.
21060 * config/rs6000/rs6000.md: Renumber some register number
21063 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21065 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21067 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21068 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21070 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21071 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21072 Delete the SPE high registers.
21073 (REG_ALLOC_ORDER): Ditto.
21074 (enum reg_class): Remove SPE_HIGH_REGS.
21075 (REG_CLASS_NAMES): Ditto.
21076 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21077 (REGISTER_NAMES): Ditto.
21078 (rs6000_reg_names): Ditto.
21079 * doc/tm.texi.in: Remove SPE as example.
21080 * doc/tm.texi: Regenerate.
21082 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21084 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21085 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21086 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21087 * config/rs6000/rs6000.md (type): Remove "brinc".
21089 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21091 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21092 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21093 * config/rs6000/linuxspe.h: Delete file.
21094 * config/rs6000/rs6000.md: Don't include spe.md.
21095 * config/rs6000/spe.h: Delete file.
21096 * config/rs6000/spe.md: Delete file.
21097 * config/rs6000/t-rs6000: Remove spe.md.
21099 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21101 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21102 (reg_or_none500mem_operand): Delete.
21103 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21104 instead of reg_or_none500mem_operand.
21106 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21108 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21109 handling of SPE flags.
21110 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21112 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21114 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21116 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21117 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21118 paired_divv2sf3): Similar.
21119 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21120 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21121 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21123 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21124 Rename the paired_* instruction patterns.
21125 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21127 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21128 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21129 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21130 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21131 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21132 PAIRED_VECTOR_MODE.
21133 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21134 (spe_func_has_64bit_regs_p): Delete.
21135 (spe_expand_predicate_builtin): Delete.
21136 (spe_expand_evsel_builtin): Delete.
21137 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21138 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21139 (invalid_e500_subreg): Delete.
21140 (rs6000_legitimize_address): Always force_reg op2 as well, for
21141 paired single memory accesses.
21142 (rs6000_member_type_forces_blk): Delete.
21143 (rs6000_spe_function_arg): Delete.
21144 (rs6000_expand_unop_builtin): Delete SPE handling.
21145 (rs6000_expand_binop_builtin): Ditto.
21146 (spe_expand_stv_builtin): Delete.
21147 (bdesc_2arg_spe): Delete.
21148 (spe_expand_builtin): Delete.
21149 (spe_expand_predicate_builtin): Delete.
21150 (spe_expand_evsel_builtin): Delete.
21151 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21152 (spe_init_builtins): Delete.
21153 (spe_func_has_64bit_regs_p): Delete.
21154 (savres_routine_name): Delete "info" parameter. Adjust callers.
21155 (rs6000_emit_stack_reset): Ditto.
21156 (rs6000_dwarf_register_span): Delete.
21157 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21158 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21159 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21161 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21163 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21164 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21165 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21166 mulv2sf3, divv2sf3): Delete expanders.
21168 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21170 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21172 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21174 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21175 * config/rs6000/rs6000.c: Ditto.
21177 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21179 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21180 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21181 comparison_operator.
21183 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21185 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21186 * config/rs6000/rs6000.opt: Ditto.
21187 * config/rs6000/t-rtems: Ditto.
21189 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21191 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21192 TARGET_E500_SINGLE by 0, simplify.
21193 * config/rs6000/rs6000.c: Ditto.
21194 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21195 (spe_build_register_parallel): Delete.
21196 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21197 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21198 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21199 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21200 (E500_CONVERT): Delete.
21201 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21203 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21205 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21206 * config/rs6000/dfp.md: Ditto.
21207 (negdd2, *negdd2_fpr): Merge.
21208 (absdd2, *absdd2_fpr): Merge.
21209 (negtd2, *negtd2_fpr): Merge.
21210 (abstd2, *abstd2_fpr): Merge.
21211 * config/rs6000/e500.h: Delete file.
21212 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21213 TARGET_FPRS by 1 and simplify.
21214 * config/rs6000/rs6000-c.c: Ditto.
21215 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21216 TARGET_DF_SPE by 0.
21217 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21219 * config/rs6000/rs6000.md: Ditto.
21220 (floatdidf2, *floatdidf2_fpr): Merge.
21221 (move_from_CR_gt_bit): Delete.
21222 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21223 (E500_CR_IOR_COMPARE): Delete.
21224 (All patterns that require !TARGET_FPRS): Delete.
21225 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21227 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21229 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21231 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21233 * graphds.c (add_edge): Intitialize edge's attached data.
21234 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21235 pointer parameter. Call pointed function on each edge during
21236 graph traversing. Skip traversing the edge when the function
21238 (graphds_dfs, graphds_scc): Ditto.
21239 (for_each_edge): New parameter. Pass the new parameter to callback
21241 * graphds.h (skip_edge_callback): New function pointer type.
21242 (graphds_dfs, graphds_scc): New function pointer parameter.
21243 (graphds_edge_callback, for_each_edge): New parameter.
21245 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21247 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21248 out code checking if runtime alias check is possible to below ...
21249 Call the new function.
21250 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21251 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21253 2017-06-07 Marek Polacek <polacek@redhat.com>
21256 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21257 TYPE_OVERFLOW_WRAPS check.
21259 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21261 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21262 if versioning is required.
21263 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21264 peeling with the check for versioning.
21266 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21268 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21269 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21270 Set true to new parameter if new ssa variable is defined.
21271 (vect_gen_vector_loop_niters): Refactor. Set range information
21272 for the new vector loop bound variable.
21273 (vect_do_peeling): Ditto.
21275 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21277 * tree-affine.c (ssa.h): Include header file.
21278 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21279 has wrapping overflow behavior.
21281 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21283 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21285 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21287 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21288 (tree_to_aff_combination): ... here.
21290 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21292 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21293 reg_pressure model function.
21294 (ivopts_global_cost_for_size): Delete.
21295 (determine_set_costs, iv_ca_recount_cost): Call new model function
21296 ivopts_estimate_reg_pressure.
21298 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21300 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21301 expensive than udiv. Remove floating point cases from mod.
21303 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21305 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21306 Increase idiv cost.
21308 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21310 * config/aarch64/aarch64.md
21311 (copysignsf3): Fix mask generation.
21313 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21315 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21317 (class dump_manager): Add register_dumps method.
21318 * dumpfile.c: Include langhooks.h.
21319 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21320 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21321 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21322 (dump_manager::dump_register): Start with 512 entries instead of 32.
21323 (dump_manager::register_dumps): New method.
21324 * toplev.c (general_init): Instead of invoking register_dumps
21325 langhook, invoke register_dumps method on the dump manager.
21326 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21329 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21331 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21332 Say that # requires an associated define_split to exist, and that
21333 the define_split must be suitable for use after register allocation.
21335 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21337 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21338 (compute_outgoing_frequencies): Also initialize zero counts.
21339 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21340 around loops; preserve more of profile when nothing changes.
21342 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21344 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21346 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21347 * config/arm/arm-cpu-cdata.h: Regenerate.
21348 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21349 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21350 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21351 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21352 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21354 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21356 * config/arm/t-rmprofile: Likewise.
21357 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21359 2017-06-06 David S. Miller <davem@davemloft.net>
21362 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21363 blockage if function uses alloca.
21365 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21367 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21368 New "uid" fields to hold pretty-print IDs of group and ref.
21369 Memory references are now identified as <group_id>:<ref_id>
21370 instead of using [random] addresses.
21371 (dump_mem_details): Simplify, no functional change.
21372 (dump_mem_ref): Simplify and make output more concise.
21373 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21374 (find_or_create_group): Initialize group uid.
21375 (record_ref): Initialize ref uid. Improve debug output.
21376 (prune_group_by_reuse, should_issue_prefetch_p,)
21377 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21378 (mark_nontemporal_store, determine_loop_nest_reuse):
21379 Improve debug output.
21381 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21383 * dbgcnt.def (prefetch): New debug counter.
21384 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21385 (schedule_prefetches): Stop issueing prefetches if debug counter
21388 2017-06-06 Tom de Vries <tom@codesourcery.com>
21390 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21391 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21393 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21395 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21396 Use aarch64_reg_or_zero predicate for operand 4.
21397 (aarch64_compare_and_swap<mode> define_insn_and_split):
21398 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21399 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21401 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21403 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21404 (arm_compute_save_core_reg_mask): This.
21405 (thumb1_compute_save_reg_mask): Rename into ...
21406 (thumb1_compute_save_core_reg_mask): This.
21407 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21408 (arm_compute_frame_layout): Likewise.
21410 2017-06-06 Richard Biener <rguenther@suse.de>
21412 PR tree-optimization/80974
21413 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21414 keep or clear leaders SSA info.
21416 2017-06-06 Tom de Vries <tom@codesourcery.com>
21418 * config/nvptx/nvptx.c (split_mode_p): New function.
21419 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21421 2017-06-06 Tom de Vries <tom@codesourcery.com>
21423 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21425 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21428 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21431 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21433 * shrink-wrap.c (handle_simple_exit): Update profile.
21434 (try_shrink_wrapping): Upate profile.
21436 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21438 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21439 (tree_guess_outgoing_edge_probabilities): New.
21440 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21441 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21443 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21445 * ipa-split.c (split_function): Initialize return bb profile.
21447 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21449 * profile.c (compute_branch_probabilities): Also initialize
21450 EXIT_BLOCK profile.
21452 2017-06-06 Richard Biener <rguenther@suse.de>
21454 PR tree-optimization/80928
21455 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21456 (vect_analyze_loop_operations): Properly guard analysis for
21458 (vect_transform_loop): Likewise.
21459 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21460 (vect_model_induction_cost): Do not cost for pure SLP.
21461 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21462 of induction in inner loop vectorization.
21463 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21464 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21465 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21467 (vect_analyze_slp_cost_1): Cost induction.
21468 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21469 (vect_get_slp_vect_defs): Likewise.
21470 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21471 (vect_transform_stmt): Handle SLP reductions.
21472 * tree-vectorizer.h (vectorizable_induction): Adjust.
21474 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21476 * config/rs6000/rs6000.c (make_resolver_func): Update
21477 init_lowered_empty_function call.
21479 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21481 * doc/invoke.texi: Document the -fprofile-abs-path option.
21482 * common.opt (fprofile-abs-path): New option.
21483 * gcov-io.h (gcov_write_filename): Declare.
21484 * gcov-io.c (gcov_write_filename): New function.
21485 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21486 * profile.c (output_location): Likewise.
21488 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21490 * shring-wrap.c: Revert accidental commit.
21492 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21494 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21496 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21498 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21500 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21501 profile in callgraph edge.
21502 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21503 (apply_scale): Likewise.
21504 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21505 Also scale profile when inlining function with zero profile.
21506 (initialize_cfun): Update exit block profile even when it is zero.
21507 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21508 when profile is read.
21510 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21512 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21513 (CLONE_*): New constants to define the processors we can generate
21514 code for with the target_clone attribute.
21515 (rs6000_clone_map): New array to identify which clone processors
21516 the current program is running on.
21517 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21518 target_clone attribute.
21519 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21520 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21521 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21522 (cpu_expand_builtin): Add support for target_clone attribute.
21523 (rs6000_valid_attribute_p): Allow "default" attribute.
21524 (get_decl_name): New debug function to simplify printing the
21525 current function name in debugging statements.
21526 (rs6000_clone_priority): New functions to support the target_clone
21527 attribute, and be able to generate code to switch between ISA 2.05
21528 through ISA 3.0 (power6 through power9).
21529 (rs6000_compare_version_priority): Likewise.
21530 (rs6000_get_function_versions_dispatcher): Likewise.
21531 (make_resolver_func): Likewise.
21532 (add_condition_to_bb): Likewise.
21533 (dispatch_function_versions): Likewise.
21534 (rs6000_generate_version_dispatcher_body): Likewise.
21535 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21536 (fusion_gpr_load_p): Fix a spacing issue.
21537 * doc/extend.texi (Common Function Attributes): Document that the
21538 PowerPC supports the target_clone attribute.
21540 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21542 * config/arm/arm.h: explain F symbol found in description of ARM
21543 register allocation in its legend.
21545 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21547 * config/mips/frame-header-opt.c: Include profile-count.h.
21548 * config/riscv/riscv.c: Include profile-count.h
21550 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21552 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21554 (sm_set_flag_if_changed): Add bbs field.
21555 (execute_sm_if_changed_flag_set): Pass BBS.
21556 (execute_sm): Update.
21558 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21560 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21563 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21565 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21566 (peephole2): New peephole2 to emit the above.
21567 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21569 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21571 * config/aarch64/aarch64.c (define_peephole2 above
21572 *sub_<shift>_<mode>): New peephole.
21574 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21576 * config/i386/i386.c (make_resolver_func): Update.
21577 * Makefile.in: Add profile-count.h and profile-count.o
21578 * auto-profile.c (afdo_indirect_call): Update to new API.
21579 (afdo_set_bb_count): Update.
21580 (afdo_propagate_edge): Update.
21581 (afdo_propagate_circuit): Update.
21582 (afdo_calculate_branch_prob): Update.
21583 (afdo_annotate_cfg): Update.
21584 * basic-block.h: Include profile-count.h
21585 (struct edge_def): Turn count to profile_count.
21586 (struct basic_block_def): Likewie.
21587 (REG_BR_PROB_BASE): Move to profile-count.h
21588 (RDIV): Move to profile-count.h
21589 * bb-reorder.c (max_entry_count): Turn to profile_count.
21590 (find_traces): Update.
21591 (rotate_loop):Update.
21592 (connect_traces):Update.
21593 (sanitize_hot_paths):Update.
21594 * bt-load.c (migrate_btr_defs): Update.
21595 * cfg.c (RDIV): Remove.
21596 (init_flow): Use alloc_block.
21597 (alloc_block): Uninitialize count.
21598 (unchecked_make_edge): Uninitialize count.
21599 (check_bb_profile): Update.
21600 (dump_edge_info): Update.
21601 (dump_bb_info): Update.
21602 (update_bb_profile_for_threading): Update.
21603 (scale_bbs_frequencies_int): Update.
21604 (scale_bbs_frequencies_gcov_type): Update.
21605 (scale_bbs_frequencies_profile_count): New.
21606 * cfg.h (update_bb_profile_for_threading): Update.
21607 (scale_bbs_frequencies_profile_count): Declare.
21608 * cfgbuild.c (compute_outgoing_frequencies): Update.
21609 (find_many_sub_basic_blocks): Update.
21610 * cfgcleanup.c (try_forward_edges): Update.
21611 (try_crossjump_to_edge): Update.
21612 * cfgexpand.c (expand_gimple_tailcall): Update.
21613 (construct_exit_block): Update.
21614 * cfghooks.c (verify_flow_info): Update.
21615 (dump_bb_for_graph): Update.
21616 (split_edge): Update.
21617 (make_forwarder_block): Update.
21618 (duplicate_block): Update.
21619 (account_profile_record): Update.
21620 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21621 (get_estimated_loop_iterations): Update.
21622 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21623 (single_likely_exit): Update.
21624 * cfgloopmanip.c (scale_loop_profile): Update.
21626 (set_zero_probability): Update.
21627 (lv_adjust_loop_entry_edge): Update.
21628 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21629 (purge_dead_edges): Update.
21630 (rtl_account_profile_record): Update.
21631 * cgraph.c (cgraph_node::create): Uninitialize count.
21632 (symbol_table::create_edge): Uninitialize count.
21633 (cgraph_update_edges_for_call_stmt_node): Update.
21634 (cgraph_edge::dump_edge_flags): Update.
21635 (cgraph_node::dump): Update.
21636 (cgraph_edge::maybe_hot_p): Update.
21637 * cgraph.h: Include profile-count.h
21638 (create_clone), create_edge, create_indirect_edge): Update.
21639 (cgraph_node): Turn count to profile_count.
21640 (cgraph_edge0: Likewise.
21641 (make_speculative, clone): Update.
21642 (create_edge): Update.
21643 (init_lowered_empty_function): Update.
21644 * cgraphclones.c (cgraph_edge::clone): Update.
21645 (duplicate_thunk_for_node): Update.
21646 (cgraph_node::create_clone): Update.
21647 * cgraphunit.c (cgraph_node::analyze): Update.
21648 (cgraph_node::expand_thunk): Update.
21649 * final.c (dump_basic_block_info): Update.
21650 * gimple-streamer-in.c (input_bb): Update.
21651 * gimple-streamer-out.c (output_bb): Update.
21652 * graphite.c (print_global_statistics): Update.
21653 (print_graphite_scop_statistics): Update.
21654 * hsa-brig.c: Include basic-block.h.
21655 * hsa-dump.c: Include basic-block.h.
21656 * hsa-gen.c (T sum_slice): Update.
21657 (convert_switch_statements):Update.
21658 * hsa-regalloc.c: Include basic-block.h.
21659 * ipa-chkp.c (chkp_produce_thunks): Update.
21660 * ipa-cp.c (struct caller_statistics): Update.
21661 (init_caller_stats): Update.
21662 (gather_caller_stats): Update.
21663 (ipcp_cloning_candidate_p): Update.
21664 (good_cloning_opportunity_p): Update.
21665 (get_info_about_necessary_edges): Update.
21666 (dump_profile_updates): Update.
21667 (update_profiling_info): Update.
21668 (update_specialized_profile): Update.
21669 (perhaps_add_new_callers): Update.
21670 (decide_about_value): Update.
21671 (ipa_cp_c_finalize): Update.
21672 * ipa-devirt.c (struct odr_type_warn_count): Update.
21673 (struct decl_warn_count): Update.
21674 (struct final_warning_record): Update.
21675 (possible_polymorphic_call_targets): Update.
21676 (ipa_devirt): Update.
21677 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21678 * ipa-icf.c (sem_function::merge): Update.
21679 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21680 * ipa-inline.c (compute_uninlined_call_time): Update.
21681 (compute_inlined_call_time): Update.
21682 (want_inline_small_function_p): Update.
21683 (want_inline_self_recursive_call_p): Update.
21684 (edge_badness): Update.
21685 (lookup_recursive_calls): Update.
21686 (recursive_inlining): Update.
21687 (inline_small_functions): Update.
21688 (dump_overall_stats): Update.
21689 (dump_inline_stats): Update.
21690 * ipa-profile.c (ipa_profile_generate_summary): Update.
21691 (ipa_propagate_frequency): Update.
21692 (ipa_profile): Update.
21693 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21694 * ipa-utils.c (ipa_merge_profiles): Update.
21695 * loop-doloop.c (doloop_modify): Update.
21696 * loop-unroll.c (report_unroll): Update.
21697 (unroll_loop_runtime_iterations): Update.
21698 * lto-cgraph.c (lto_output_edge): Update.
21699 (lto_output_node): Update.
21700 (input_node): Update.
21701 (input_edge): Update.
21702 (merge_profile_summaries): Update.
21703 * lto-streamer-in.c (input_cfg): Update.
21704 * lto-streamer-out.c (output_cfg): Update.
21705 * mcf.c (create_fixup_graph): Update.
21706 (adjust_cfg_counts): Update.
21707 (sum_edge_counts): Update.
21708 * modulo-sched.c (sms_schedule): Update.
21709 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21710 * predict.c (maybe_hot_count_p): Update.
21711 (probably_never_executed): Update.
21712 (dump_prediction): Update.
21713 (combine_predictions_for_bb): Update.
21714 (propagate_freq): Update.
21715 (handle_missing_profiles): Update.
21716 (counts_to_freqs): Update.
21717 (rebuild_frequencies): Update.
21718 (force_edge_cold): Update.
21719 * predict.h: Include profile-count.h
21720 (maybe_hot_count_p, counts_to_freqs): UPdate.
21721 * print-rtl-function.c: Do not include cfg.h
21722 * print-rtl.c: Include basic-block.h
21723 * profile-count.c: New file.
21724 * profile-count.h: New file.
21725 * profile.c (is_edge_inconsistent): Update.
21726 (correct_negative_edge_counts): Update.
21727 (is_inconsistent): Update.
21728 (set_bb_counts): Update.
21729 (read_profile_edge_counts): Update.
21730 (compute_frequency_overlap): Update.
21731 (compute_branch_probabilities): Update; Initialize and deinitialize
21733 (branch_prob): Update.
21734 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21735 (edge_gcov_count): New.
21736 (bb_gcov_count): New.
21737 * shrink-wrap.c (try_shrink_wrapping): Update.
21738 * tracer.c (better_p): Update.
21739 * trans-mem.c (expand_transaction): Update.
21740 (ipa_tm_insert_irr_call): Update.
21741 (ipa_tm_insert_gettmclone_call): Update.
21742 * tree-call-cdce.c: Update.
21743 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21744 (gimple_duplicate_sese_tail): Update.
21745 (gimple_account_profile_record): Update.
21746 (execute_fixup_cfg): Update.
21747 * tree-inline.c (copy_bb): Update.
21748 (copy_edges_for_bb): Update.
21749 (initialize_cfun): Update.
21750 (freqs_to_counts): Update.
21751 (copy_cfg_body): Update.
21752 (expand_call_inline): Update.
21753 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21754 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21755 (try_unroll_loop_completely): Update.
21756 (try_peel_loop): Update.
21757 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21758 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21759 * tree-ssa-loop-split.c (connect_loops): Update.
21760 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21761 * tree-ssa-reassoc.c (branch_fixup): Update.
21762 * tree-ssa-tail-merge.c (replace_block_by): Update.
21763 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21764 (compute_path_counts): Update.
21765 (update_profile): Update.
21766 (recompute_probabilities): Update.
21767 (update_joiner_offpath_counts): Update.
21768 (estimated_freqs_path): Update.
21769 (freqs_to_counts_path): Update.
21770 (clear_counts_path): Update.
21771 (ssa_fix_duplicate_block_edges): Update.
21772 (duplicate_thread_path): Update.
21773 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21774 (struct switch_conv_info): Update.
21775 * tree-tailcall.c (decrease_profile): Update.
21776 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21777 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21778 * value-prof.c (check_counter): Update.
21779 (gimple_divmod_fixed_value): Update.
21780 (gimple_mod_pow2): Update.
21781 (gimple_mod_subtract): Update.
21782 (gimple_ic_transform): Update.
21783 (gimple_stringop_fixed_value): Update.
21784 * value-prof.h (gimple_ic): Update.
21786 2017-06-02 Carl Love <cel@us.ibm.com>
21788 * config/rs6000/rs6000-c: Add support for built-in functions
21789 vector double vec_doublee (vector signed int);
21790 vector double vec_doublee (vector unsigned int);
21791 vector double vec_doublee (vector float);
21792 vector double vec_doubleh (vector signed int);
21793 vector double vec_doubleh (vector unsigned int);
21794 vector double vec_doubleh (vector float);
21795 vector double vec_doublel (vector signed int);
21796 vector double vec_doublel (vector unsigned int);
21797 vector double vec_doublel (vector float);
21798 vector double vec_doubleo (vector signed int);
21799 vector double vec_doubleo (vector unsigned int);
21800 vector double vec_doubleo (vector float);.
21801 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21802 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21804 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21805 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21806 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21808 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21809 vec_doublel, vec_doubleh.
21810 * doc/extend.texi: Update the built-in documentation file for the
21811 new built-in functions.
21813 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21816 * ipa-inline-analysis.c (free_growth_caches): Set
21817 edge_removal_hook_holder to NULL after removing it.
21819 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21821 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21822 comparision with zero.
21824 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21825 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21826 for early expansion of vec_min and vec_max builtins.
21827 (builtin_function_type): Add min/max unsigned variants to those
21828 identified as having unsigned arguments.
21830 2017-06-02 Olivier Hainque <hainque@adacore.com>
21832 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21834 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21836 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21837 Use VALL_F16 iterator rather than VALL.
21839 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21841 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21842 Emit CBNZ inside loop when doing a strong exchange and comparing
21843 against zero. Generate the CC flags after the loop.
21845 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21847 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21848 (dl_section_ref): New.
21849 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21850 On AIX, append an expression to subtract the size of the
21851 section length to dl_section_ref.
21853 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21855 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21856 for early expansion of vector absolute builtins.
21858 2017-06-02 Richard Biener <rguenther@suse.de>
21860 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21861 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21863 2017-06-02 Richard Biener <rguenther@suse.de>
21865 PR tree-optimization/80948
21866 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21867 stmt order as well.
21869 2017-06-02 Richard Biener <rguenther@suse.de>
21871 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21873 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21874 for inductions as relevant.
21876 2017-06-02 Richard Biener <rguenther@suse.de>
21878 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21879 (vectorizable_induction): ... this. Remove dead code.
21881 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21883 * builtins. (expand_builtin_alloca): Remove second parameter and
21884 infer its value from the first parameter instead.
21885 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21887 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21889 PR rtl-optimization/80903
21890 * loop-doloop.c (add_test): Unshare sequence.
21892 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21894 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21896 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21898 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21900 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21901 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21902 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21903 (xlogue_layout::get_stub_rtx): Make static.
21904 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21905 (xlogue_layout::compute_stub_managed_regs): Rename to...
21906 (xlogue_layout::count_stub_managed_regs): ...this.
21907 (xlogue_layout::is_stub_managed_reg): New function.
21908 (xlogue_layout::m_stub_names): Rename to...
21909 (xlogue_layout::s_stub_names): ...this, make static.
21910 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21911 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21912 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21913 xlogue_layout::s_stub_names): Instantiate statics.
21914 (stub_managed_regs): Remove.
21915 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21916 (disable_call_ms2sysv_xlogues): Rename to...
21917 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21918 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21920 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21921 change after reload_completed.
21922 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21924 (ix86_expand_prologue): Likewise.
21925 (ix86_expand_epilogue): Likewise.
21926 (ix86_expand_split_stack_prologue): Likewise.
21927 (ix86_compute_frame_layout): Remove frame parameter ...
21928 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21929 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21931 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21932 (ix86_frame): Move from here ...
21933 * config/i386/i386.h (ix86_frame): ... to here.
21934 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21935 complete ix86_frame data structure instead. Remove some_ld_name.
21937 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21939 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21940 symbols that hold a DECL_VALUE_EXPR.
21942 2017-06-01 Martin Jambor <mjambor@suse.cz>
21944 PR tree-optimization/80898
21945 * tree-sra.c (process_subtree_disqualification): Removed.
21946 (disqualify_candidate): Do not acll
21947 process_subtree_disqualification.
21948 (subtree_mark_written_and_enqueue): New function.
21949 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21950 RHS has been disqualified and re-queue LHS if necessary. Apart
21951 from that, ignore disqualified RHS.
21953 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21955 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21956 address fetch for z10 or later.
21958 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21960 * config/arc/arc.md (tst_movb): Add guard when splitting.
21962 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21964 * config/arc/arc.c (arc_can_eliminate): Test against
21965 arc_frame_pointer_needed.
21967 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21969 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21970 to prevent store reordering.
21971 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21972 (type): Add block type.
21973 (stack_tie): Define special instruction to be used in
21976 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21978 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21979 constraint. It is not valid for the pattern.
21980 (noncommutative_binary_comparison): Likewise.
21982 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21984 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21987 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21989 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21990 be used by the reg-alloc.
21992 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21994 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21995 reg-alloc when having mul64 or mul32x16 instructions.
21996 (mulsidi3): Likewise.
21997 (umulsidi3): Likewise.
21998 (mulsi32x16): New pattern.
21999 (mulsi64): Likewise.
22000 (mulsidi64): Likewise.
22001 (umulsidi64): Likewise.
22002 (MUL32x16_REG): Define.
22003 (mul64_600): Use MUL32x16_REG.
22004 (mac64_600): Likewise.
22005 (umul64_600): Likewise.
22006 (umac64_600): Likewise.
22008 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22010 * config/arc/arc.md (mulsi3_700): Make it commutative.
22012 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22014 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22016 (*sign_extendsidi2_insn): Likewise for movstosw.
22018 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22020 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22021 the type of the input discriminant value. Convert the
22022 discriminant value of signedness vary.
22024 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22026 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22027 Add to -Wall section.
22029 2017-06-01 Richard Biener <rguenther@suse.de>
22031 PR middle-end/66313
22032 * fold-const.c (fold_plusminus_mult_expr): If the factored
22033 factor may be zero use a wrapping type for the inner operation.
22034 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22035 and handle moved defs.
22036 (process_assignment): Properly guard the unary op case. Return a
22037 tri-state indicating that moving the stmt before the call may allow
22038 to continue. Pass through to_move.
22039 (find_tail_calls): Handle moving unrelated defs before
22042 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22045 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22046 splitter result in the canonical way.
22048 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22050 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22051 also for 32bit target. Update insn attributes.
22052 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22054 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22056 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22057 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22058 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22059 (_mm_maskz_min_ss): New intrinsics.
22061 2017-05-31 Martin Liska <mliska@suse.cz>
22063 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22064 Change comment style to one we normally use.
22065 (vectorizable_reduction): Likewise.
22066 (vectorizable_induction): Likewise.
22067 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22068 (vectorizable_call): Likewise.
22069 (vectorizable_simd_clone_call): Likewise.
22070 (vectorizable_conversion): Likewise.
22071 (vectorizable_assignment): Likewise.
22072 (vectorizable_shift): Likewise.
22073 (vectorizable_operation): Likewise.
22074 (vectorizable_store): Likewise.
22075 (vectorizable_load): Likewise.
22076 * tree-vectorizer.h: Likewise.
22078 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22080 * passes.c (emergency_dump_function): New.
22081 * tree-pass.h (emergency_dump_function): Declare.
22082 * plugin.c (plugins_internal_error_function): Remove.
22083 * plugin.h (plugins_internal_error_function): Remove declaration.
22084 * toplev.c (internal_error_function): New static function. Use it...
22085 (general_init): ...here.
22087 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22089 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22090 (arc_rtx_costs): Add costs for new patterns.
22091 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22092 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22094 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22096 * tree-ssa-strlen.c (get_next_strinfo): New function.
22097 (get_stridx_plus_constant): Use it.
22098 (zero_length_string): Likewise.
22099 (adjust_related_strinfos): Likewise.
22100 (adjust_last_stmt): Likewise.
22102 2017-05-31 Richard Biener <rguenther@suse.de>
22105 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22106 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22108 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22110 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22111 loop_vinfo argument and use of dependence distance vectors.
22112 Check instead whether the two references differ only in their
22113 initial value and assume that they have the same alignment if the
22114 difference is a multiple of the vector alignment.
22115 (vect_analyze_data_refs_alignment): Update call accordingly.
22117 2017-05-31 Martin Liska <mliska@suse.cz>
22120 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22122 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22124 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22125 (create_intersect_range_checks): Move from ...
22126 * tree-data-ref.c (create_intersect_range_checks_index)
22127 (create_intersect_range_checks): ... to here.
22128 (create_runtime_alias_checks): New function factored from ...
22129 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22130 here. Call above function.
22131 * tree-data-ref.h (create_runtime_alias_checks): New function.
22133 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22135 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22136 segment length for dr_b and compute it in wide_int.
22138 2017-05-31 Richard Biener <rguenther@suse.de>
22140 PR tree-optimization/80906
22141 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22142 and pass through iv_map.
22143 (copy_bb_and_scalar_dependences): Adjust.
22144 (translate_pending_phi_nodes): Likewise.
22145 (copy_loop_close_phi_args): Handle code-generating IVs instead
22148 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22150 * diagnostic-color.c (color_dict): Add "type-diff".
22151 (parse_gcc_colors): Update comment.
22152 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22153 -fdiagnostics-show-template-tree and -fno-elide-type.
22154 (GCC_COLORS): Add type-diff to example.
22156 (-fdiagnostics-show-template-tree): New.
22157 (-fno-elide-type): New.
22158 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22159 the pp_format_decoder callback. Call any m_format_postprocessor's
22161 (pretty_printer::pretty_printer): Initialize
22162 m_format_postprocessor.
22163 (pretty_printer::~pretty_printer): Delete any
22164 m_format_postprocessor.
22165 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22166 (class format_postprocessor): New class.
22167 (struct pretty_printer::format_decoder): Document the new parameters.
22168 (struct pretty_printer::m_format_postprocessor): New field.
22169 * tree-diagnostic.c (default_tree_printer): Update for new
22170 bool and const char ** params.
22171 * tree-diagnostic.h (default_tree_printer): Likewise.
22173 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22175 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22176 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22177 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22178 rs6000_gen_cell_microcode code.
22179 (rs6000_final_prescan_insn): Delete.
22180 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22181 "warn-cell-microcode" entries.
22182 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22183 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22184 throughout. Change cc_reg_not_micro_cr0_operand to
22185 cc_reg_not_cr0_operand throughout.
22186 (*extendhi<mode>2_noload): Delete.
22187 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22188 (mwarn-cell-microcode): Delete.
22189 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22190 -mgen-cell-microcode and -mwarn-cell-microcode.
22192 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22195 * config/i386/constraints.md (Yd): New constraint.
22197 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22198 and (?Yd, r) alternatives. Update insn attributes.
22199 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22200 and (?*Yd, r) alternatives. Update insn attributes.
22201 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22203 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22205 * gimplify.c (gimplify_modify_expr): Don't create a
22206 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22209 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22211 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22213 2017-05-30 Richard Biener <rguenther@suse.de>
22215 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22216 and reduc_def fields.
22217 (STMT_VINFO_REDUC_TYPE): New define.
22218 (STMT_VINFO_REDUC_DEF): Likewise.
22219 (vect_force_simple_reduction): Adjust prototype.
22220 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22221 (vect_is_simple_reduction): Remove check_reduction argument.
22222 (vect_force_simple_reduction): Adjust and set
22223 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22224 (vectorizable_reduction): Do not re-do reduction analysis
22225 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22226 * tree-parloops.c (gather_scalar_reductions): Adjust.
22228 2017-05-30 Richard Biener <rguenther@suse.de>
22230 PR middle-end/80901
22231 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22234 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22236 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22237 Introduce unknown_misalignment parameter and remove vf.
22238 (vect_peeling_hash_get_lowest_cost):
22239 Pass unknown_misalignment parameter.
22240 (vect_enhance_data_refs_alignment):
22241 Fix unsupportable data ref treatment.
22243 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22245 * tree-vect-data-refs.c (vect_get_data_access_cost):
22246 Workaround for SLP handling.
22247 (vect_enhance_data_refs_alignment):
22248 Compute costs for doing no peeling at all, compare to the best
22249 peeling costs so far and avoid peeling if cheaper.
22251 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22253 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22254 Return peeling info and set costs to zero for unlimited cost
22256 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22257 with unknown misalignment. Compute and costs for unknown
22258 misalignment, compare them to the costs for known misalignment
22259 and choose the cheapest for peeling.
22261 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22263 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22264 (vect_get_peeling_costs_all_drs): Create function.
22265 (vect_peeling_hash_get_lowest_cost):
22266 Use vect_get_peeling_costs_all_drs.
22267 (vect_peeling_supportable): Create function.
22268 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22270 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22272 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22273 DR_HAS_NEGATIVE_STEP.
22274 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22275 (vect_enhance_data_refs_alignment): Use.
22276 (vect_duplicate_ssa_name_ptr_info): Use.
22277 * tree-vectorizer.h (dr_misalignment): Use.
22278 (known_alignment_for_access_p): Use.
22280 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22283 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22285 (has_section_name): New function.
22287 2017-05-30 Martin Liska <mliska@suse.cz>
22290 * auto-profile.c (get_function_decl_from_block): Fix
22293 2017-05-30 Richard Biener <rguenther@suse.de>
22295 PR middle-end/80876
22296 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22298 2017-05-30 Martin Liska <mliska@suse.cz>
22300 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22301 * dumpfile.h (struct dump_file_info): Remove ctors.
22303 2017-05-30 Martin Liska <mliska@suse.cz>
22305 * predict.def: Fix GNU coding style.
22307 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22309 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22310 Mark 'to' argument with ATTRIBUTE_UNUSED.
22312 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22314 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22315 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22316 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22319 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22321 * doc/install.texi (Options specification): Restore entry of
22322 --enable-sjlj-exceptions.
22324 2017-05-27 Michael Eager <eager@eagercon.com>
22327 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22329 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22331 * config/microblaze/microblaze.h
22332 (FIXED_REGISTERS): Update in macro.
22333 (CALL_USED_REGISTERS): Update in macro.
22335 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22337 * doc/install.texi: Add links to macOS binary distributions.
22339 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22343 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22345 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22347 2017-05-26 Martin Liska <mliska@suse.cz>
22349 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22351 2017-05-26 Martin Liska <mliska@suse.cz>
22353 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22354 always leading ';; '.
22355 (dump_bb_info): Likewise.
22356 (brief_dump_cfg): Likewise.
22357 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22358 * dumpfile.c: Remove usage of TDF_VERBOSE.
22359 * dumpfile.h (enum dump_kind): Likewise.
22360 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22361 * print-tree.c (debug_verbose): Remove.
22362 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22363 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22364 * tree-diagnostic.c (default_tree_printer): Replace
22365 TDF_DIAGNOSTIC with TDF_SLIM.
22367 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22369 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22370 in parameter loop, rather than loop_vinfo.
22371 (create_intersect_range_checks): Ditto.
22372 (vect_create_cond_for_alias_checks): Update call to above functions.
22374 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22376 PR tree-optimization/80815
22377 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22378 for merging runtime alias checks. Handle negative DR_STEPs.
22380 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22382 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22384 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22385 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22386 out code pruning runtime alias checks.
22387 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22388 factored out from above.
22389 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22391 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22393 (prune_runtime_alias_test_list): New decalaration.
22395 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22397 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22398 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22399 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22400 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22401 (operator==, comp_dr_with_seg_len_pair): Ditto.
22402 (vect_prune_runtime_alias_test_list): Ditto.
22404 2017-05-26 Martin Liska <mliska@suse.cz>
22407 * params.def: Bound partial-inlining-entry-probability param.
22409 2017-05-26 Marek Polacek <polacek@redhat.com>
22412 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22415 2017-05-26 Richard Biener <rguenther@suse.de>
22417 PR tree-optimization/80842
22418 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22421 2017-05-26 Richard Biener <rguenther@suse.de>
22423 PR tree-optimization/80844
22424 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22426 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22428 * doc/md.texi (Machine Constraints): Update x86 family
22429 machine constraints section to match 'config/i386/constraints.md'.
22431 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22433 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22435 2017-05-25 Nathan Sidwell <nathan@acm.org>
22437 * doc/invoke.texi (--enable-languages): Update documentation.
22439 2017-05-25 Martin Liska <mliska@suse.cz>
22441 * dumpfile.c: Add TDF_FOLDING.
22442 * dumpfile.h (enum dump_kind): Likewise.
22443 * genmatch.c (dt_simplify::gen_1): Use it.
22445 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22447 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22449 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22451 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22452 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22454 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22456 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22457 * match.pd (X == C): Rewrite it here.
22458 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22459 with_certain_nonzero_bits2): New predicates.
22460 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22462 2017-05-24 Nathan Sidwell <nathan@acm.org>
22464 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22467 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22470 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22472 * config/powerpcspe: New port. Files are copied from the rs6000
22473 port, with "rs6000" in filenames replaced by "powerpcspe".
22475 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22477 PR rtl-optimization/80754
22478 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22480 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22482 * config/sparc/sparc.md (length): Return the correct value for -mflat
22483 sibcalls to match output_sibcall.
22485 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22489 * config/rs6000/rs6000.c (struct machine_function): Add new field
22491 (rs6000_get_separate_components): Init that field, use it.
22492 (rs6000_components_for_bb): Use the field.
22494 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22496 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22498 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22500 PR middle-end/80823
22501 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22503 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22506 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22507 address against address_operand predicate.
22508 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22510 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22512 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22513 parameters passed indirectly.
22515 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22517 * config/i386/i386.md (*movdi_internal): Remove SSE4
22518 alternative 18 (?r, *v). Update insn attributes.
22519 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22520 Update insn attributes.
22521 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22522 Update insn attributes.
22523 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22524 alternative 1 (r, v). Remove isa attribute.
22525 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22526 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22527 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22529 2017-05-23 Tom de Vries <tom@codesourcery.com>
22531 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22534 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22536 * cgraphunit.c (symbol_table::process_new_functions): Update.
22537 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22538 (inline_generate_summary): Rename to ...
22539 (ipa_fn_summary_generate): ... this one.
22540 (inline_read_summary): Rename to ...
22541 (ipa_fn_summary_read): ... this one.
22542 (inline_write_summary): Rename to ...
22543 (ipa_fn_summary_write): ... this one.
22544 (inline_free_summary): Rename to ...
22545 (ipa_free_fn_summary): ... this one.
22546 (pass_data_local_fn_summary, pass_local_fn_summary,
22547 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22548 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22549 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22550 make_pass_ipa_fn_summary): New.
22551 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22552 inline_write_summary, inline_free_summary): Remove.
22553 (ipa_free_fn_summary) : New.
22554 * ipa-inline.c (ipa_inline): Update.
22555 (pass_ipa_inline): Do not generate summaries.
22556 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22558 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22559 and add pass_ipa_fn_summary.
22560 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22562 (make_pass_inline_parameters): Remove.
22564 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22566 * omp-low.c (struct omp_context): Remove "default_kind" member.
22569 * omp-offload.c (execute_oacc_device_lower): Remove the
22570 parallelism dimensions function attributes for unparallelized
22571 OpenACC kernels constructs.
22573 2017-05-23 Martin Liska <mliska@suse.cz>
22575 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22577 (cgraph_edge::make_speculative): Likewise.
22578 (cgraph_edge::resolve_speculation): Likewise.
22579 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22580 (cgraph_node::dump): Likewise.
22581 * cgraph.h: Likewise.
22582 * cgraphunit.c (analyze_functions): Likewise.
22583 (symbol_table::compile): Likewise.
22584 * ipa-cp.c (print_all_lattices): Likewise.
22585 (determine_versionability): Likewise.
22586 (initialize_node_lattices): Likewise.
22587 (ipcp_verify_propagated_values): Likewise.
22588 (estimate_local_effects): Likewise.
22589 (update_profiling_info): Likewise.
22590 (create_specialized_node): Likewise.
22591 (perhaps_add_new_callers): Likewise.
22592 (decide_about_value): Likewise.
22593 (decide_whether_version_node): Likewise.
22594 (identify_dead_nodes): Likewise.
22595 (ipcp_store_bits_results): Likewise.
22596 * ipa-devirt.c (dump_targets): Likewise.
22597 (ipa_devirt): Likewise.
22598 * ipa-icf.c (sem_item::dump): Likewise.
22599 (sem_function::equals): Likewise.
22600 (sem_variable::equals): Likewise.
22601 (sem_item_optimizer::read_section): Likewise.
22602 (sem_item_optimizer::execute): Likewise.
22603 (congruence_class::dump): Likewise.
22604 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22605 (dump_inline_summary): Likewise.
22606 (estimate_node_size_and_time): Likewise.
22607 (inline_analyze_function): Likewise.
22608 * ipa-inline-transform.c (inline_call): Likewise.
22609 * ipa-inline.c (report_inline_failed_reason): Likewise.
22610 (want_early_inline_function_p): Likewise.
22611 (edge_badness): Likewise.
22612 (update_edge_key): Likewise.
22613 (inline_small_functions): Likewise.
22614 * ipa-profile.c (ipa_profile): Likewise.
22615 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22616 (ipa_make_edge_direct_to_target): Likewise.
22617 (remove_described_reference): Likewise.
22618 (ipa_impossible_devirt_target): Likewise.
22619 (propagate_controlled_uses): Likewise.
22620 (ipa_print_node_params): Likewise.
22621 (ipcp_transform_function): Likewise.
22622 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22623 (propagate_pure_const): Likewise.
22624 * ipa-reference.c (generate_summary): Likewise.
22625 (read_write_all_from_decl): Likewise.
22626 (propagate): Likewise.
22627 (ipa_reference_read_optimization_summary): Likewise.
22628 * ipa-utils.c (ipa_merge_profiles): Likewise.
22629 * ipa.c (walk_polymorphic_call_targets): Likewise.
22630 (symbol_table::remove_unreachable_nodes): Likewise.
22631 (ipa_single_use): Likewise.
22632 * passes.c (execute_todo): Likewise.
22633 * predict.c (drop_profile): Likewise.
22634 * symtab.c (symtab_node::get_dump_name): New function.
22635 (symtab_node::dump_name): Likewise.
22636 (symtab_node::dump_asm_name): Likewise.
22637 (symtab_node::dump_references): Likewise.
22638 (symtab_node::dump_referring): Likewise.
22639 (symtab_node::dump_base): Likewise.
22640 (symtab_node::debug_symtab): Likewise.
22641 * tree-sra.c (convert_callers_for_node): Likewise.
22642 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22643 * value-prof.c (init_node_map): Likewise.
22645 2017-05-23 Martin Liska <mliska@suse.cz>
22647 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22648 and symtab_node::debug_symtab to symbol_table::debug.
22649 * cgraphunit.c (analyze_functions): Use the renamed function.
22650 (symbol_table::compile): Likewise.
22651 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22652 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22653 * passes.c (execute_todo): Likewise.
22654 * symtab.c (symbol_table::dump): New function.
22655 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22657 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22659 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22660 that nonconst implies exec.
22662 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22664 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22665 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22666 (inline_edge_summary_vec): Turn into ...
22667 (ipa_call_summaries): ... this one.
22668 (redirect_to_unreachable, edge_set_predicate,
22669 evaluate_properties_for_edge, inline_summary_alloc,
22670 reset_ipa_call_summary, reset_inline_summary,
22671 inline_summary_t::duplicate): Update.
22672 (inline_edge_duplication_hook): Turn to ...
22673 (ipa_call_summary_t::duplicate): ... this one.
22674 (inline_edge_removal_hook): Turn to ...
22675 (ipa_call_summary_t::remove): ... this one.
22676 (dump_inline_edge_summary): Turn to ...
22677 (dump_ipa_call_summary): ... this one.
22678 (estimate_function_body_sizes): Update.
22679 (inline_update_callee_summaries): Update.
22680 (remap_edge_change_prob): Update.
22681 (remap_edge_summaries): Update.
22682 (inline_merge_summary): Update.
22683 (do_estimate_edge_time): Update.
22684 (inline_generate_summary): Update.
22685 (inline_read_section): Update.
22686 (inline_read_summary): Update.
22687 (inline_free_summary): Update.
22688 * ipa-inline.c (can_inline_edge_p): Update.
22689 (compute_inlined_call_time): Update.
22690 (want_inline_small_function_p): Update.
22691 (edge_badness): Update.
22692 (early_inliner): Update.
22693 * ipa-inline.h (inline_edge_summary): Turn to ...
22694 (ipa_call_summary): ... this one.
22695 (ipa_call_summary_t): New class.
22696 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22697 (ipa_call_summaries): New.
22698 (inline_edge_summary): Remove.
22699 (estimate_edge_growth): Update.
22700 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22701 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22702 * ipa-split.c (execute_split_functions): Update.
22703 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22705 2017-05-23 Tom de Vries <tom@codesourcery.com>
22707 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22708 attributes): Document rdrand effective target.
22710 2017-05-23 Tom de Vries <tom@codesourcery.com>
22712 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22713 attributes): Sort alphabetically.
22715 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22717 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22719 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22722 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22723 V2DF/V2DI splat into two separate patterns, one that handles
22724 registers, and the other that only handles memory. Drop support
22725 for splatting from a GPR on ISA 2.07 and then splitting the
22726 splat into direct move and splat.
22727 (vsx_splat_<mode>_reg): Likewise.
22728 (vsx_splat_<mode>_mem): Likewise.
22730 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22732 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22734 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22736 PR middle-end/80809
22737 * omp-low.c (finish_taskreg_remap): New function.
22738 (finish_taskreg_scan): If unit size of ctx->record_type
22739 is non-constant, unshare the size expression and replace
22740 decls in it with possible outer var refs.
22742 PR middle-end/80809
22743 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22744 GOVD_SHARED rather than GOVD_PRIVATE with it.
22745 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22746 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22748 PR middle-end/80853
22749 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22750 as last argument to build_outer_var_ref for pointer bases of array
22751 section reductions.
22753 2017-05-19 Martin Sebor <msebor@redhat.com>
22755 * print-tree.c (print_node): Print DECL_READ_P flag.
22757 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22759 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22760 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22761 * cgraph.c: Likewise.
22762 * cgraphunit.c: Likewise.
22763 * gengtype.c: Likewise.
22764 * ipa-cp.c: Likewise.
22765 * ipa-devirt.c: Likewise.
22766 * ipa-icf.c: Likewise.
22767 * ipa-predicate.c: Likewise.
22768 * ipa-profile.c: Likewise.
22769 * ipa-prop.c: Likewise.
22770 * ipa-split.c: Likewise.
22772 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22773 edge_predicate_pool, dump_inline_hints,
22774 inline_summary::account_size_time, redirect_to_unreachable,
22775 edge_set_predicate, set_hint_predicate,
22776 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22777 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22778 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22779 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22780 ipa_call_summary_t::remove, initialize_growth_caches,
22781 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22782 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22783 mark_modified, unmodified_parm_1, unmodified_parm,
22784 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22785 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22786 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22787 will_be_nonconstant_predicate, record_modified_bb_info,
22788 get_minimal_bb, record_modified, param_change_prob,
22789 phi_result_unknown_predicate, predicate_for_phi_result,
22790 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22791 estimate_function_body_sizes, compute_inline_parameters,
22792 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22793 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22794 inline_update_callee_summaries, remap_edge_change_prob,
22795 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22796 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22797 inline_analyze_function, inline_summary_t::insert,
22798 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22799 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22800 inline_free_summary): Move to ipa-fnsummary.h
22801 (predicate_t): Remove.
22802 * ipa-fnsummary.c: New file.
22803 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22804 (enum inline_hints_vals, inline_hints, agg_position_info,
22805 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22806 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22807 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22808 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22809 inline_read_summary, inline_write_summary, inline_free_summary,
22810 inline_analyze_function, initialize_inline_failed,
22811 inline_merge_summary, inline_update_overall_summary,
22812 compute_inline_parameters): Move to ipa-fnsummary.h
22813 * ipa-fnsummary.h: New file.
22814 * ipa-inline-transform.h: Include ipa-inline.h.
22815 * ipa-inline.c: LIkewise.
22817 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22819 * ipa-inline.c (edge_badness): Use inlined_time instead of
22820 inline_summaries->get.
22822 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22824 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22826 2017-05-22 Nathan Sidwell <nathan@acm.org>
22828 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22829 (fdump-lang): Document 'raw' option.
22830 * dumpfile.h (TDI_tu): Delete.
22831 * dumpfile.c (dump_files): Remove translation-unit.
22832 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22834 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22836 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22837 command option from $(AWK) call.
22838 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22839 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22840 [FORMAT]: Remove handling of variable.
22841 * config/avr/t-multilib: Regenerate.
22843 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22845 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22847 (dump_inline_summary): Do not print self_time.
22848 (estimate_function_body_sizes): Do not set self_time.
22849 (compute_inline_parameters): Likewise.
22850 (inline_read_section, inline_write_summary): Do not stream self_time.
22851 * ipa-inline.h (inline_summary): Drop self_time.
22853 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22855 * ipa-inline-analysis.c (account_size_time): Rename to ...
22856 (inline_summary::account_size_time): ... this one.
22857 (reset_ipa_call_summary): Turn to ...
22858 (ipa_call_summary::reset): ... this one.
22859 (reset_inline_summary): Turn to ...
22860 (inline_summary::reset): ... this one.
22861 (inline_summary_t::remove): Update.
22862 (inline_summary_t::duplicate): Update.
22863 (ipa_call_summary_t::remove): Update.
22864 (dump_inline_summary): Update.
22865 (estimate_function_body_sizes): Update.
22866 (compute_inline_parameters): Update.
22867 (estimate_node_size_and_time): Update.
22868 (inline_merge_summary): Update.
22869 (inline_update_overall_summary): Update.
22870 (inline_read_section): Update.
22871 (inline_write_summary): Update.
22872 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22873 add account_size_time and reset member functions.
22874 (ipa_call_summary): Add reset function.
22875 * ipa-predicate.h (predicate::operator &): Constify.
22877 2017-05-22 Richard Biener <rguenther@suse.de>
22879 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22881 2017-05-19 Jason Merrill <jason@redhat.com>
22883 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22885 2017-05-19 Marek Polacek <polacek@redhat.com>
22888 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22889 TYPE_OVERFLOW_WRAPS checks.
22891 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22893 * tree-core.h (enum omp_clause_default_kind): Add
22894 "OMP_CLAUSE_DEFAULT_PRESENT".
22895 * tree-pretty-print.c (dump_omp_clause): Handle it.
22896 * gimplify.c (enum gimplify_omp_var_data): Add
22897 "GOVD_MAP_FORCE_PRESENT".
22898 (gimplify_adjust_omp_clauses_1): Map it to
22899 "GOMP_MAP_FORCE_PRESENT".
22900 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22902 * gimplify.c (oacc_default_clause): Clarify.
22904 2017-05-19 Nathan Sidwell <nathan@acm.org>
22906 LANG_HOOK_REGISTER_DUMPS
22907 * toplev.c (general_init): Call register dump lang hook.
22908 * doc/invoke.texi: Document -fdump-lang option family.
22909 * dumpfile.c (dump_files): Remove class dump here.
22910 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22911 * dumpfile.h (tree_dump_index): Remove TDI_class.
22912 * langhooks-def.h (lhd_register_dumps): Declare.
22913 (LANG_HOOKS_REGISTER_DUMPS): Define.
22914 (LANG_HOOKS_INITIALIZER): Add it.
22915 * langhooks.c (lhd_register_dumps): Define.
22916 * langhooks.h (struct lang_hooks): Add register_dumps.
22918 2017-05-19 Nathan Sidwell <nathan@acm.org>
22920 * context.h (context::set_passes): New.
22921 * context.c (context::context): Do not create pass manager.
22922 * toplev.c (general_init): Create pass manager here.
22924 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22926 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22927 use this splitter if two add or or instructions would also work for
22928 the constant we want to generate.
22930 2017-05-19 Richard Biener <rguenther@suse.de>
22933 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22934 predicate evaluation.
22936 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22938 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22940 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22943 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22945 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22946 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22947 (inline_edge_summary_vec): Turn into ...
22948 (ipa_call_summaries): ... this one.
22949 (redirect_to_unreachable, edge_set_predicate,
22950 evaluate_properties_for_edge, inline_summary_alloc,
22951 reset_ipa_call_summary, reset_inline_summary,
22952 inline_summary_t::duplicate): Update.
22953 (inline_edge_duplication_hook): Turn to ...
22954 (ipa_call_summary_t::duplicate): ... this one.
22955 (inline_edge_removal_hook): Turn to ...
22956 (ipa_call_summary_t::remove): ... this one.
22957 (dump_inline_edge_summary): Turn to ...
22958 (dump_ipa_call_summary): ... this one.
22959 (estimate_function_body_sizes): Update.
22960 (inline_update_callee_summaries): Update.
22961 (remap_edge_change_prob): Update.
22962 (remap_edge_summaries): Update.
22963 (inline_merge_summary): Update.
22964 (do_estimate_edge_time): Update.
22965 (inline_generate_summary): Update.
22966 (inline_read_section): Update.
22967 (inline_read_summary): Update.
22968 (inline_free_summary): Update.
22969 * ipa-inline.c (can_inline_edge_p): Update.
22970 (compute_inlined_call_time): Update.
22971 (want_inline_small_function_p): Update.
22972 (edge_badness): Update.
22973 (early_inliner): Update.
22974 * ipa-inline.h (inline_edge_summary): Turn to ...
22975 (ipa_call_summary): ... this one.
22976 (ipa_call_summary_t): New class.
22977 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22978 (ipa_call_summaries): New.
22979 (inline_edge_summary): Remove.
22980 (estimate_edge_growth): Update.
22981 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22982 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22983 * ipa-split.c (execute_split_functions): Update.
22984 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22986 2017-05-19 Richard Biener <rguenther@suse.de>
22988 PR middle-end/80764
22989 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22991 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22993 * config/rs6000/rs6000.c (struct machine_function): Add field
22994 fpr_is_wrapped_separately.
22995 (rs6000_get_separate_components): Use 64 components. Handle the
22996 new FPR components.
22997 (rs6000_components_for_bb): Handle the FPR components.
22998 (rs6000_emit_prologue_components): Handle the FPR components.
22999 (rs6000_emit_epilogue_components): Handle the FPR components.
23000 (rs6000_set_handled_components): Handle the FPR components.
23001 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23002 that are already separately shrink-wrapped.
23003 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23004 that are already separately shrink-wrapped.
23006 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23009 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23012 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23013 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23014 cases where the register allocator uses a move and an offsettable
23015 memory operation to/from a FPR register on ISA 2.06/2.07.
23016 (define_peephole2 for Altivec d-form store): Likewise.
23018 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23021 * config/i386/mmx.md (*mov<mode>_internal): Enable
23022 alternatives 11, 12, 13 and 14 also for 32bit targets.
23023 Remove alternatives 15, 16, 17 and 18.
23024 * config/i386/sse.md (vec_concatv2di): Change
23025 alternative (!x, *y) to (x, ?!*Yn).
23027 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23029 * dumpfile.h (enum dump_kind): Remove stray comma.
23031 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23033 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23034 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23035 predicate::num_conditions
23036 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23037 (CHANGED): turn into predicate::changed.
23038 (agg_position_info): Move to ipa-predicate.h
23039 (add_condition, predicate::add_clause, predicate::operator &=,
23040 predicate::or_with, predicate::evaluate, predicate::probability,
23041 dump_condition, dump_clause, predicate::dump,
23042 predicate::remap_after_duplication, predicate::remap_after_inlining,
23043 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23044 (evaluate_conditions_for_known_args): Update.
23045 (set_cond_stmt_execution_predicate): Update.
23046 * ipa-inline.h: Include ipa-predicate.h
23047 (condition, inline_param_summary, conditions, agg_position_info,
23048 predicate): Move to ipa-predicate.h
23049 * ipa-predicate.c: New file.
23050 * ipa-predicate.h: New file.
23052 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23054 * final.c (leaf_function_p): Check we are not in a sequence.
23056 2017-05-18 Martin Liska <mliska@suse.cz>
23058 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23059 * dumpfile.c (dump_register): Use new enum dump_kind.
23060 (get_dump_file_name): Likewise.
23061 (dump_enable_all): Likewise.
23062 (dump_switch_p_1): Likewise.
23063 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23064 * dumpfile.h (enum dump_kind): New enum type.
23065 (struct dump_file_info): Create constructor and
23066 format fields and comments.
23067 * passes.c (pass_manager::register_one_dump_file):
23069 * statistics.c (statistics_early_init): Likewise.
23070 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23071 TDF_TREE with TDF_SLIM.
23072 (gather_memory_references_ref): Likewise.
23074 2017-05-18 Martin Liska <mliska@suse.cz>
23076 * vec.h (struct vnull): Use it.
23078 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23080 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23081 (true_predicate, false_predicate, true_predicate_p,
23082 false_predicate_p): Remove.
23083 (single_cond_predicate, not_inlined_predicate): Turn to member function
23085 (add_condition): Update.
23086 (add_clause): Turn to...
23087 (predicate::add_clause): ... this one; update; allow passing NULL
23089 (and_predicates): Turn to ...
23090 (predicate::operator &=): ... this one.
23091 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23092 (or_predicates): Turn to ...
23093 (predicate::or_with): ... this one.
23094 (evaluate_predicate): Turn to ...
23095 (predicate::evaluate): ... this one.
23096 (predicate_probability): Turn to ...
23097 (predicate::probability): ... this one.
23098 (dump_condition): Update.
23099 (dump_predicate): Turn to ...
23100 (predicate::dump): ... this one.
23101 (account_size_time): Update.
23102 (edge_set_predicate): Update.
23103 (set_hint_predicate): UPdate.
23104 (evaluate_conditions_for_known_args): Update.
23105 (evaluate_properties_for_edge): Update.
23106 (remap_predicate_after_duplication): Turn to...
23107 (predicate::remap_after_duplication): ... this one.
23108 (remap_hint_predicate_after_duplication): Update.
23109 (inline_summary_t::duplicate): UPdate.
23110 (dump_inline_edge_summary): Update.
23111 (dump_inline_summary): Update.
23112 (set_cond_stmt_execution_predicate): Update.
23113 (set_switch_stmt_execution_predicate): Update.
23114 (compute_bb_predicates): Update.
23115 (will_be_nonconstant_expr_predicate): Update.
23116 (will_be_nonconstant_predicate): Update.
23117 (phi_result_unknown_predicate): Update.
23118 (predicate_for_phi_result): Update.
23119 (array_index_predicate): Update.
23120 (estimate_function_body_sizes): Update.
23121 (estimate_node_size_and_time): Update.
23122 (estimate_ipcp_clone_size_and_time): Update.
23123 (remap_predicate): Rename to ...
23124 (predicate::remap_after_inlining): ... this one.
23125 (remap_hint_predicate): Update.
23126 (inline_merge_summary): Update.
23127 (inline_update_overall_summary): Update.
23128 (estimate_size_after_inlining): Update.
23129 (read_predicate): Rename to ...
23130 (predicate::stream_in): ... this one.
23131 (read_inline_edge_summary): Update.
23132 (write_predicate): Rename to ...
23133 (predicate::stream_out): ... this one.
23134 (write_inline_edge_summary): Update.
23135 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23136 (clause_t): Turn to uint32_t
23137 (predicate): Turn to class; implement constructor and operators
23139 (size_time_entry): Update.
23140 (inline_summary): Update.
23141 (inline_edge_summary): Update.
23143 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23145 * fold-const.c (fold_binary_loc): Move transformation...
23146 * match.pd (C - X CMP X): ... here.
23148 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23150 * config/sparc/sparc.c (sparc_option_override): Set function
23151 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23152 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23154 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23156 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23158 2017-05-18 Marek Polacek <polacek@redhat.com>
23161 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23162 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23163 gimple_assign_load_p.
23165 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23167 PR middle-end/80692
23168 * real.c (do_compare): Give decimal_do_compare preference over
23169 comparing just the signs.
23171 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23173 * doc/md.texi (Canonicalization of Instructions): Describe the
23174 canonical form of instructions that inherently set a condition
23177 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23179 PR middle-end/80775
23180 * tree-cfg.c: Move deletion of unreachable case statements to after
23181 the merging of consecutive case labels.
23183 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23185 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23186 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23187 restoring of callee-saved registers.
23189 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23191 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23192 * config/visium/visium.c (single_set_and_flags): Likewise.
23193 * config/visium/visium.md (Substitutions): Likewise.
23195 2017-05-17 Martin Liska <mliska@suse.cz>
23197 * cfg.c: Introduce dump_flags_t type and
23198 use it instead of int type.
23200 * cfghooks.c: Likewise.
23201 * cfghooks.h (struct cfg_hooks): Likewise.
23202 * cfgrtl.c: Likewise.
23203 * cfgrtl.h: Likewise.
23204 * cgraph.c (cgraph_node::get_body): Likewise.
23205 * coretypes.h: Likewise.
23206 * domwalk.c: Likewise.
23207 * domwalk.h: Likewise.
23208 * dumpfile.c (struct dump_option_value_info): Likewise.
23209 (dump_enable_all): Likewise.
23210 (dump_switch_p_1): Likewise.
23211 (opt_info_switch_p): Likewise.
23212 * dumpfile.h (enum tree_dump_index): Likewise.
23213 (struct dump_file_info): Likewise.
23214 * genemit.c: Likewise.
23215 * generic-match-head.c: Likewise.
23216 * gengtype.c (open_base_files): Likewise.
23217 * gimple-pretty-print.c: Likewise.
23218 * gimple-pretty-print.h: Likewise.
23219 * graph.c (print_graph_cfg): Likewise.
23220 * graphite-scop-detection.c (dot_all_sese): Likewise.
23221 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23222 * loop-unroll.c (report_unroll): Likewise.
23223 * passes.c (pass_manager::register_one_dump_file): Likewise.
23224 * print-tree.c: Likewise.
23225 * statistics.c: Likewise.
23226 * tree-cfg.c: Likewise.
23227 * tree-cfg.h: Likewise.
23228 * tree-dfa.c: Likewise.
23229 * tree-dfa.h: Likewise.
23230 * tree-dump.c (dump_function): Likewise.
23231 * tree-dump.h (struct dump_info): Likewise.
23232 * tree-pretty-print.c: Likewise.
23233 * tree-pretty-print.h: Likewise.
23234 * tree-ssa-live.c: Likewise.
23235 * tree-ssa-live.h: Likewise.
23236 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23237 * tree-vect-loop.c: Likewise.
23238 * tree-vect-slp.c: Likewise.
23240 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23241 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23243 PR tree-optimization/80457
23244 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23245 of all arguments to a statement as scalar_to_vec operations.
23246 (vectorizable_call): Adjust call to vect_model_simple_cost for
23248 (vectorizable_conversion): Likewise.
23249 (vectorizable_assignment): Likewise.
23250 (vectorizable_shift): Likewise.
23251 (vectorizable_operation): Likewise.
23252 (vectorizable_comparison): Likewise.
23253 (vect_is_simple_cond): Record the def types for operands.
23254 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23255 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23256 for statement argument count.
23258 2017-05-16 Carl Love <cel@us.ibm.com>
23260 * config/rs6000/rs6000-c: Add support for built-in functions
23261 vector unsigned long long vec_bperm (vector unsigned long long,
23262 vector unsigned char)
23263 vector signed long long vec_mule (vector signed int,
23265 vector unsigned long long vec_mule (vector unsigned int,
23266 vector unsigned int)
23267 vector signed long long vec_mulo (vector signed int,
23269 vector unsigned long long vec_mulo (vector unsigned int,
23270 vector unsigned int)
23271 vector signed char vec_sldw (vector signed char,
23272 vector signed char,
23274 vector unsigned char vec_sldw (vector unsigned char,
23275 vector unsigned char,
23277 vector signed short vec_sldw (vector signed short,
23278 vector signed short,
23280 vector unsigned short vec_sldw (vector unsigned short,
23281 vector unsigned short,
23283 vector signed int vec_sldw (vector signed int,
23286 vector unsigned int vec_sldw (vector unsigned int,
23287 vector unsigned int,
23289 vector signed long long vec_sldw (vector signed long long,
23290 vector signed long long,
23292 vector unsigned long long vec_sldw (vector unsigned long long,
23293 vector unsigned long long,
23295 * config/rs6000/rs6000-c: Add support for built-in functions
23296 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23297 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23298 * doc/extend.texi: Update the built-in documentation for the
23299 new built-in functions.
23301 2017-05-16 Marek Polacek <polacek@redhat.com>
23305 * tree.c (save_expr): Don't fold the expression.
23307 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23309 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23310 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23311 and (?*y,m). Update insn attributes.
23313 2017-05-16 Martin Liska <mliska@suse.cz>
23315 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23316 flags argument of print_gimple_stmt, print_gimple_expr,
23317 print_generic_stmt and print_generic_expr.
23318 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23319 * coretypes.h: Likewise.
23320 * except.c (dump_eh_tree): Likewise.
23321 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23322 * gimple-pretty-print.h: Likewise.
23323 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23324 (backprop::push_to_worklist): Likewise.
23325 (backprop::pop_from_worklist): Likewise.
23326 (backprop::process_use): Likewise.
23327 (backprop::intersect_uses): Likewise.
23328 (note_replacement): Likewise.
23329 * gimple-ssa-store-merging.c
23330 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23331 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23332 (pass_store_merging::execute): Likewise.
23333 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23334 (ssa_base_cand_dump_callback): Likewise.
23335 (dump_incr_vec): Likewise.
23336 (replace_refs): Likewise.
23337 (replace_mult_candidate): Likewise.
23338 (create_add_on_incoming_edge): Likewise.
23339 (create_phi_basis): Likewise.
23340 (insert_initializers): Likewise.
23341 (all_phi_incrs_profitable): Likewise.
23342 (introduce_cast_before_cand): Likewise.
23343 (replace_one_candidate): Likewise.
23344 * gimplify.c (gimplify_expr): Likewise.
23345 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23346 (set_rename): Likewise.
23347 (rename_uses): Likewise.
23348 (copy_loop_phi_nodes): Likewise.
23349 (add_close_phis_to_merge_points): Likewise.
23350 (copy_loop_close_phi_args): Likewise.
23351 (copy_cond_phi_args): Likewise.
23352 (graphite_copy_stmts_from_block): Likewise.
23353 (translate_pending_phi_nodes): Likewise.
23354 * graphite-poly.c (print_pdr): Likewise.
23355 (dump_gbb_cases): Likewise.
23356 (dump_gbb_conditions): Likewise.
23357 (print_scop_params): Likewise.
23358 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23359 (build_cross_bb_scalars_use): Likewise.
23360 (gather_bbs::before_dom_children): Likewise.
23361 * hsa-dump.c (dump_hsa_immed): Likewise.
23362 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23363 (get_replacement_map): Likewise.
23364 * ipa-inline-analysis.c (dump_condition): Likewise.
23365 (estimate_function_body_sizes): Likewise.
23366 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23367 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23368 * ipa-prop.c (ipa_dump_param): Likewise.
23369 (ipa_print_node_jump_functions_for_edge): Likewise.
23370 (ipa_modify_call_arguments): Likewise.
23371 (ipa_modify_expr): Likewise.
23372 (ipa_dump_param_adjustments): Likewise.
23373 (ipa_dump_agg_replacement_values): Likewise.
23374 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23375 * ipa-pure-const.c (check_stmt): Likewise.
23376 (pass_nothrow::execute): Likewise.
23377 * ipa-split.c (execute_split_functions): Likewise.
23378 * omp-offload.c (dump_oacc_loop_part): Likewise.
23379 (dump_oacc_loop): Likewise.
23380 * trans-mem.c (tm_log_emit): Likewise.
23381 (tm_memopt_accumulate_memops): Likewise.
23382 (dump_tm_memopt_set): Likewise.
23383 (dump_tm_memopt_transform): Likewise.
23384 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23385 (print_loop): Likewise.
23386 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23387 (chkp_gather_checks_info): Likewise.
23388 (chkp_get_check_result): Likewise.
23389 (chkp_remove_check_if_pass): Likewise.
23390 (chkp_use_outer_bounds_if_possible): Likewise.
23391 (chkp_reduce_bounds_lifetime): Likewise.
23392 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23393 (chkp_mark_completed_bounds): Likewise.
23394 (chkp_register_incomplete_bounds): Likewise.
23395 (chkp_mark_invalid_bounds): Likewise.
23396 (chkp_maybe_copy_and_register_bounds): Likewise.
23397 (chkp_build_returned_bound): Likewise.
23398 (chkp_get_bound_for_parm): Likewise.
23399 (chkp_build_bndldx): Likewise.
23400 (chkp_get_bounds_by_definition): Likewise.
23401 (chkp_generate_extern_var_bounds): Likewise.
23402 (chkp_get_bounds_for_decl_addr): Likewise.
23403 * tree-chrec.c (chrec_apply): Likewise.
23404 * tree-data-ref.c (dump_data_reference): Likewise.
23405 (dump_subscript): Likewise.
23406 (dump_data_dependence_relation): Likewise.
23407 (analyze_overlapping_iterations): Likewise.
23408 * tree-inline.c (expand_call_inline): Likewise.
23409 (tree_function_versioning): Likewise.
23410 * tree-into-ssa.c (dump_defs_stack): Likewise.
23411 (dump_currdefs): Likewise.
23412 (dump_names_replaced_by): Likewise.
23413 (dump_update_ssa): Likewise.
23414 (update_ssa): Likewise.
23415 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23416 * tree-parloops.c (build_new_reduction): Likewise.
23417 (try_create_reduction_list): Likewise.
23418 (ref_conflicts_with_region): Likewise.
23419 (oacc_entry_exit_ok_1): Likewise.
23420 (oacc_entry_exit_single_gang): Likewise.
23421 * tree-pretty-print.h: Likewise.
23422 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23423 (get_scalar_evolution): Likewise.
23424 (add_to_evolution): Likewise.
23425 (get_loop_exit_condition): Likewise.
23426 (analyze_evolution_in_loop): Likewise.
23427 (analyze_initial_condition): Likewise.
23428 (analyze_scalar_evolution): Likewise.
23429 (instantiate_scev): Likewise.
23430 (number_of_latch_executions): Likewise.
23431 (gather_chrec_stats): Likewise.
23432 (final_value_replacement_loop): Likewise.
23433 (scev_const_prop): Likewise.
23434 * tree-sra.c (dump_access): Likewise.
23435 (disqualify_candidate): Likewise.
23436 (create_access): Likewise.
23437 (reject): Likewise.
23438 (maybe_add_sra_candidate): Likewise.
23439 (create_access_replacement): Likewise.
23440 (analyze_access_subtree): Likewise.
23441 (analyze_all_variable_accesses): Likewise.
23442 (sra_modify_assign): Likewise.
23443 (initialize_constant_pool_replacements): Likewise.
23444 (find_param_candidates): Likewise.
23445 (decide_one_param_reduction): Likewise.
23446 (replace_removed_params_ssa_names): Likewise.
23447 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23448 * tree-ssa-copy.c (dump_copy_of): Likewise.
23449 (copy_prop_visit_cond_stmt): Likewise.
23450 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23451 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23452 (record_equivalences_from_stmt): Likewise.
23453 * tree-ssa-dse.c (compute_trims): Likewise.
23454 (delete_dead_call): Likewise.
23455 (delete_dead_assignment): Likewise.
23456 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23457 (forward_propagate_into_cond): Likewise.
23458 (pass_forwprop::execute): Likewise.
23459 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23460 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23462 (move_computations_worker): Likewise.
23463 (execute_sm): Likewise.
23464 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23465 (remove_exits_and_undefined_stmts): Likewise.
23466 (remove_redundant_iv_tests): Likewise.
23467 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23468 (adjust_iv_update_pos): Likewise.
23469 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23470 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23471 (value_replacement): Likewise.
23472 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23473 * tree-ssa-pre.c (print_pre_expr): Likewise.
23474 (get_representative_for): Likewise.
23475 (create_expression_by_pieces): Likewise.
23476 (insert_into_preds_of_block): Likewise.
23477 (eliminate_insert): Likewise.
23478 (eliminate_dom_walker::before_dom_children): Likewise.
23479 (eliminate): Likewise.
23480 (remove_dead_inserted_code): Likewise.
23481 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23482 * tree-ssa-reassoc.c (get_rank): Likewise.
23483 (eliminate_duplicate_pair): Likewise.
23484 (eliminate_plus_minus_pair): Likewise.
23485 (eliminate_not_pairs): Likewise.
23486 (undistribute_ops_list): Likewise.
23487 (eliminate_redundant_comparison): Likewise.
23488 (update_range_test): Likewise.
23489 (optimize_range_tests_var_bound): Likewise.
23490 (optimize_vec_cond_expr): Likewise.
23491 (rewrite_expr_tree): Likewise.
23492 (rewrite_expr_tree_parallel): Likewise.
23493 (linearize_expr): Likewise.
23494 (break_up_subtract): Likewise.
23495 (linearize_expr_tree): Likewise.
23496 (attempt_builtin_powi): Likewise.
23497 (attempt_builtin_copysign): Likewise.
23498 (transform_stmt_to_copy): Likewise.
23499 (transform_stmt_to_multiply): Likewise.
23500 (dump_ops_vector): Likewise.
23501 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23502 (print_scc): Likewise.
23503 (set_ssa_val_to): Likewise.
23504 (visit_reference_op_store): Likewise.
23505 (visit_use): Likewise.
23506 (sccvn_dom_walker::before_dom_children): Likewise.
23507 (run_scc_vn): Likewise.
23508 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23510 (expr_hash_elt::print): Likewise.
23511 (const_and_copies::pop_to_marker): Likewise.
23512 (const_and_copies::record_const_or_copy_raw): Likewise.
23513 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23514 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23515 (dump_predicates): Likewise.
23516 (find_uninit_use): Likewise.
23517 (warn_uninitialized_phi): Likewise.
23518 (pass_late_warn_uninitialized::execute): Likewise.
23519 * tree-ssa.c (verify_vssa): Likewise.
23520 (verify_ssa): Likewise.
23521 (maybe_optimize_var): Likewise.
23522 * tree-vrp.c (dump_value_range): Likewise.
23523 (dump_all_value_ranges): Likewise.
23524 (dump_asserts_for): Likewise.
23525 (register_edge_assert_for_2): Likewise.
23526 (vrp_visit_cond_stmt): Likewise.
23527 (vrp_visit_switch_stmt): Likewise.
23528 (vrp_visit_stmt): Likewise.
23529 (vrp_visit_phi_node): Likewise.
23530 (simplify_cond_using_ranges_1): Likewise.
23531 (fold_predicate_in): Likewise.
23532 (evrp_dom_walker::before_dom_children): Likewise.
23533 (evrp_dom_walker::push_value_range): Likewise.
23534 (evrp_dom_walker::pop_value_range): Likewise.
23535 (execute_early_vrp): Likewise.
23537 2017-05-16 Richard Biener <rguenther@suse.de>
23539 * dwarf2out.c (loc_list_from_tree_1): Do not create
23540 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23542 2017-05-16 Richard Biener <rguenther@suse.de>
23544 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23546 (note_variable_value_in_expr): If we resolved the decl ref
23547 do not push to the stack.
23549 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23551 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23552 operations in fast-math mode.
23553 (vaddq_f16): Likewise.
23554 (vmul_f16): Likewise.
23555 (vmulq_f16): Likewise.
23556 (vsub_f16): Likewise.
23557 (vsubq_f16): Likewise.
23558 * config/arm/neon.md (add<mode>3): New.
23560 (fma:<VH:mode>3): New. Also remove outdated comment.
23563 2017-05-16 Martin Liska <mliska@suse.cz>
23567 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23568 (odr_types_equivalent_p): Likewise.
23570 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23572 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23574 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23577 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23578 non-interunit SSE move alternatives with '?'.
23579 (zero-extendsidi peephole2): New peephole to skip intermediate
23580 general register in SSE zero-extend sequence.
23582 2017-05-15 Jeff Law <law@redhat.com>
23584 * reorg.c (relax_delay_slots): Create a new variable to hold
23585 the temporary target rather than clobbering TARGET_LABEL.
23587 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23588 missing argument to extract_bit_field call.
23589 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23591 2017-05-15 Martin Liska <mliska@suse.cz>
23594 * gcc.c (process_command): Do not allow empty argument of -o option.
23596 2017-05-15 Renlin Li <renlin.li@arm.com>
23598 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23599 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23600 * config/aarch64/constraints.md (Usf): Add long call check.
23601 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23602 (call_value): Likewise.
23603 (sibcall): Likewise.
23604 (sibcall_value): Likewise.
23606 (call_value_insn): New.
23607 (sibcall_insn): Update rtx pattern.
23608 (sibcall_value_insn): Likewise.
23609 (call_internal): Remove.
23610 (call_value_internal): Likewise.
23611 (sibcall_internal): Likewise.
23612 (sibcall_value_internal): Likewise.
23613 (call_reg): Likewise.
23614 (call_symbol): Likewise.
23615 (call_value_reg): Likewise.
23616 (call_value_symbol): Likewise.
23618 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23621 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23623 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23625 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23626 compatible with CCGOCmode and with CCZmode.
23628 2017-05-14 Martin Sebor <msebor@redhat.com>
23630 PR middle-end/77671
23631 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23632 (gimple_fold_builtin_snprintf): Same.
23633 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23634 (gimple_fold_builtin_snprintf): Same.
23635 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23636 of character types.
23637 (is_call_safe): New function.
23638 (try_substitute_return_value): Call it.
23639 (try_simplify_call): New function.
23640 (pass_sprintf_length::handle_gimple_call): Call it.
23642 2017-05-14 Martin Sebor <msebor@redhat.com>
23644 PR middle-end/80669
23645 * builtins.c (expand_builtin_stpncpy): Simplify.
23647 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23649 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23650 * config/i386/i386.h
23651 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23652 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23653 (struct machine_function): Add new members call_ms2sysv,
23654 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23655 (struct machine_frame_state): New fields sp_realigned and
23656 sp_realigned_offset.
23657 * config/i386/i386.c
23658 (enum xlogue_stub): New enum.
23659 (enum xlogue_stub_sets): New enum.
23660 (class xlogue_layout): New class.
23661 (struct ix86_frame): New fields stack_realign_allocate_offset,
23662 stack_realign_offset and outlined_save_offset. Modify comments to
23663 detail stack layout when using out-of-line stubs.
23664 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23665 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23666 -mcall-ms2sysv-xlogues.
23667 (stub_managed_regs): New static variable.
23668 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23669 registers managed by out-of-line stub.
23670 (disable_call_ms2sysv_xlogues): New function.
23671 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23672 m->call_ms2sysv when appropriate and compute frame layout for
23674 (sp_valid_at, fp_valid_at): New inline functions.
23675 (choose_basereg): New function.
23676 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23678 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23679 Use align parameter of choose_baseaddr to generated aligned SSE movs
23681 (pro_epilogue_adjust_stack): Modify to track
23682 machine_frame_state::sp_realigned.
23683 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23684 (ix86_nsaved_sseregs): Likewise.
23685 (ix86_emit_save_regs): Likewise.
23686 (ix86_emit_save_regs_using_mov): Likewise.
23687 (ix86_emit_save_sse_regs_using_mov): Likewise.
23688 (get_scratch_register_on_entry): Likewise.
23689 (gen_frame_set): New function.
23690 (gen_frame_load): Likewise.
23691 (gen_frame_store): Likewise.
23692 (emit_outlined_ms2sysv_save): Likewise.
23693 (emit_outlined_ms2sysv_restore): Likewise.
23694 (ix86_expand_prologue): Modify stack re-alignment code and call
23695 emit_outlined_ms2sysv_save when appropriate.
23696 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23697 parameter rtx_insn *insn, which allows the function to be used to only
23698 generate the notes.
23699 (ix86_expand_epilogue): Modify validity checks of frame and stack
23700 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23701 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23702 * config/i386/predicates.md
23703 (save_multiple): New predicate.
23704 (restore_multiple): Likewise.
23705 * config/i386/sse.md
23706 (save_multiple<mode>): New pattern.
23707 (save_multiple_realign<mode>): Likewise.
23708 (restore_multiple<mode>): Likewise.
23709 (restore_multiple_and_return<mode>): Likewise.
23710 (restore_multiple_leave_return<mode>): Likewise.
23711 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23713 2017-05-14 Julia Koval <julia.koval@intel.com>
23715 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23716 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23717 (__builtin_ia32_xsetbv): New builtins.
23718 * config/i386/i386.c (ix86_expand_special_args_builtin):
23720 (ix86_expand_builtin): Special expand for new intrinsics.
23721 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23722 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23723 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23725 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23727 * cfganal.c (inverted_post_order_compute): Change argument type
23729 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23730 * df-core.c (rest_of_handle_df_initialize): Adjust.
23731 (rest_of_handle_df_finish): Likewise.
23732 (df_analyze_1): Likewise.
23733 (df_analyze): Likewise.
23734 (loop_inverted_post_order_compute): Change argument to be a vec *.
23735 (df_analyze_loop): Adjust.
23736 (df_get_n_blocks): Likewise.
23737 (df_get_postorder): Likewise.
23738 * df.h (struct df_d): Change field to be a vec.
23739 * lcm.c (compute_laterin): Adjust.
23740 (compute_available): Likewise.
23741 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23742 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23743 * tree-ssa-pre.c (compute_antic): Likewise.
23745 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23747 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23748 (depth_first_search::depth_first_search): Change structure init
23749 function to this constructor.
23750 (depth_first_search::add_bb): Rename function to this member.
23751 (depth_first_search::execute): Likewise.
23752 (flow_dfs_compute_reverse_finish): Adjust.
23754 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23756 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23757 (longest_simple_path): Likewise.
23758 * shrink-wrap.c (spread_components): Likewise.
23759 (disqualify_problematic_components): Likewise.
23760 (emit_common_heads_for_components): Likewise.
23761 (emit_common_tails_for_components): Likewise.
23762 (insert_prologue_epilogue_for_components): Likewise.
23764 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23766 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23769 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23771 * df-core.c (df_set_blocks): Start using auto_bitmap.
23772 (df_compact_blocks): Likewise.
23773 * df-problems.c (df_rd_confluence_n): Likewise.
23774 * df-scan.c (df_insn_rescan_all): Likewise.
23775 (df_process_deferred_rescans): Likewise.
23776 (df_update_entry_block_defs): Likewise.
23777 (df_update_exit_block_uses): Likewise.
23778 (df_entry_block_bitmap_verify): Likewise.
23779 (df_exit_block_bitmap_verify): Likewise.
23780 (df_scan_verify): Likewise.
23781 * lra-constraints.c (lra_constraints): Likewise.
23782 (undo_optional_reloads): Likewise.
23783 (lra_undo_inheritance): Likewise.
23784 * lra-remat.c (calculate_gen_cands): Likewise.
23785 (do_remat): Likewise.
23786 * lra-spills.c (assign_spill_hard_regs): Likewise.
23787 (spill_pseudos): Likewise.
23788 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23789 (bitmap_set_subtract_values): Likewise.
23791 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23793 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23794 management with auto_bitmap.
23795 (fix_inter_tick): Likewise.
23796 (fix_recovery_deps): Likewise.
23797 * ira.c (add_store_equivs): Likewise.
23798 (find_moveable_pseudos): Likewise.
23799 (split_live_ranges_for_shrink_wrap): Likewise.
23800 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23801 (rtx_reuse_manager::seen_def_p): Likewise.
23802 (rtx_reuse_manager::set_seen_def): Likewise.
23803 * print-rtl.h (class rtx_reuse_manager): Likewise.
23805 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23807 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23809 (migrate_btr_def): Likewise.
23810 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23811 * df-core.c (loop_post_order_compute): Likewise.
23812 (loop_inverted_post_order_compute): Likewise.
23813 * hsa-common.h: Likewise.
23814 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23815 * init-regs.c (initialize_uninitialized_regs): Likewise.
23816 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23817 (inline_small_functions): Likewise.
23818 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23819 * ira.c (combine_and_move_insns): Likewise.
23820 (build_insn_chain): Likewise.
23821 * loop-invariant.c (find_invariants): Likewise.
23822 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23823 * predict.c (tree_predict_by_opcode): Likewise.
23824 (predict_paths_leading_to): Likewise.
23825 (predict_paths_leading_to_edge): Likewise.
23826 (estimate_loops_at_level): Likewise.
23827 (estimate_loops): Likewise.
23828 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23829 (spread_components): Likewise.
23830 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23831 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23832 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23833 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23834 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23835 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23836 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23837 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23838 (mark_threaded_blocks): Likewise.
23839 (thread_through_all_blocks): Likewise.
23840 * tree-ssa.c (verify_ssa): Likewise.
23841 (execute_update_addresses_taken): Likewise.
23842 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23844 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23846 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23848 (post_order_compute): Likewise.
23849 (inverted_post_order_compute): Likewise.
23850 (pre_and_rev_post_order_compute_fn): Likewise.
23852 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23854 * genrecog.c (int_set::int_set): Explicitly construct our
23855 auto_vec base class.
23856 * vec.h (auto_vec::auto_vec): New constructor.
23858 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23860 * bitmap.h (class auto_bitmap): New constructor taking
23861 bitmap_obstack * argument.
23863 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23865 * bitmap.h (class auto_bitmap): Change type of m_bits to
23866 bitmap_head, and adjust ctor / dtor and member operators.
23868 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23870 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23871 when returned register mode doesn't match original mode.
23873 2017-05-12 Jeff Law <law@redhat.com>
23874 Jakub Jelinek <jakub@redhat.com>
23876 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23877 we look for cc setter after the compare-elim changes.
23878 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23879 within the vector to match what compare-elim now expects.
23880 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23881 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23883 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23884 after the compare-elim changes.
23885 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23886 the vector to match what compare-elim now expects.
23887 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23888 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23889 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23890 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23891 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23893 * config/visium/visium.c (single_set_and_flags): Fix where
23894 we look for cc setter after the compare-elim changes.
23895 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23896 with the vector to match what compare-elim now expects.
23897 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23898 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23899 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23900 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23901 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23902 (neg<mode>2_insn_set_overflow): Likewise.
23904 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23906 PR middle-end/79794
23907 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23908 maybe_expand_insn call, set ops[0].target. If still set after call,
23909 set alt_rtl. Add extra arg to recursive calls.
23910 (extract_bit_field): Add alt_rtl argument. Pass to
23912 * expmed.h (extract_bit_field): Fix prototype.
23913 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23914 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23915 to extract_bit_field_calls.
23916 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23917 Pass alt_rtl to extract_bit_field calls.
23918 * calls.c (store_unaligned_arguments_into_psuedos)
23919 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23920 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23922 * optabs.h (struct expand_operand): Add target bitfield.
23924 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23926 * compare-elim.c (try_eliminate_compare): Canonicalize
23927 operation with embedded compare to
23928 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23929 (set (reg) (operation)].
23931 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23933 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23936 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23937 cost of adding a carry flag for ADC instruction.
23938 [case MINUS]: Ignore the cost of subtracting a carry flag
23939 for SBB instruction.
23941 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23943 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23945 * config/rs6000/bmiintrin.h: New file.
23946 * config/rs6000/bmi2intrin.h: New file.
23947 * config/rs6000/x86intrin.h: New file.
23949 2017-05-12 Jeff Law <law@redhat.com>
23951 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23954 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23956 PR middle-end/80707
23957 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23959 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23961 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23962 early expansion of vector divide builtins.
23963 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23964 builtins identified as having unsigned arguments.
23966 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23968 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23969 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23970 expansion of vector logical operations (and, andc, or, xor,
23973 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23975 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23976 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23978 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23980 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23981 early GIMPLE expansion of vector multiplies.
23983 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23985 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23986 TARGET_HAVE_MOVT conditional.
23987 (movt splitter): Likewise.
23989 2017-05-12 Richard Biener <rguenther@suse.de>
23991 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23992 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23993 Fold all stmts not inplace.
23995 2017-05-12 Richard Biener <rguenther@suse.de>
23997 PR tree-optimization/80713
23998 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23999 inserted_exprs bit for not removed stmts.
24001 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24003 PR middle-end/69921
24004 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24005 parallelized" attribute for parallelized OpenACC kernels.
24006 * omp-offload.c (execute_oacc_device_lower): Use it.
24008 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24009 Set "oacc kernels" attribute.
24010 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24011 parameter. Adjust all users.
24012 (oacc_fn_attrib_kernels_p): Remove function.
24013 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24014 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24015 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24016 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24017 assert "oacc kernels" attribute is set.
24019 2017-05-11 Carl Love <cel@us.ibm.com>
24021 * config/rs6000/rs6000-c: Add support for built-in functions
24022 vector unsigned char vec_popcnt (vector signed char)
24023 vector unsigned char vec_popcnt (vector unsigned char)
24024 vector unsigned short vec_popcnt (vector signed short)
24025 vector unsigned short vec_popcnt (vector unsigned short)
24026 vector unsigned int vec_popcnt (vector signed int)
24027 vector unsigned int vec_popcnt (vector unsigned int)
24028 vector unsigned long long vec_popcnt (vector signed long long)
24029 vector unsigned long long vec_popcnt (vector unsigned long long)
24030 vector signed long long vec_slo (vector signed long long,
24031 vector signed char)
24032 vector signed long long vec_slo (vector signed long long,
24033 vector unsigned char)
24034 vector unsigned long long vec_slo (vector unsigned long long,
24035 vector signed char)
24036 vector unsigned long long vec_slo (vector unsigned long long,
24037 vector unsigned char)
24038 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24039 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24040 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24041 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24042 * doc/extend.texi: Update the built-in documentation file for the
24043 new built-in functions.
24045 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24047 * attribs.h (sorted_attr_string): Move machine independent
24048 functions for target clone support from the i386 port to common
24049 code. Rename ix86_function_versions to common_function_versions.
24050 Rename make_name to make_unique_name.
24051 (common_function_versions): Likewise.
24052 (make_unique_name): Likewise.
24053 (make_dispatcher_decl): Likewise.
24054 (is_function_default_version): Likewise.
24055 * attribs.c (attr_strcmp): Likewise.
24056 (sorted_attr_string): Likewise.
24057 (common_function_versions): Likewise.
24058 (make_unique_name): Likewise.
24059 (make_dispatcher_decl): Likewise.
24060 (is_function_default_version): Likewise.
24061 * config/i386/i386.c (attr_strcmp): Likewise.
24062 (sorted_attr_string): Likewise.
24063 (ix86_function_versions): Likewise.
24064 (make_name): Likewise.
24065 (make_dispatcher_decl): Likewise.
24066 (is_function_default_version): Likewise.
24067 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24069 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24072 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24073 Account for direct move costs for vec_construct of integer
24076 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24079 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24080 (UNSPEC_STX_ATOMIC): Ditto.
24081 (loaddi_via_sse): New insn.
24082 (storedi_via_sse): Ditto.
24083 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24084 Update corresponding peephole2 patterns.
24085 (atomic_storedi_fpu): Ditto.
24087 2017-05-11 Julia Koval <julia.koval@intel.com>
24089 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24090 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24092 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24093 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24094 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24096 2017-05-11 Nathan Sidwell <nathan@acm.org>
24098 * graphite-poly.c: Include dumpfile.h.
24100 * dumpfle.h (dump_function): Declare here ...
24101 * tree-dump.h (dump_function): ... not here.
24102 * dumpfile.c: #include tree-cfg.h.
24103 (dump_function): Move here from ...
24104 * tree-dump.c (dump_function): ... here.
24105 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24106 * graphite-poly.c: Don't include tree-dump.h.
24107 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24108 * print-tree.c: Likewise.
24109 * stor-layout.c: Likewise.
24110 * tree-nested.c: Likewise.
24112 * dumpfile.c (dump_start): Use TDF_FLAGS.
24113 (dump_enable_all): Fix TDF_KIND check thinko.
24115 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24117 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24118 array entries to represent two legal parameterizations of the
24119 overloaded __builtin_cmpb function, as represented by the
24120 P6_OV_BUILTIN_CMPB constant.
24121 (altivec_resolve_overloaded_builtin): Add special case handling
24122 for the __builtin_cmpb function, as represented by the
24123 P6_OV_BUILTIN_CMPB constant.
24124 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24125 (BU_P6_64BIT_2): New macro.
24126 (BU_P6_OVERLOAD_2): New macro
24127 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24128 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24129 (CMPB): Add overload support to represent both 32-bit and 64-bit
24130 compare-bytes function.
24131 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24132 support for TARGET_CMPB.
24133 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24134 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24135 documentation of the __builtin_cmpb overloaded built-in function.
24137 2017-05-11 Richard Biener <rguenther@suse.de>
24139 PR tree-optimization/80705
24140 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24141 bases are not vectorizable.
24143 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24145 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24146 when counting register pressure.
24148 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24150 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24151 (iv_ca_more_deps): Renamed to ...
24152 (iv_ca_compare_deps): ... this.
24153 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24155 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24157 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24159 (determine_group_iv_costs): ... here.
24160 (find_inv_vars_cb): Record inv var if it's not recorded before.
24162 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24164 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24165 (get_shiftadd_cost): Ditto.
24167 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24169 * tree-ssa-address.c: Include header file.
24170 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24172 (add_to_parts): Refactor.
24173 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24174 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24177 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24179 PR tree-optimization/53090
24180 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24182 (extract_cond_operands): Detect condition with IV on both sides
24183 and return COMP_IV_EXPR_2.
24184 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24185 (rewrite_use_compare): Simplify by removing call to function
24186 extract_cond_operands.
24188 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24190 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24191 (extract_cond_operands): Detect condition comparing against non-
24192 invariant bound and return appropriate enum value.
24193 (find_interesting_uses_cond): Update use of extract_cond_operands.
24194 Handle its return value accordingly.
24195 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24197 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24199 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24200 nonlinear iv_use computation in loop invariant sensitive way.
24202 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24204 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24205 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24207 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24209 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24210 (dump_cand): Support iv_cand.inv_exprs.
24211 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24213 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24216 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24218 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24220 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24221 as local function. Include necessary header files.
24222 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24224 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24226 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24228 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24230 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24231 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24232 RSHIFT_EXPR and BIT_NOT_EXPR.
24234 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24236 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24237 (adjust_setup_cost): New parameter supporting round up adjustment.
24238 (struct address_cost_data): Delete.
24239 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24240 (split_address_cost, ptr_difference_cost): Delete.
24241 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24242 (struct ainc_cost_data): New struct.
24243 (get_address_cost_ainc): New function.
24244 (get_address_cost, get_computation_cost): Reimplement.
24245 (determine_group_iv_cost_address): Record inv_expr for all uses of
24247 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24248 (iv_ca_has_deps): Reimplemented to ...
24249 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24251 (iv_ca_extend): Call iv_ca_more_deps.
24253 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24255 * tree-ssa-address.c (struct mem_address): Move to header file.
24256 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24257 * tree-ssa-address.h (struct mem_address): Move from C file.
24258 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24260 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24262 * tree-affine.h (aff_combination_type): New interface.
24263 (aff_combination_zero_p): Remove static.
24264 (aff_combination_const_p): New interface.
24265 (aff_combination_singleton_var_p): New interfaces.
24267 2017-05-11 Richard Biener <rguenther@suse.de>
24269 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24270 Skip unreachable blocks and destinations.
24271 (eliminate): Move stmt removal and fixup ...
24272 (fini_eliminate): ... here. Skip inserted exprs.
24273 (pass_pre::execute): Move fini_pre after fini_eliminate.
24274 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24275 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24276 PRE to get rid of dead code that has invalid SSA form and
24277 split critical edges again.
24279 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24281 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24283 2017-05-11 Richard Biener <rguenther@suse.de>
24285 * passes.c (execute_function_todo): Verify loops if they are
24286 said to be up-to-date.
24287 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24288 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24290 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24293 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24294 handle calling assemble_external ourself.
24297 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24298 modes with zero size. Enhance comment.
24300 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24302 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24303 built-ins for vec_xl and vec_xst with short and char pointer
24306 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24308 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24309 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24310 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24311 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24312 (_mm_maskz_min_round_ss): New intrinsics.
24313 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24314 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24315 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24316 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24317 (__builtin_ia32_minss_mask_round): New builtins.
24318 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24319 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24320 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24322 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24323 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24325 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24328 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24330 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24331 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24332 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24333 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24334 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24335 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24336 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24337 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24338 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24339 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24340 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24341 (__builtin_ia32_mulss_mask_round): New builtins.
24342 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24343 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24344 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24346 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24347 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24349 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24352 2017-05-10 Julia Koval <julia.koval@intel.com>
24354 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24355 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24356 (_mm256_setr_m128i): New intrinsics.
24358 2017-05-10 Julia Koval <julia.koval@intel.com>
24360 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24361 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24362 (_mm_maskz_rcp14_ss): New intrinsics.
24363 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24364 (__builtin_ia32_rcp14ss_mask): New builtins.
24365 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24367 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24369 PR tree-optimization/51513
24370 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24371 (assert_unreachable_fallthru_edge_p): Use it.
24372 (group_case_labels_stmt): Likewise.
24373 * tree-cfg.h: Prototype it.
24374 * stmt.c: Include cfghooks.h and tree-cfg.h.
24375 (emit_case_dispatch_table) <gap_label>: New local variable.
24376 Use it to fill dispatch table gaps.
24377 Test for default_label before updating probabilities.
24378 (expand_case) <default_label>: Remove unneeded initialization.
24379 Test for unreachable default case statement and remove its edge.
24380 Set default_label accordingly.
24381 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24383 2017-05-10 Carl Love <cel@us.ibm.com>
24385 * config/rs6000/rs6000-c: Add support for built-in functions
24386 vector signed char vec_neg (vector signed char)
24387 vector signed short int vec_neg (vector short int)
24388 vector signed int vec_neg (vector signed int)
24389 vector signed long long vec_neg (vector signed long long)
24390 vector float vec_neg (vector float)
24391 vector double vec_neg (vector double)
24392 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24394 * config/rs6000/altivec.h: Add define for vec_neg
24395 * doc/extend.texi: Update the built-in documentation for the
24396 new built-in functions.
24398 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24400 PR tree-optimization/77644
24401 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24403 2017-05-10 Nathan Sidwell <nathan@acm.org>
24405 * dumpfile.h (TDI_lang_all): New.
24406 (TDF_KIND): New. Renumber others
24407 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24409 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24411 (get_dump_file_name): Adjust suffix generation.
24412 (dump_enable_all): Use TDF_KIND.
24413 * doc/invoke.texi (-fdump-lang-all): Document.
24415 * dumpfile.h: Tabify.
24417 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24420 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24421 Move member access before delete.
24423 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24425 * tree-inline.c (expand_call_inline): Split block at stmt
24428 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24431 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24432 are now unused after splitting mov{sf,sd}_hardfloat.
24433 (f32_lr2): Likewise.
24434 (f32_lm): Likewise.
24435 (f32_lm2): Likewise.
24436 (f32_li): Likewise.
24437 (f32_li2): Likewise.
24438 (f32_lv): Likewise.
24439 (f32_sr): Likewise.
24440 (f32_sr2): Likewise.
24441 (f32_sm): Likewise.
24442 (f32_sm2): Likewise.
24443 (f32_si): Likewise.
24444 (f32_si2): Likewise.
24445 (f32_sv): Likewise.
24446 (f32_dm): Likewise.
24447 (f32_vsx): Likewise.
24448 (f32_av): Likewise.
24449 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24450 For movsf, order stores so the VSX stores occur before the GPR
24451 store which encourages the register allocator to use a traditional
24452 FPR instead of a GPR. For movsd, order the stores so that the GPR
24453 store comes before the VSX stores to allow the power6 to work.
24454 This is due to the power6 not having a 32-bit integer store
24455 instruction from a FPR.
24456 (movsf_hardfloat): Likewise.
24457 (movsd_hardfloat): Likewise.
24459 2017-05-09 Martin Sebor <msebor@redhat.com>
24461 PR translation/80280
24462 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24465 PR translation/80280
24466 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24467 data member added in r247778.
24468 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24470 2017-05-09 Nathan Sidwell <nathan@acm.org>
24472 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24474 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24477 2017-05-09 Marek Polacek <polacek@redhat.com>
24479 * doc/invoke.texi: Fix typo.
24481 2017-05-09 Richard Biener <rguenther@suse.de>
24483 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24484 (vrp_val_is_min): Likewise.
24485 (set_value_range_to_value): Likewise.
24486 (set_value_range_to_nonnegative): Likewise.
24487 (gimple_assign_nonzero_p): Likewise.
24488 (gimple_stmt_nonzero_p): Likewise.
24489 (vrp_int_const_binop): Likewise. Remove unreachable case.
24490 (adjust_range_with_scev): Adjust comments.
24491 (compare_range_with_value): Likewise.
24492 (extract_range_from_phi_node): Likewise.
24493 (test_for_singularity): Likewise.
24495 2017-05-09 Richard Biener <rguenther@suse.de>
24497 * tree-vrp.c (get_single_symbol): Add assert that we don't
24498 get overflowed constants as invariant part.
24499 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24500 checks. Use wi::cmp instead of recursing for integer constants.
24501 (compare_values): Just ignore whether we assumed undefined
24502 overflow instead of failing the compare.
24503 (extract_range_for_var_from_comparison_expr): Add comment before the
24504 TREE_NO_WARNING sets.
24505 (test_for_singularity): Likewise.
24506 (extract_range_from_comparison): Do not disable optimization
24507 when we assumed undefined overflow.
24508 (extract_range_basic): Remove init of unused var.
24510 2017-05-09 Richard Biener <rguenther@suse.de>
24512 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24513 (extract_range_from_multiplicative_op_1): Adjust.
24514 (extract_range_from_binary_expr_1): Use int_const_binop.
24516 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24519 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24520 rs6000_store_data_bypass_p in seven define_bypass directives and
24521 in several comments.
24522 * config/rs6000/rs6000-protos.h: Add prototype for
24523 rs6000_store_data_bypass_p function.
24524 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24525 function implements slightly different (rs6000-specific) semantics
24526 than store_data_bypass_p, returning false rather than aborting
24527 with assertion error when arguments do not satisfy the
24528 requirements of store data bypass.
24529 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24530 rs6000_store_data_bypass_p.
24532 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24534 * config/xtensa/xtensa-protos.h
24535 (xtensa_initial_elimination_offset): New declaration.
24536 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24537 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24538 macro definition, add case for FRAME_POINTER_REGNUM when
24539 FRAME_GROWS_DOWNWARD.
24540 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24541 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24542 xtensa_initial_elimination_offset.
24544 2017-05-08 Nathan Sidwell <nathan@acm.org>
24546 * doc/invoke.texi: Alphabetize -fdump options.
24548 2017-05-08 Martin Sebor <msebor@redhat.com>
24550 PR translation/80280
24551 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24553 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24555 * target.def (compute_frame_layout): New optional target hook.
24556 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24557 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24558 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24560 * reload1.c (verify_initial_elim_offsets): Likewise.
24561 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24562 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24563 (arm_get_frame_offsets): Split up into this ...
24564 (arm_compute_frame_layout): ... and this function.
24566 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24568 * config/aarch64/constraints.md (Usa): New constraint.
24569 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24571 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24573 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24574 with_multilib_list after it has been checked.
24576 2017-05-08 Richard Biener <rguenther@suse.de>
24578 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24579 (bitmap_set_subtract_values): Likewise.
24581 2017-05-08 Richard Biener <rguenther@suse.de>
24583 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24584 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24586 (gimple_stmt_nonzero_warnv_p): Rename to ...
24587 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24589 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24590 (extract_range_basic): Adjust, do not disable propagation on
24591 strict overflow sensitive simplification.
24592 (vrp_visit_cond_stmt): Likewise.
24594 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24596 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24597 body size unconditionally.
24599 2017-05-07 Jeff Law <law@redhat.com>
24602 2017-05-06 Jeff Law <law@redhat.com>
24603 PR tree-optimization/78496
24604 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24607 PR tree-optimization/78496
24608 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24609 (simplify_stmt_using_ranges): Call it.
24610 (vrp_dom_walker::before_dom_children): Extract equivalences
24611 from an ASSERT_EXPR with an equality comparison against a
24614 2017-05-06 Jeff Law <law@redhat.com>
24616 PR tree-optimization/78496
24617 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24620 PR tree-optimization/78496
24621 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24622 (simplify_stmt_using_ranges): Call it.
24623 (vrp_dom_walker::before_dom_children): Extract equivalences
24624 from an ASSERT_EXPR with an equality comparison against a
24627 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24629 * lra-constraints.c (lra_copy_reg_equiv): New function.
24630 (split_reg): Use it to copy equivalence information from the
24631 original register to the spill register.
24633 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24635 PR rtl-optimization/75964
24636 * simplify-rtx.c (simplify_const_relational_operation): Remove
24637 invalid handling of comparisons of integer ABS.
24639 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24641 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24642 initialize to zero.
24643 (init_regs): Remove declaration.
24644 (function_arg_advance_32): Initialize error_p as boolean variable.
24646 2017-05-05 Nathan Sidwell <nathan@acm.org>
24648 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24649 lines. Use for (;;).
24651 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24653 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24654 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24655 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24656 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24657 VF=2 that require versioning.
24659 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24661 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24664 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24666 * diagnostic.h (diagnostic_override_option_index): Convert from
24667 macro to inline function.
24669 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24671 * diagnostic.c (last_module_changed_p): New function.
24672 (set_last_module): New function.
24673 (diagnostic_report_current_module): Convert macro usage to
24674 the above functions.
24675 * diagnostic.h (diagnostic_context::last_module): Strengthen
24676 from const line_map * to const line_map_ordinary *.
24677 (diagnostic_last_module_changed): Delete macro.
24678 (diagnostic_set_last_module): Delete macro.
24680 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24682 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24683 with diagnostic_report_diagnostic.
24684 (diagnostic_n_impl_richloc): Likewise.
24685 * diagnostic.h (report_diagnostic): Delete macro.
24686 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24687 with diagnostic_report_diagnostic.
24688 * substring-locations.c (format_warning_va): Likewise.
24690 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24692 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24693 save/restor of format_spec. Move option-printing code to...
24694 (print_option_information): ...this new function, and
24695 reimplement by simply printing to the pretty_printer,
24696 rather than appending to the format string.
24698 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24700 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24701 handling logic into...
24702 (update_effective_level_from_pragmas): ...this new function.
24704 2017-05-04 Andrew Waterman <andrew@sifive.com>
24706 * config/riscv/riscv.opt (mstrict-align): New option.
24707 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24708 (SLOW_UNALIGNED_ACCESS): Define.
24709 (riscv_slow_unaligned_access): Declare.
24710 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24712 (riscv_slow_unaligned_access): New variable.
24713 (rocket_tune_info): Set slow_unaligned_access to true.
24714 (optimize_size_tune_info): Set slow_unaligned_access to false.
24715 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24716 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24717 (riscv_option_override): Set riscv_slow_unaligned_access.
24718 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24720 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24722 * config/riscv/riscv.md: Unify indentation.
24724 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24729 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24731 (extendsi<mode>2): Add support for doing sign extension via
24732 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24733 don't have ISA 3.0 instructions.
24734 (extendsi<mode>2 splitter): Likewise.
24735 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24736 generate the normal insns since SImode can now go in vector
24737 registers. Disallow the special UNSPECs needed for previous
24738 machines to hide SImode being used. Add new insns
24739 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24740 (fix_trunc<mode>si2_stfiwx): Likewise.
24741 (fix_trunc<mode>si2_internal): Likewise.
24742 (fixuns_trunc<mode>si2): Likewise.
24743 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24744 (fctiw<u>z_<mode>_smallint): Likewise.
24745 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24746 of floating point to 32-bit integer from doing a direct move to
24747 the GPR registers to do a store.
24748 (fctiwz_<mode>): Break long line.
24750 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24752 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24753 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24754 (addr_list, addr_offset_valid_p): New.
24755 (split_address_groups): Check offset validity with above function.
24756 (gt-tree-ssa-loop-ivopts.h): Include header file.
24758 2017-05-05 Nathan Sidwell <nathan@acm.org>
24760 * config.gcc (arm*-*-*): Add missing 'fi'.
24762 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24764 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24765 included in -fopt-info does not matter.
24766 * doc/optinfo.texi (-fopt-info): Fix description of default
24767 behavour. Explicitly say order of options included in -fopt-info
24770 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24772 * config.gcc: Allow combinations of aprofile and rmprofile values for
24773 --with-multilib-list.
24774 * config/arm/t-multilib: New file.
24775 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24776 variables. Remove setting of ISA and floating-point ABI in
24777 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24778 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24779 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24780 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24782 * config/arm/t-rmprofile: Likewise except for the matches changes.
24783 * doc/install.texi (--with-multilib-list): Document the combination of
24784 aprofile and rmprofile values and warn about pitfalls in doing that.
24786 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24788 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24789 (movdi_aarch64): Likewise.
24791 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24793 PR tree-optimization/80632
24794 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24796 (build_arrays): Initialize it for virtual phis.
24797 (fix_phi_nodes): Use it for virtual phis.
24799 PR tree-optimization/80558
24800 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24801 [x, y] op z into [x op, y op z] for op & or | if conditions
24804 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24805 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24808 * config/arm/arm.md (use_literal_pool): Remove.
24809 (64-bit immediate split): No longer takes cost into consideration
24810 if arm_disable_literal_pool is enabled.
24811 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24812 used when arm_disable_literal_pool is enabled.
24813 (arm_max_const_double_inline_cost): Remove use of
24814 arm_disable_literal_pool.
24815 (push_minipool_fix): Add assert.
24816 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24817 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24818 (no_literal_pool_sf_immediate): New.
24820 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24822 PR tree-optimization/80613
24823 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24824 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24826 2017-05-05 Richard Biener <rguenther@suse.de>
24828 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24830 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24832 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24833 of this flag from insn conditions due to removal from r247495.
24835 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24837 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24839 (arm_early_store_addr_dep_ptr): Likewise.
24840 * config/arm/aarch-common-protos.h
24841 (arm_early_load_addr_dep_ptr): Add prototype.
24842 (arm_early_store_addr_dep_ptr): Likewise.
24843 * config/arm/cortex-a53.md: Add new bypasses.
24845 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24847 * tree.c (next_type_uid): Change type to unsigned.
24848 (type_hash_canon): Decrement back next_type_uid if
24849 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24850 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24853 2017-05-04 Martin Sebor <msebor@redhat.com>
24855 * builtins.c: Fix a trivial typo in a comment.
24857 PR middle-end/79234
24858 * builtins.c (check_sizes): Adjust to handle reading past the end.
24859 Avoid printing excessive upper bound of ranges. Use %E to print
24860 tree nodes instead of converting them to %wu.
24861 (expand_builtin_memchr): New function.
24862 (compute_dest_size): Rename...
24863 (compute_objsize): ...to this.
24864 (expand_builtin_memcpy): Adjust.
24865 (expand_builtin_mempcpy): Adjust.
24866 (expand_builtin_strcat): Adjust.
24867 (expand_builtin_strcpy): Adjust.
24868 (check_strncat_sizes): Adjust.
24869 (expand_builtin_strncat): Adjust.
24870 (expand_builtin_strncpy): Adjust and simplify.
24871 (expand_builtin_memset): Adjust.
24872 (expand_builtin_bzero): Adjust.
24873 (expand_builtin_memcmp): Adjust.
24874 (expand_builtin): Handle memcmp.
24875 (maybe_emit_chk_warning): Check strncat just once.
24877 2017-05-04 Martin Sebor <msebor@redhat.com>
24879 PR preprocessor/79214
24880 PR middle-end/79222
24881 PR middle-end/79223
24882 * builtins.c (check_sizes): Add inlining context and issue
24883 warnings even when -Wno-system-headers is set.
24884 (check_strncat_sizes): Same.
24885 (expand_builtin_strncat): Same.
24886 (expand_builtin_memmove): New function.
24887 (expand_builtin_stpncpy): Same.
24888 (expand_builtin): Handle memmove and stpncpy.
24890 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24892 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24893 which is not used any more.
24895 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24897 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24899 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24901 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24902 (cortexa53_tunings): Likewise.
24903 (cortexa57_tunings): Likewise.
24904 (cortexa72_tunings): Likewise.
24905 (cortexa73_tunings): Likewise.
24907 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24909 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24910 Set loop alignment to 8.
24912 2017-05-04 Martin Sebor <msebor@redhat.com>
24914 PR translation/80280
24915 * builtins.c (expand_builtin_object_size): Add missing quoting to
24916 %D and like directives.
24917 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24918 (hsa_type_for_tree_type): Same.
24919 (verify_function_arguments): Same.
24920 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24921 * varasm.c (get_section): Same.
24924 2017-05-04 Martin Sebor <msebor@redhat.com>
24926 PR translation/80280
24927 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24929 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24931 * config/aarch64/aarch64.c (generic_addrcost_table):
24932 Change HI/TI mode setting.
24934 2017-05-04 Martin Jambor <mjambor@suse.cz>
24936 PR tree-optimization/80622
24937 * tree-sra.c (comes_initialized_p): New function.
24938 (build_accesses_from_assign): Only set write lazily when
24939 comes_initialized_p is false.
24940 (analyze_access_subtree): Use comes_initialized_p.
24941 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24942 instead of testing for PARM_DECL.
24944 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24946 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24947 constraint on operand 0 to allow more general addressing modes.
24948 Adjust output template.
24949 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24951 * config/aarch64/aarch64-protos.h
24952 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24953 * config/aarch64/constraints.md (Dp): New address constraint.
24954 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24957 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24959 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24960 update use of estimate_ipcp_clone_size_and_time.
24961 (estimate_local_effects): Update use of
24962 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24963 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24964 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24965 Return nonspecialized time.
24967 2017-05-04 Richard Biener <rguenther@suse.de>
24969 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24970 for the last VUSE which def dominates the PHI. Directly call
24972 (get_continuation_for_phi_1): Remove.
24974 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24976 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24977 to explain the use of truncating division. Cap the number of
24978 iterations to the maximum given by nb_iterations_upper_bound,
24981 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24983 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24984 * configure: Regenerate.
24985 * config.in: Regenerate.
24986 * config/i386/driver-mingw32.c: new file.
24987 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24988 * config.host: Link driver-mingw32.o on MinGW host.
24989 * doc/install.texi: Document new --enable-mingw-wildcard configure
24992 2017-05-04 Marek Polacek <polacek@redhat.com>
24994 PR tree-optimization/80612
24995 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24997 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24998 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25000 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25001 (movt splitter): Likewise.
25002 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25003 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25004 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25005 block for Thumb-1 with MOVT.
25006 (thumb2_legitimate_address_p): Move code block ...
25007 (can_avoid_literal_pool_for_label_p): ... into this new function.
25008 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25010 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25011 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25012 "M-profile targets with the MOVT instruction".
25014 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25016 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25017 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25018 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25020 2017-05-04 Martin Liska <mliska@suse.cz>
25022 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25023 variable cond_code.
25025 2017-05-04 Richard Biener <rguenther@suse.de>
25027 * tree.c (array_at_struct_end_p): Handle arrays at struct
25028 end with flexarrays more conservatively. Refactor and treat
25029 arrays of arrays or aggregates more strict. Fix
25030 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25031 * tree.c (array_at_struct_end_p): Adjust prototype.
25032 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25033 * gimple-fold.c (get_range_strlen): Likewise.
25034 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25036 2017-05-04 Richard Biener <rguenther@suse.de>
25038 PR tree-optimization/31130
25039 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25041 (supports_overflow_infinity): Likewise.
25042 (is_negative_overflow_infinity): Likewise.
25043 (is_positive_overflow_infinity): Likewise.
25044 (is_overflow_infinity): Likewise.
25045 (stmt_overflow_infinity): Likewise.
25046 (overflow_infinity_range_p): Likewise.
25047 (usable_range_p): Remove as always returning true.
25048 (make_overflow_infinity): Remove.
25049 (negative_overflow_infinity): Likewise.
25050 (positive_overflow_infinity): Likewise.
25051 (avoid_overflow_infinity): Likewise.
25052 (set_value_range): Adjust accordingly.
25053 (set_value_range_to_nonnegative): Likewise, remove now unused
25054 overflow_infinity arg.
25055 (vrp_operand_equal_p): Adjust.
25056 (update_value_range): Likewise.
25057 (range_int_cst_singleton_p): Likewise.
25058 (operand_less_p): Likewise.
25059 (compare_values_warnv): Likewise.
25060 (extract_range_for_var_from_comparison_expr): Likewise.
25061 (vrp_int_const_binop): Likewise.
25062 (zero_nonzero_bits_from_vr): Likewise.
25063 (extract_range_from_multiplicative_op_1): Likewise.
25064 (extract_range_from_binary_expr_1): Likewise.
25065 (extract_range_from_unary_expr): Likewise.
25066 (extract_range_from_comparison): Likewise.
25067 (extract_range_basic): Likewise.
25068 (adjust_range_with_scev): Likewise.
25069 (compare_ranges): Likewise.
25070 (compare_range_with_value): Likewise.
25071 (dump_value_range): Likewise.
25072 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25074 (simplify_cond_using_ranges): Adjust.
25076 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25078 * brig-builtins.def: Added a builtin for class_f64.
25079 * builtin-types.def: Added a builtin type needed by class_f64.
25081 2017-05-03 Jason Merrill <jason@redhat.com>
25083 * timevar.def: Add TV_CONSTEXPR.
25085 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25087 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25089 2017-05-03 Martin Jambor <mjambor@suse.cz>
25091 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25092 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25093 * ipa-cp.c (ipcp_propagate_stage): Do not call
25094 ipa_update_after_lto_read.
25095 * ipa-inline.c (ipa_inline): Likewise.
25097 2017-05-03 Martin Jambor <mjambor@suse.cz>
25099 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25100 tag. Added a default constructor and a destructor.
25101 (ipa_edge_args_sum_t): New class;
25102 (ipa_edge_args_sum): Declare.
25103 (ipa_edge_args_vector): Remove declaration.
25104 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25105 (ipa_free_edge_args_substructures): Remove declaration.
25106 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25107 (ipa_edge_args_info_available_for_edge_p): Likewise.
25108 * ipa-prop.c (ipa_edge_args_vector): Removed.
25109 (edge_removal_hook_holder): Likewise.
25110 (edge_duplication_hook_holder): Likewise.
25111 (ipa_edge_args_sum): New variable.
25112 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25113 ipa_edge_args_vector.
25114 (ipa_free_edge_args_substructures): Likewise.
25115 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25116 ipa_edge_args_vector.
25117 (ipa_edge_removal_hook): Turned into method
25118 ipa_edge_args_sum_t::remove.
25119 (ipa_edge_duplication_hook): Turned into method
25120 ipa_edge_args_sum_t::duplicate.
25121 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25122 registering edge hooks.
25123 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25124 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25125 ipa_edge_args_sum instead of ipa_edge_args_vector.
25126 * ipa-profile.c (ipa_profile): Likewise.
25128 2017-05-03 Martin Jambor <mjambor@suse.cz>
25130 * symbol-summary.h (function_summary): New method exists.
25131 (function_summary::symtab_removal): Deallocate through release.
25132 (call_summary): New class.
25133 (gt_ggc_mx): New overload.
25134 (gt_pch_nx): Likewise.
25135 (gt_pch_nx): Likewise.
25137 2017-05-03 Jeff Law <law@redhat.com>
25139 PR tree-optimization/78496
25140 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25141 from simplify_cond_using_ranges. Split off code to walk
25142 backwards through casts into ...
25143 (simplify_cond_using_ranges_2): New function.
25144 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25145 (execute_vrp): After identifying jump threads, call
25146 simplify_cond_using_ranges_2.
25148 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25151 * ipa-inline.h (inline_summary): Add ctor.
25152 (create_ggc): Do not use ggc_cleared_alloc.
25154 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25155 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25157 * gcc.c (handle_braces): Support escaping in switch matching
25159 * doc/invoke.texi (Spec Files): Document it.
25160 Remove superfluous @code markup in items.
25162 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25164 * diagnostic-show-locus.c (struct column_range): New struct.
25165 (get_affected_columns): New function.
25166 (get_printed_columns): New function.
25167 (struct correction): New struct.
25168 (correction::ensure_capacity): New function.
25169 (correction::ensure_terminated): New function.
25170 (struct line_corrections): New struct.
25171 (line_corrections::~line_corrections): New dtor.
25172 (line_corrections::add_hint): New function.
25173 (layout::print_trailing_fixits): Reimplement in terms of the new
25175 (selftest::test_overlapped_fixit_printing): New function.
25176 (selftest::diagnostic_show_locus_c_tests): Call it.
25178 2017-05-03 Nathan Sidwell <nathan@acm.org>
25180 Canonicalize canonical type hashing
25181 * tree.h (type_hash_canon_hash): Declare.
25182 * tree.c (type_hash_list, attribute_hash_list): Move into
25183 type_hash_canon_hash.
25184 (build_type_attribute_qual_variant): Break out hash code calc into
25185 type_hash_canon_hash.
25186 (type_hash_canon_hash): New. Generic type hash computation.
25187 (build_range_type_1, build_array_type_1, build_function_type,
25188 build_method_type_directly, build_offset_type, build_complex_type,
25189 make_vector_type): Call it.
25191 2017-05-03 Richard Biener <rguenther@suse.de>
25193 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25194 When all DRs have unknown misaligned do not always peel
25195 when there is a store but apply the same costing model as if
25196 there were only loads.
25198 2017-05-03 Richard Biener <rguenther@suse.de>
25201 PR tree-optimization/80492
25202 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25203 compare_base_decls returning dont-know properly.
25205 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25207 * config/arm/iterators.md (CCSI): New mode iterator.
25208 (arch): New mode attribute.
25209 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25210 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25211 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25212 code iterator for success result mode.
25213 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25214 the corresponding new insn generators.
25216 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25219 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25220 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25222 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25224 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25225 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25226 (DDR_A): Wrap DDR argument in brackets.
25227 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25228 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25229 (DDR_REVERSED_P): Likewise.
25231 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25233 PR tree-optimization/79472
25234 * tree-switch-conversion.c (struct switch_conv_info): Add
25235 contiguous_range and default_case_nonstandard fields.
25236 (collect_switch_conv_info): Compute contiguous_range and
25237 default_case_nonstandard fields, don't clear final_bb if
25238 contiguous_range and only the default case doesn't have the required
25240 (check_all_empty_except_final): Set default_case_nonstandard instead
25241 of failing if contiguous_range and the default case doesn't have empty
25243 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25244 and only the default case doesn't have the required constants. Skip
25246 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25247 if default_case_nonstandard.
25248 (build_constructors): Build constant 1 just once. Assert that default
25249 values aren't inserted in between cases if contiguous_range. Skip
25251 (build_arrays): Skip virtual phis.
25252 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25253 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25254 Handle virtual phis.
25255 (gen_inbound_check): Handle default_case_nonstandard case.
25256 (process_switch): Adjust check_final_bb caller. Call
25257 gather_default_values with the first non-default case instead of
25258 default case if default_case_nonstandard.
25260 2017-05-02 Nathan Sidwell <nathan@acm.org>
25262 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25263 check. Fix formatting.
25265 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25267 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25268 errors when comparing specialized and unspecialized times.
25270 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25272 * diagnostic-show-locus.c
25273 (layout::should_print_annotation_line_p): Make private.
25274 (layout::print_annotation_line): Make private.
25275 (layout::annotation_line_showed_range_p): Make private.
25276 (layout::show_ruler): Make private.
25277 (layout::print_source_line): Make private. Pass in line and
25278 line_width, rather than calling location_get_source_line. Drop
25280 (layout::print_leading_fixits): New method.
25281 (layout::print_any_fixits): Rename to...
25282 (layout::print_trailing_fixits): ...this, and make private.
25283 Don't print newline fixits.
25284 (diagnostic_show_locus): Move logic for printing one row into...
25285 (layout::print_line): ...this new function. Move the
25286 location_get_source_line call and error-handling from
25287 print_source_line to here. Call print_leading_fixits, and rename
25288 print_any_fixits to print_trailing_fixits.
25289 (selftest::test_fixit_insert_containing_newline): Update now that
25290 newlines are partially supported.
25291 (selftest::test_fixit_insert_containing_newline_2): New test.
25292 (selftest::test_fixit_replace_containing_newline): Update comments.
25293 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25294 * edit-context.c (class added_line): New class.
25295 (class edited_line): Describe newline handling in comment.
25296 (edited_line::actually_edited_p): New method.
25297 (edited_line::print_content): Delete redundant decl.
25298 (edited_line::m_predecessors): New field.
25299 (edited_file::print_content): Call edited_line::print_content.
25300 (edited_file::print_diff): Update to support newlines.
25301 (edited_file::print_diff_hunk): Likewise.
25302 (edited_file::print_run_of_changed_lines): New function.
25303 (edited_file::print_diff_line): Convert to...
25304 (print_diff_line): ...this.
25305 (edited_file::get_effective_line_count): New function.
25306 (edited_line::edited_line): Initialize new field m_predecessors.
25307 (edited_line::~edited_line): Clean up m_predecessors.
25308 (edited_line::apply_fixit): Handle newlines.
25309 (edited_line::get_effective_line_count): New function.
25310 (edited_line::print_content): New function.
25311 (edited_line::print_diff_lines): New function.
25312 (selftest::test_applying_fixits_insert_containing_newline): New
25314 (selftest::test_applying_fixits_replace_containing_newline): New
25316 (selftest::insert_line): New function.
25317 (selftest::test_applying_fixits_multiple_lines): Add example of
25319 (selftest::edit_context_c_tests): Call the new tests.
25321 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25323 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25324 parameter cand. Update dump information.
25325 (get_computation_cost): Update uses.
25327 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25329 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25330 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25331 (get_computation_at, rewrite_use_address): Update use of
25332 get_computation_aff.
25334 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25336 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25337 (get_computation): Delete.
25338 (get_computation_cost): Implement like get_computation_cost_at.
25339 Use get_computation_at.
25340 (get_computation_cost_at): Delete.
25341 (rewrite_use_nonlinear_expr): Use get_computation_at.
25342 (rewrite_use_compare, remove_unused_ivs): Ditto.
25344 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25346 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25348 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25350 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25351 (ivopts_global_cost_for_size): Rename parameter and update uses.
25352 (iv_ca_recount_cost): Update uses.
25353 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25354 candidates seperately in n_invs and n_cands.
25355 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25357 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25359 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25360 (find_inv_vars_cb): New.
25361 (find_depends): Renamed to ...
25362 (find_inv_vars): ... this.
25363 (add_candidate_1, force_var_cost): Call find_inv_vars.
25364 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25366 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25368 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25369 inv_vars. Add inv_exprs.
25370 (struct iv_cand): Rename depends_on to inv_vars.
25371 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25372 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25373 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25374 (dump_cand): Dump inv_vars.
25375 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25376 (record_invariant, find_depends, add_candidate_1): Ditto.
25377 (set_group_iv_cost, force_var_cost): Ditto.
25378 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25379 (get_computation_cost_at, get_computation_cost): Ditto.
25380 (determine_group_iv_cost_generic): Ditto.
25381 (determine_group_iv_cost_address): Ditto.
25382 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25383 (determine_group_iv_costs): Ditto.
25384 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25385 (iv_ca_set_remove_invariants): Renamed to ...
25386 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25387 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25388 (iv_ca_set_add_invariants): Renamed to ...
25389 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25390 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25391 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25392 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25393 (create_new_ivs): Remove useless dump.
25395 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25397 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25399 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25400 (iv_ca_set_no_cp, create_new_iv): Ditto.
25402 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25404 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25406 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25408 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25409 function tree_check2.
25411 2017-05-02 Martin Liska <mliska@suse.cz>
25413 * doc/gcov.texi: Add missing preposition.
25414 * gcov.c (function_info::function_info): Properly fill up
25415 all member variables.
25417 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25419 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25421 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25423 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25425 2017-05-02 Martin Liska <mliska@suse.cz>
25428 * lto-streamer-in.c (lto_read_tree_1): Remove
25429 LTO_STREAMER_DEBUG.
25430 * lto-streamer.c (struct tree_hash_entry): Likewise.
25431 (struct tree_entry_hasher): Likewise.
25432 (tree_entry_hasher::hash): Likewise.
25433 (tree_entry_hasher::equal): Likewise.
25434 (lto_streamer_init): Likewise.
25435 (lto_orig_address_map): Likewise.
25436 (lto_orig_address_get): Likewise.
25437 (lto_orig_address_remove): Likewise.
25438 * lto-streamer.h: Likewise.
25439 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25440 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25442 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25444 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25445 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25446 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25447 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25448 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25449 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25450 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25451 (mm_maskz_sub_ss): New intrinsics.
25452 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25453 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25454 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25455 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25456 (__builtin_ia32_subss_mask_round): New builtins.
25457 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25458 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25459 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25461 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25462 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25464 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25467 2017-05-02 Martin Jambor <mjambor@suse.cz>
25469 PR tree-optimization/78687
25470 * tree-sra.c (access): New field parent.
25471 (process_subtree_disqualification): New function.
25472 (disqualify_candidate): Call it.
25473 (build_accesses_from_assign): Reset write flag if creating an
25475 (build_access_subtree): Fill in parent field and also prpagate
25476 down grp_write flag.
25477 (create_artificial_child_access): New parameter set_grp_write, set
25478 grp_write to its value.
25479 (propagate_subaccesses_across_link): Also propagate grp_write flag
25481 (propagate_all_subaccesses): Push the closest parent back to work
25482 queue if add_access_to_work_queue returned true.
25484 2017-05-02 Richard Biener <rguenther@suse.de>
25486 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25487 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25488 -fstrict-overflow documentation.
25489 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25490 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25491 flag_strict_overflow.
25492 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25493 * lto-opts.c (lto_write_options): Do not stream it.
25494 * lto-wrapper.c (merge_and_complain): Do not handle it.
25495 * opts.c (default_options_table): Do not set -fstrict-overflow.
25496 (finish_options): Likewise do not clear it when sanitizing.
25497 * simplify-rtx.c (simplify_const_relational_operation): Do not
25498 test flag_strict_overflow.
25500 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25502 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25503 using enabled attribute.
25504 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25505 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25506 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25507 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25508 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25509 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25510 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25511 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25512 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25513 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25515 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25517 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25519 2017-05-02 Richard Biener <rguenther@suse.de>
25521 PR tree-optimization/80591
25523 2017-04-10 Richard Biener <rguenther@suse.de>
25525 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25528 2017-05-02 Richard Biener <rguenther@suse.de>
25530 PR tree-optimization/80549
25531 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25532 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25533 headers if they do not have a preheader.
25535 2017-05-02 Martin Liska <mliska@suse.cz>
25538 * common.opt: Fix typo.
25539 * doc/invoke.texi: Likewise.
25541 2017-05-01 Jan Beulich <jbeulich@suse.com>
25543 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25544 swapping, add (x,x,m,x,n) alternative.
25546 2017-05-01 Nathan Sidwell <nathan@acm.org>
25548 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25549 unnecessary unadjusted_alignment check.
25551 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25554 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25556 * gimplify.c (gimplify_cilk_detach): New function.
25557 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25558 * tree-core.h: Document EXPR_CILK_SPAWN.
25559 * tree.h (EXPR_CILK_SPAWN): Define.
25561 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25563 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25564 to use new fixit_hint representation, using the "replace" logic.
25565 (get_line_span_for_fixit_hint): Likewise.
25566 (layout::print_any_fixits): Likewise.
25567 (selftest::test_one_liner_many_fixits): Rename to...
25568 (selftest::test_one_liner_many_fixits_1): ...this, and update
25569 comment and expected output to reflect that the multiple fix-it
25570 hints are now consolidated into one insertion.
25571 (selftest::test_one_liner_many_fixits_2): New test.
25572 (selftest::test_diagnostic_show_locus_one_liner): Update for
25574 (selftest::test_fixit_consolidation): Update for fix-it API
25576 * diagnostic.c (print_parseable_fixits): Likewise.
25577 * edit-context.c (edited_line::m_line_events): Convert from
25578 auto_vec <line_event *> to auto_vec <line_event>.
25579 (class line_event): Convert from abstract base class to a concrete
25580 class, taking over the role of replace_event.
25581 (class insert_event): Delete.
25582 (class replace_event): Rename to class line_event. Convert to
25584 (edit_context::add_fixits): Reimplement.
25585 (edit_context::apply_insert): Delete.
25586 (edit_context::apply_replace): Rename to...
25587 (edit_context::apply_fixit): ...this. Convert to half-open range.
25588 (edited_file::apply_insert): Delete.
25589 (edited_file::apply_replace): Rename to...
25590 (edited_file::apply_fixit): ...this.
25591 (edited_line::~edited_line): Drop deletion of events.
25592 (edited_line::apply_insert): Delete.
25593 (edited_line::apply_replace): Rename to...
25594 (edited_line::apply_fixit): ...this. Convert to half-open range.
25595 Update for change to type of m_line_events.
25596 * edit-context.h (edit_context::apply_insert): Delete.
25597 (edit_context::apply_replace): Rename to...
25598 (edit_context::apply_fixit): ...this.
25600 2017-05-01 Martin Sebor <msebor@redhat.com>
25602 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25605 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25608 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25609 __get_cpuid_max returns 0.
25610 (__get_cpuid_count): Ditto.
25612 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25614 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25615 replacement expression is another instance of one of its arguments.
25617 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25620 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25621 check for stack push/pop autoinc.
25622 * config/i386/i386.c (ix86_agi_dependent): Return false
25623 if the only reason why modified_in_p returned true is that
25624 addr is SP based and set_insn is a push or pop.
25626 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25628 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25631 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25634 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25635 (account_size_time): Use two predicates - exec_pred and
25637 (evaluate_conditions_for_known_args): Compute both clause and
25639 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25640 (inline_summary_t::duplicate): Update.
25641 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25643 (compute_inline_parameters): Likewise.
25644 (estimate_edge_size_and_time): Update caluclation of time.
25645 (estimate_node_size_and_time): Compute both time and nonspecialized
25647 (estimate_ipcp_clone_size_and_time): Update.
25648 (inline_merge_summary): Update.
25649 (do_estimate_edge_time): Update.
25650 (do_estimate_edge_size): Update.
25651 (do_estimate_edge_hints): Update.
25652 (inline_read_section, inline_write_summary): Stream both new predicates.
25653 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25655 (compute_inlined_call_time): Cleanup.
25656 (big_speedup_p): Update.
25657 (edge_badness): Update.
25658 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25659 (size_time_entry): Replace predicate by exec_predicate and
25660 nonconst_predicate.
25661 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25662 (estimate_edge_time): Return also nonspec_time.
25663 (reset_edge_growth_cache): Update.
25665 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25667 PR rtl-optimization/80491
25668 * ifcvt.c (noce_process_if_block): When looking for x setter
25669 with missing else_bb, don't check only the insn right before
25670 cond_earliest, but look for the last insn that x is modified in
25671 within the same bb.
25673 PR rtl-optimization/80491
25674 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25676 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25678 PR tree-optimization/80487
25679 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25681 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25683 PR tree-optimization/79697
25684 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25685 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25686 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25688 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25689 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25691 2017-04-28 Martin Sebor <msebor@redhat.com>
25693 PR tree-optimization/80523
25694 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25695 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25697 (maybe_warn, format_directive, parse_directive): Use new functions.
25698 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25700 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25702 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25704 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25706 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25707 target_header_dir): Set correctly.
25708 * configure: Regenerated.
25709 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25710 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25711 instead of SYSTEM_HEADER_DIR.
25713 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25715 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25716 (estimate_local_effects): Likewise.
25717 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25718 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25719 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25720 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25721 do_estimate_edge_time, estimate_edge_time): Likewise.
25722 * ipa-inline-analysis.c (estimate_node_size_and_time,
25723 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25724 (estimate_time_after_inlining): Remove.
25726 2017-04-28 Martin Liska <mliska@suse.cz>
25728 * doc/gcov.texi: Enhance documentation of gcov.
25730 2017-04-28 Martin Liska <mliska@suse.cz>
25732 * doc/gcov.texi: Sort options in alphabetic order.
25733 * doc/gcov-dump.texi: Likewise.
25734 * doc/gcov-tool.texi: Likewise.
25735 * gcov.c (print_usage): Likewise.
25736 * gcov-dump.c (print_usage): Likewise.
25737 * gcov-tool.c (print_merge_usage_message): Likewise.
25738 (print_rewrite_usage_message): Likewise.
25739 (print_overlap_usage_message): Likewise.
25741 2017-04-28 Martin Liska <mliska@suse.cz>
25743 PR gcov-profile/53915
25744 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25746 2017-04-28 Martin Liska <mliska@suse.cz>
25748 PR gcov-profile/79891
25749 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25750 is marked by compiler as living on a line.
25751 (get_cycles_count): Remove usage of the union.
25752 (output_intermediate_file): Likewise.
25753 (find_source): Fix GNU coding style.
25754 (accumulate_line_counts): Remove old non-all block mode.
25755 (output_lines): Remove usage of the union.
25756 * profile.c (output_location): Include all BBs, even if
25757 belonging to a same line (and file) as a previous BB.
25759 2017-04-28 Martin Liska <mliska@suse.cz>
25761 * gcov.c (process_args): Handle new argument 'w'.
25762 (read_graph_file): Assign ID to BBs.
25763 (output_branch_count): Display BB # if verbose flag is set.
25764 (output_lines): Likewise for arcs.
25765 (print_usage): Add '--verbose' option help.
25766 * doc/gcov.texi: Document --verbose (-w) option.
25768 2017-04-28 Martin Liska <mliska@suse.cz>
25770 * gcov.c (struct block_location_info): New struct.
25771 (process_file): Fill up the new structure.
25772 (read_graph_file): Replace usage of encoding by the newly added
25774 (add_line_counts): Likewise.
25775 (accumulate_line_counts): Remove usage of the union.
25776 (function_info::function_info): New function.
25777 (function_info::~function_info): Likewise.
25778 (process_file): Call delete instead of release_function.
25779 (release_function): Release the function.
25780 (release_structures): Call delete instead of release_function.
25781 (solve_flow_graph): Replace usage of num_blocks.
25782 (find_exception_blocks): Likewise.
25783 (output_lines): Fix GNU coding style.
25785 2017-04-28 Martin Liska <mliska@suse.cz>
25788 * coverage.c (coverage_remove_note_file): New function.
25789 * coverage.h: Declare the function.
25790 * toplev.c (finalize): Clean if an error has been seen.
25792 2017-04-28 Martin Liska <mliska@suse.cz>
25794 PR gcov-profile/80031
25795 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25796 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25797 * gcov.c (read_graph_file): Read just number of blocks.
25798 * profile.c (branch_prob): Do not stream 0 flags per a basic
25801 2017-04-28 Martin Liska <mliska@suse.cz>
25803 * gcov-dump.c (tag_*): Add new argument to declarations.
25804 (dump_gcov_file): Likewise.
25805 (tag_blocks): Add and use new argument depth.
25806 (tag_arcs): Likewise.
25807 (tag_lines): Likewise.
25808 (tag_counters): Likewise.
25809 (tag_summary): Likewise.
25810 (dump_working_sets): Use depth to do a proper indentation.
25812 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25815 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25816 * symtab.c (symtab_node::debug_symtab): Move definition here.
25818 2017-04-28 Richard Biener <rguenther@suse.de>
25820 * lto-streamer.h (LTO_major_version): Bump to 7.
25822 2017-04-28 Richard Biener <rguenther@suse.de>
25824 * tree-vrp.c (assert_info): New struct.
25825 (add_assert_info): New helper.
25826 (register_edge_assert_for_2): Refactor to add asserts to a vector
25828 (register_edge_assert_for_1): Likewise.
25829 (register_edge_assert_for): Likewise.
25830 (finish_register_edge_assert_for): New helper actually registering
25831 asserts where live on edge.
25832 (find_conditional_asserts): Adjust.
25833 (find_switch_asserts): Likewise.
25834 (evrp_dom_walker::try_find_new_range): Generalize.
25835 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25837 2017-04-27 Marek Polacek <polacek@redhat.com>
25840 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25841 arg10 and arg11 to itype.
25843 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25845 * doc/extend.texi (Object Size Checking): Improve grammar.
25847 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25850 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25851 that the logic for permitting reciprocal estimates matches that
25854 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25857 * tree.c (type_cache_hasher::equal): Only compare
25858 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25859 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25860 non-aggregate element types.
25861 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25862 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25864 2017-04-27 Richard Biener <rguenther@suse.de>
25866 PR middle-end/80533
25867 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25868 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25869 keeping a reference to a trailing array.
25871 2017-04-27 Richard Biener <rguenther@suse.de>
25873 PR middle-end/80539
25874 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25875 being in loop-closed SSA form conservatively.
25876 (chrec_fold_multiply_poly_poly): Likewise.
25878 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25880 PR middle-end/79665
25881 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25882 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25884 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25887 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25888 (aarch64_function_arg_alignment): Return unsigned int again, but still
25889 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25890 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25891 Don't emit -Wpsabi note.
25892 (aarch64_function_arg_boundary): Likewise.
25893 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25896 2017-04-26 Nathan Sidwell <nathan@acm.org>
25898 * tree.h (crc32_unsigned_n): Declare.
25899 (crc32_unsigned, crc32_unsigned): Make inline.
25900 * tree.c (crc32_unsigned_bits): Replace with ...
25901 (crc32_unsigned_n): ... this.
25902 (crc32_unsigned, crc32_byte): Remove.
25903 (crc32_string): Remove unnecessary braces.
25905 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25907 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25908 * ipa-inline-analysis.c (MAX_TIME): Remove.
25909 (account_size_time): Use sreal for time.
25910 (dump_inline_summary): Update.
25911 (estimate_function_body_sizes): Update.
25912 (estimate_edge_size_and_time): Update.
25913 (estimate_calls_size_and_time): Update.
25914 (estimate_node_size_and_time): Update.
25915 (inline_merge_summary): Update.
25916 (inline_update_overall_summary): Update.
25917 (estimate_time_after_inlining): Update.
25918 (inline_read_section): Update.
25919 (inline_write_summary): Update.
25920 * ipa-inline.c (compute_uninlined_call_time): Update.
25921 (compute_inlined_call_time): Update.
25922 (recursive_inlining): Update.
25923 (inline_small_functions): Update.
25924 (dump_overall_stats): Update.
25925 * ipa-inline.h: Include sreal.h.
25926 (size_time_entry): Turn time to sreal.
25927 (inline_summary): Turn self_time nad time to sreal.
25929 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25931 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25933 (sreal::stream_out, sreal::stream_in): New.
25934 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25936 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25938 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25941 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25944 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25946 (dimode_scalar_chain::compute_convert_gain): Ditto.
25947 (dimode_scalar_chain::make_vector_copies): Ditto.
25948 (dimode_scalar_chain::convert_reg): Ditto.
25949 (dimode_scalar_chain::convert_insn): Ditto.
25950 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25951 (VI248_AVX512BW_1): New mode iterator.
25952 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25953 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25956 2017-04-25 Martin Sebor <msebor@redhat.com>
25958 PR tree-optimization/80497
25959 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25960 constants are representable in HOST_WIDE_INT.
25961 (parse_directive): Ditto.
25963 2017-04-25 Martin Sebor <msebor@redhat.com>
25966 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25967 (new_zero_array): Adjust signature.
25968 (dom_info::dom_init): Used unsigned rather that size_t.
25969 (dom_info::dom_info): Same.
25971 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25972 Jakub Jelinek <jakub@redhat.com>
25975 * config/arm/arm.c: Include gimple.h.
25976 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25977 returns negative, increment ncrn only if it returned positive.
25978 (arm_needs_doubleword_align): Return int instead of bool,
25979 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25980 members, but if there is any such non-FIELD_DECL
25981 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25982 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25983 returns negative, increment nregs only if it returned positive.
25984 (arm_setup_incoming_varargs): Likewise.
25985 (arm_function_arg_boundary): Emit -Wpsabi note if
25986 arm_needs_doubleword_align returns negative, return
25987 DOUBLEWORD_ALIGNMENT only if it returned positive.
25989 2017-04-25 Marek Polacek <polacek@redhat.com>
25992 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25993 first argument to type.
25995 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25998 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25999 type checks to test for compatibility instead of equality.
26001 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26002 Jakub Jelinek <jakub@redhat.com>
26005 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26007 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26008 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26009 the alignment computation, but return their maximum in warn_alignment.
26010 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26011 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26013 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26014 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26017 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26019 * config/arc/simdext.md (dmpyh): Fix typo.
26021 2017-04-25 Richard Biener <rguenther@suse.de>
26023 PR tree-optimization/80492
26024 * alias.c (compare_base_decls): Handle registers with asm
26025 specification conservatively.
26026 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26027 compare_base_decls returning dont-know properly.
26029 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26031 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26032 (legitimate_offset_address_p): New function.
26033 (arc_legitimate_address_p): Use above function.
26035 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26037 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26039 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26041 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26042 ACCH registers whenever they are available.
26044 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26046 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26047 double regs fix when not used.
26049 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26051 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26053 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26054 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26056 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26058 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26059 when emitting short ADD instructions.
26061 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26063 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26065 (cmpsi_cc_c_insn): Likewise.
26066 (cbranchsi4_scratch): Compute proper instruction length using
26067 compact_hreg_operand.
26068 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26070 2017-04-25 Richard Biener <rguenther@suse.de>
26072 PR middle-end/80509
26073 * passes.c (pass_manager::pass_manager): Initialize
26074 m_name_to_pass_map.
26076 2017-04-25 Richard Biener <rguenther@suse.de>
26078 PR tree-optimization/79201
26079 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26081 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26084 * config/s390/vector.md: Split MEM->GPR vector moves for
26085 non-s_operand addresses.
26087 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26090 * config/s390/predicates.md (reload_const_wide_int_operand): New
26092 * config/s390/s390.md ("movti"): Remove d/P alternative.
26093 ("movti_bigconst"): New pattern definition.
26095 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26098 * s390-protos.h (s390_expand_cs_hqi): Removed.
26099 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26100 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26101 modes as well as CCZ1mode and CCZmode.
26102 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26103 signature of s390_emit_compare_and_swap.
26104 (s390_expand_cs_hqi): Likewise, make static.
26105 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26106 compare-and-swap, in some cases.
26107 (s390_expand_cs): Wrapper function.
26108 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26110 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26111 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26112 patterns for small and large integers. Forbid symref memory operands.
26113 Move expander to s390.c. Require cc register.
26114 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26115 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26116 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26117 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26118 symref memory operands. Remove CC mode and call s390_match_ccmode
26120 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26122 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26124 * config/s390/s390.md (define_peephole2): New peephole to help
26125 combining the load-and-test pattern with volatile memory.
26127 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26129 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26130 with CCZmode for TARGET_Z196.
26132 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26134 PR rtl-optimization/80501
26135 * combine.c (make_compound_operation_int): Set subreg_code to SET
26136 even for AND with mask of the sign bit of mode.
26138 PR rtl-optimization/80500
26139 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26140 sum's initial value.
26142 2017-04-25 Julian Brown <julian@codesourcery.com>
26143 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26145 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26147 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26149 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26151 2017-04-25 Julian Brown <julian@codesourcery.com>
26152 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26154 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26155 (thunderx2t99_sha): New Reservation.
26157 2017-04-25 Julian Brown <julian@codesourcery.com>
26158 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26160 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26161 type for 1-element load.
26163 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26165 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26167 2017-04-24 Martin Jambor <mjambor@suse.cz>
26169 PR tree-optimization/80293
26170 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26171 char arrays not totally scalarizable if it is false.
26172 (analyze_all_variable_accesses): Pass correct value in the new
26173 parameter. Add a statistics counter.
26175 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26177 PR middle-end/79931
26178 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26180 2017-04-24 Richard Biener <rguenther@suse.de>
26182 PR tree-optimization/80494
26183 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26184 out for complex types.
26186 2017-04-24 Richard Biener <rguenther@suse.de>
26188 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26189 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26190 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26191 (DFS): Adjust and never fail.
26192 (sccvn_dom_walker::fail): Remove.
26193 (sccvn_dom_walker::before_dom_children): Adjust.
26194 (run_scc_vn): Likewise and never fail.
26195 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26196 (pass_fre::execute): Likewise.
26198 2017-04-24 Richard Biener <rguenther@suse.de>
26200 PR tree-optimization/79725
26201 * tree-ssa-sink.c (statement_sink_location): Return whether
26202 failure reason was zero uses. Move that check later.
26203 (sink_code_in_bb): Deal with zero uses by removing the stmt
26206 2017-04-24 Richard Biener <rguenther@suse.de>
26209 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26210 pointer-based references.
26212 2017-04-24 Richard Biener <rguenther@suse.de>
26215 * pass_manager.h (pass_manager::operator new): Remove.
26216 (pass_manager::operator delete): Likewise.
26217 * passes.c (pass_manager::operator new): Remove.
26218 (pass_manager::operator delete): Likewise.
26219 (pass_manager::pass_manager): Zero individual pass members.
26221 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26224 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26225 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26226 Check "XEXP (src, 1)" operand here.
26227 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26228 Check "XEXP (src, 1)" operand here.
26229 (dimode_scalar_chain::make_vector_copies): Detect count register
26230 of a shift instruction. Zero extend count register from QImode
26231 to DImode to satisfy vector shift pattern count operand predicate.
26232 Substitute vector shift count operand with a DImode copy.
26233 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26236 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26238 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26239 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26240 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26241 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26242 (UNSPEC_NOREX_MEM): Remove definition.
26244 2017-04-21 Richard Biener <rguenther@suse.de>
26246 PR tree-optimization/79547
26247 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26248 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26249 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26250 without any constraints.
26252 2017-04-21 Richard Biener <rguenther@suse.de>
26254 PR tree-optimization/78847
26255 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26257 2017-04-21 Richard Biener <rguenther@suse.de>
26259 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26260 (build_distinct_type_copy): Likewise.
26261 (build_variant_type_copy): Likewise.
26262 * tree.c (build_qualified_type): Pass down mem-stat info.
26263 (build_distinct_type_copy): Likewise.
26264 (build_variant_type_copy): Likewise.
26266 2017-04-21 Richard Biener <rguenther@suse.de>
26268 PR tree-optimization/80237
26269 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26271 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26272 for a simplified result.
26274 2016-04-21 Richard Biener <rguenther@suse.de>
26276 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26277 sth as strict as a simple_iv but a chrec without symbols and an
26278 operand defined in the loop we are peeling (and not some subloop).
26279 (propagate_constants_for_unrolling): Propagate all constants.
26281 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26284 * config/i386/i386.c (print_reg): Remove assert for disalowed
26285 regno values, call output_operand_lossage instead.
26287 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26290 * config/i386/constraints.md (Yc): New register constraint.
26291 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26292 Use Yc constraint for alternative 2 of operand 0. Remove
26293 preferred_for_speed attribute.
26295 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26297 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26298 lastprivate clauses in SIMT case.
26300 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26302 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26304 2017-04-20 Richard Biener <rguenther@suse.de>
26306 PR tree-optimization/57796
26307 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26308 as N scalar stores.
26309 (vect_model_load_cost): Cost gathers as N scalar loads.
26311 2017-04-20 Richard Biener <rguenther@suse.de>
26313 * ggc-page.c (ggc_allocated_p): Rename to ...
26314 (safe_lookup_page_table_entry): ... this and return the lookup
26316 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26318 2017-04-20 Richard Biener <rguenther@suse.de>
26320 PR tree-optimization/80453
26321 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26322 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26323 from the conditions.
26324 (vn_phi_eq): Pass them down.
26325 (vn_phi_lookup): Record them.
26326 (vn_phi_insert): Likewise.
26328 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26330 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26331 uninitialized variable warning to avoid buffer overrun.
26333 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26336 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26337 is suppressed for '{ 0 }' in C.
26339 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26341 * BASE-VER: Set to 8.0.0.
26343 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26345 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26346 priority .init_array and .fini_array section with SECTION_NOTYPE
26349 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26351 PR middle-end/80423
26352 * tree.h (build_array_type): Add typeless_storage default argument.
26353 * tree.c (type_cache_hasher::equal): Also compare
26354 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26355 (build_array_type): Add typeless_storage argument, set
26356 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26358 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26359 (build_array_type): Likewise. Add typeless_storage argument.
26361 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26362 Jakub Jelinek <jakub@redhat.com>
26364 PR tree-optimization/80426
26365 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26366 operation on symbolic operands, also compute the overflow for the
26367 invariant part when the operation degenerates into a negation.
26369 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26372 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26373 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26376 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26378 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26381 * config/avr/avr.c (tree.h): Include it.
26382 (cgraph.h): Include it.
26383 (avr_encode_section_info): Don't warn for uninitialized progmem
26384 variable if it's just an alias.
26386 2017-04-19 Richard Biener <rguenther@suse.de>
26389 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26390 when needed by AutoPGO.
26392 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26395 * doc/lto.texi: Remove an extra 'that'.
26397 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26399 PR rtl-optimization/80429
26400 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26401 are only used in debug insns.
26403 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26404 Vladimir Makarov <vmakarov@redhat.com>
26406 * config/sparc/predicates.md (input_operand): Add comment. Return
26407 true for any memory operand when LRA is in progress.
26408 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26410 2017-04-18 Jeff Law <law@redhat.com>
26413 * mips.md ({return,simple_return}_internal): Do not overwrite
26416 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26418 PR tree-optimization/80443
26419 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26420 instead of adding 1, subtract -1 and similarly instead of subtracting
26423 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26425 PR rtl-optimization/80357
26426 * haifa-sched.c (tmp_bitmap): New variable.
26427 (model_recompute): Handle duplicate use records.
26428 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26429 (free_global_sched_pressure_data): Free it.
26431 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26434 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26435 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26436 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26437 instead of SYSTEM_HEADER_DIR.
26439 2017-04-18 Jeff Law <law@redhat.com>
26441 PR middle-end/80422
26442 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26443 predecessors after walking up the insn chain.
26445 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26448 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26449 sizetype type into debug info.
26451 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26454 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26455 unneeded test for TARGET_UPPER_REGS_SF.
26456 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26458 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26461 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26462 instead of gsi_after_labels.
26464 2017-04-18 Jeff Law <law@redhat.com>
26466 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26470 2017-04-13 Jeff Law <law@redhat.com>
26471 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26472 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26474 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26477 * config/avr/avr.c (intl.h): Include it.
26478 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26480 2017-04-18 Martin Liska <mliska@suse.cz>
26482 PR gcov-profile/78783
26483 * gcov-tool.c (gcov_output_files): Validate that destination
26484 file is either removed by the tool or by a user.
26486 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26487 Guy Benyei <guybe@mellanox.com>
26489 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26490 block, and do not negate it, the stored id is already negative.
26492 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26494 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26496 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26499 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26500 masks of options that should be turned off if the VSX vector
26501 options are turned off.
26502 (OTHER_P8_VECTOR_MASKS): Likewise.
26503 (OTHER_VSX_VECTOR_MASKS): Likewise.
26504 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26505 rs6000_disable_incompatible_switches to validate no type switches
26507 (rs6000_incompatible_switch): New function to disallow turning on
26508 other vector options if -mno-vsx, -mno-power8-vector, or
26509 -mno-power9-vector are specified.
26511 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26513 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26515 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26517 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26518 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26519 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26520 (ARG_POINTER_CFA_OFFSET): Likewise.
26522 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26524 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26525 conditions to take advantage of various optimizations.
26527 2017-04-13 Jeff Law <law@redhat.com>
26529 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26530 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26531 (zero_extendsidi2_dext): Likewise.
26533 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26536 * fold-const.c (fold_ternary_loc): Revert
26537 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26540 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26542 PR rtl-optimization/80343
26543 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26544 new scratch pseudo.
26546 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26549 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26550 to ubsan_encode_value.
26552 2017-04-13 Jeff Law <law@redhat.com>
26554 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26555 appearing in DEBUG_INSNs.
26557 2017-04-13 Martin Liska <mliska@suse.cz>
26559 PR gcov-profile/80413
26560 * gcov-io.c (gcov_write_string): Copy to buffer just when
26561 allocated size is greater than zero.
26563 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26566 * dwarf2out.c (decls_for_scope): Ignore declarations of
26567 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26569 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26572 * ipa-visibility.c (non_local_p): Fix typos.
26573 (localize_node): When localizing symbol in same comdat group,
26574 dissolve the group only when we know external symbols are going
26576 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26578 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26580 PR tree-optimization/79390
26581 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26582 order does not result in usable sequence, retry with reversed operand
26588 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26589 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26590 op0 instead of fold_convert_loc (loc, type, arg0).
26592 2017-04-12 Jeff Law <law@redhat.com>
26594 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26595 has a delay slot in the generated code.
26597 * config/cris/cris.md (cris_preferred_reload_class): Return
26598 GENNONACR_REGS rather than GENERAL_REGS.
26600 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26603 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26604 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26605 signedness of the result type.
26607 2017-04-12 Richard Biener <rguenther@suse.de>
26608 Jeff Law <law@redhat.com>
26610 PR tree-optimization/80359
26611 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26612 trim stores to TARGET_MEM_REFs.
26614 2017-04-12 Richard Biener <rguenther@suse.de>
26616 PR tree-optimization/79390
26617 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26618 threading case even more.
26620 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26623 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26624 for quad_address_p for TImode, instead of just not indexed_address.
26626 2017-04-12 Richard Biener <rguenther@suse.de>
26627 Bernd Edlinger <bernd.edlinger@hotmail.de>
26629 PR middle-end/79671
26630 * alias.c (component_uses_parent_alias_set_from): Handle
26631 TYPE_TYPELESS_STORAGE.
26632 (get_alias_set): Likewise.
26633 * tree-core.h (tree_type_common): Add typeless_storage flag.
26634 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26635 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26636 for types containing members with TYPE_TYPELESS_STORAGE.
26637 (place_field): Likewise.
26638 (layout_type): Likewise for ARRAY_TYPE.
26639 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26640 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26641 TYPE_TYPELESS_STORAGE.
26642 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26644 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26647 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26648 first argument to type.
26650 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26654 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26655 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26656 (rs6000_expand_binop_builtin): Likewise.
26657 (rs6000_expand_ternop_builtin): Likewise; also add missing
26658 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26659 vshasigma built-ins.
26660 * doc/extend.texi: Document that vec_xxpermdi's third argument
26661 must be a constant.
26663 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26665 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26666 Use shift_const cost parameter when calculating gain of STV shifts.
26668 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26670 PR rtl-optimization/70478
26671 * lra-constraints.c (process_alt_operands): Check memory for
26672 disfavoring memory insn operand.
26674 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26676 PR middle-end/80100
26677 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26678 left shift in unsigned HOST_WIDE_INT type.
26680 PR rtl-optimization/80385
26681 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26682 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26685 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26686 if they have any depend clauses.
26688 2017-04-11 Martin Liska <mliska@suse.cz>
26691 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26692 * ipa-split.c (split_function): Create a local comdat symbol
26693 if caller is in a comdat group.
26695 2017-04-11 Martin Liska <mliska@suse.cz>
26698 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26701 2017-04-11 Martin Sebor <msebor@redhat.com>
26703 PR middle-end/80364
26704 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26705 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26707 (directive::set_width, directive::set_precision, format_character):
26709 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26712 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26715 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26716 conflict, set target->arch_name instead of target->cpu_name.
26718 2017-04-11 Richard Biener <rguenther@suse.de>
26720 PR tree-optimization/80374
26721 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26722 build_zero_cst, remove fold_convertible_p check again.
26724 2017-04-11 Martin Liska <mliska@suse.cz>
26727 * ubsan.c (instrument_object_size): Do not instrument register
26730 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26733 * config/i386/i386-builtin-types.def
26734 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26735 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26736 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26737 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26738 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26739 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26740 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26741 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26742 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26743 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26744 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26745 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26746 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26747 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26748 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26749 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26750 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26751 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26752 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26753 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26754 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26755 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26756 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26757 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26758 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26759 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26760 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26761 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26762 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26763 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26764 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26765 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26766 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26767 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26768 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26769 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26770 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26771 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26772 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26773 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26774 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26775 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26776 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26777 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26778 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26779 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26781 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26782 flag to second_arg_count, handle 4 argument function type _COUNT
26783 aliases, handle second_arg_count on second argument rather than last.
26785 2017-04-10 Jeff Law <law@redhat.com>
26787 PR tree-optimization/80374
26788 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26789 record anything if we can not convert integer_zero_node to the
26792 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26795 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26796 Enhance special handling given to the TARGET_P9_MINMAX option in
26797 relation to certain other options.
26799 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26801 PR tree-optimization/80153
26802 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26803 remove POINTER_PLUS_EXPR's base part directly, rather than through
26806 2017-04-10 Richard Biener <rguenther@suse.de>
26807 Bin Cheng <bin.cheng@arm.com>
26809 PR tree-optimization/80153
26810 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26811 the first element of pointer type aff_tree. Build result expr in
26813 (add_elt_to_tree): Convert to type unconditionally. Remove other
26814 fold_convert calls.
26815 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26816 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26818 2017-04-10 Richard Biener <rguenther@suse.de>
26820 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26823 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26825 PR rtl-optimization/70478
26826 * lra-constraints.c (curr_small_class_check): New.
26827 (update_and_check_small_class_inputs): New.
26828 (process_alt_operands): Update curr_small_class_check. Disfavor
26829 alternative insn memory operands. Check available regs for small
26832 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26835 * config/mips/mips.opt (-mvirt): Update description.
26836 * doc/invoke.texi (-mvirt): Likewise.
26838 2017-04-10 Richard Biener <rguenther@suse.de>
26840 PR middle-end/80362
26841 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26842 looking for NEGATE_EXPR in -A / -B to A / B folding.
26844 2017-04-10 Martin Liska <mliska@suse.cz>
26846 PR gcov-profile/80224
26847 * gcov.c (print_usage): Fix usage string.
26848 (get_gcov_intermediate_filename): Remove.
26849 (output_gcov_file): Use both for normal and intermediate format.
26850 (generate_results): Do not initialize special file for
26851 intermediate format.
26853 2017-04-10 Richard Biener <rguenther@suse.de>
26855 PR tree-optimization/80304
26856 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26859 2017-04-10 Nathan Sidwell <nathan@acm.org>
26862 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26863 (rs6000_init_builtins): Use it.
26865 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26867 * config/arm/arm.md (<mrc>): Add mode to SET source.
26868 (<mrrc>): Likewise.
26870 2017-04-10 Richard Biener <rguenther@suse.de>
26872 PR middle-end/80344
26873 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26875 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26878 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26879 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26880 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26881 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26882 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26883 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26884 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26885 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26886 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26887 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26888 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26889 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26890 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26891 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26892 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26893 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26894 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26895 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26896 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26897 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26898 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26899 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26900 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26902 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26904 PR rtl-optimization/70478
26905 * lra-constraints.c: Reverse the last patch.
26907 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26909 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26910 Add comment for WCHAR_T.
26912 2017-04-08 Martin Liska <mliska@suse.cz>
26915 2017-04-07 Martin Liska <mliska@suse.cz>
26918 * ipa-split.c (split_function): Add function part to a same comdat
26921 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26924 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26926 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26928 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26929 (vec_pairing): ...to this.
26930 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26931 (rs6000_sched_init): Adjust for name change.
26932 (struct rs6000_sched_context): Likewise.
26933 (rs6000_init_sched_context): Likewise.
26934 (rs6000_set_sched_context): Likewise.
26936 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26942 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26944 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26945 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26946 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26948 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26950 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26952 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26954 PR rtl-optimization/70703
26955 * ira-color.c (update_conflict_hard_regno_costs): Use
26956 int64_t instead of HOST_WIDE_INT.
26958 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26960 PR rtl-optimization/70478
26961 * lra-constraints.c (process_alt_operands): Disfavor alternative
26962 insn memory operands.
26964 2017-04-07 Jeff Law <law@redhat.com>
26966 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26967 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26969 2017-04-07 Martin Liska <mliska@suse.cz>
26972 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26973 Show error message instead of an ICE.
26975 2017-04-07 Martin Liska <mliska@suse.cz>
26978 * ipa-split.c (split_function): Add function part to a same comdat
26981 2017-04-07 Richard Biener <rguenther@suse.de>
26983 PR middle-end/80341
26984 * tree.c (get_unwidened): Also handle ! for_type case for
26986 * convert.c (do_narrow): Split out from ...
26987 (convert_to_integer_1): ... here. Do not pass final truncation
26988 type to get_unwidened for TRUNC_DIV_EXPR.
26990 2017-04-07 Richard Biener <rguenther@suse.de>
26992 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26994 (aff_combination_const): Adjust.
26995 (aff_combination_scale): Likewise.
26996 (aff_combination_add_elt): Likewise.
26997 (aff_combination_add_cst): Likewise.
26998 (aff_combination_convert): Likewise.
26999 (add_elt_to_tree): Likewise. Remove unused argument.
27000 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27002 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27004 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27006 * config/arm/arm.c (arm_default_short_enums): Use
27007 ARM_DEFAULT_SHORT_ENUMS.
27008 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27010 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27013 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27014 members with redundant out-of-class redeclaration.
27016 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27019 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27020 * config/i386/i386.md (*zero_extendsidi2):
27021 Add (?*x,*x) and (?*v,*v) alternatives.
27023 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27026 * config/i386/i386.c (ix86_expand_builtin)
27027 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27028 mode from insn data. Convert operands to insn operand mode.
27029 Copy operands that don't satisfy insn predicate to a register.
27031 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27033 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27036 2017-04-06 Richard Biener <rguenther@suse.de>
27038 PR tree-optimization/80334
27039 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27040 preserve alignment of accesses.
27042 2017-04-06 Richard Biener <rguenther@suse.de>
27044 PR tree-optimization/80262
27045 * tree-sra.c (build_ref_for_offset): Preserve address-space
27047 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27048 Drop useless address-space information on MEM_REF offsets.
27050 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27052 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27054 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27056 PR rtl-optimization/70703
27057 * ira-color.c (update_conflict_hard_regno_costs): Use
27058 HOST_WIDE_INT instead of long.
27060 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27063 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27064 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27066 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27067 for x86_64 target. Handle -m3dnowa option.
27069 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27071 PR rtl-optimization/70703
27072 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27073 (update_conflict_hard_regno_costs): Use long instead of unsigned
27074 arithmetic for cost calculation.
27076 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27077 Bernd Edlinger <bernd.edlinger@hotmail.de>
27080 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27083 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27086 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27087 ptr_mode with Pmode throughout.
27088 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27089 into probe_stack_range and use DImode.
27091 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27094 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27095 call_eh_return is true.
27097 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27099 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27100 Initialize last_match_fntype_index.
27102 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27105 * tree-nvr.c: Include internal-fn.h.
27106 (pass_return_slot::execute): Ignore internal calls without
27109 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27110 Richard Biener <rguenther@suse.de>
27113 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27114 captures used multiple times, except for the last use.
27115 * generic-match-head.c: Include gimplify.h.
27117 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27119 PR tree-optimization/79390
27120 * target.h (struct noce_if_info): Declare.
27121 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27122 * target.def (noce_conversion_profitable_p): New target hook.
27123 * ifcvt.h (struct noce_if_info): New type, moved from ...
27124 * ifcvt.c (struct noce_if_info): ... here.
27125 (noce_conversion_profitable_p): Renamed to ...
27126 (default_noce_conversion_profitable_p): ... this. No longer
27128 (noce_try_store_flag_constants, noce_try_addcc,
27129 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27130 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27131 instead of noce_conversion_profitable_p.
27132 * config/i386/i386.c: Include ifcvt.h.
27133 (ix86_option_override_internal): Don't override
27134 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27135 (ix86_noce_conversion_profitable_p): New function.
27136 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27137 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27138 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27139 * doc/tm.texi: Regenerated.
27141 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27143 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27146 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27149 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27150 instructions for small multiply cores.
27152 2017-04-04 Jeff Law <law@redhat.com>
27154 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27156 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27157 that are not set by the loop over the elements.
27159 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27162 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27163 int mode, convert_modes it to mode as unsigned, otherwise use
27164 lowpart_subreg to mode rather than SImode.
27165 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27166 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27167 Use DImode instead of SImode for the shift count operand.
27168 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27171 2017-04-04 Richard Biener <rguenther@suse.de>
27173 PR middle-end/80281
27174 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27175 arithmetic done for the negate or the plus. Simplify.
27176 (A - (-B) -> A + B): Likewise.
27177 * fold-const.c (split_tree): Make sure to not negate pointers.
27179 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27181 PR rtl-optimization/60818
27182 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27183 a compare of comparisons with the thing compared if this results
27184 in a different machine mode.
27186 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27188 * alias.c (base_alias_check): Fix typo in comment.
27189 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27190 * cgraphunit.c (symbol_table::compile): Likewise.
27191 * collect2.c (maybe_run_lto_and_relink): Likewise.
27192 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27193 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27194 * config/avr/avr.c (avr_map_op_t): Likewise.
27195 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27196 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27197 * config/epiphany/epiphany.md (movcc): Likewise.
27198 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27199 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27201 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27202 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27203 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27204 * config/sh/sh.c (sh_rtx_costs): Likewise.
27205 * fold-const.c (fold_truth_andor): Likewise.
27206 * genautomata.c (collapse_flag): Likewise.
27207 * gengtype.h (struct type::u::s): Likewise.
27208 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27209 * input.c (FORMAT_AMOUNT): Likewise.
27210 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27211 (known_aggs_to_agg_replacement_list): Likewise.
27212 * ipa-inline-analysis.c: Likewise.
27213 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27214 * ipa-polymorphic-call.c
27215 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27216 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27217 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27219 * modulo-sched.c (apply_reg_moves): Likewise.
27220 * omp-expand.c (build_omp_regions_1): Likewise.
27221 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27222 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27223 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27224 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27225 * value-prof.c: Likewise.
27226 * var-tracking.c (val_reset): Likewise.
27228 2017-04-03 Richard Biener <rguenther@suse.de>
27230 PR tree-optimization/80275
27231 * fold-const.c (split_address_to_core_and_offset): Handle
27234 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27236 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27237 descriptors is at least equal to that of functions.
27239 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27241 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27243 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27246 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27247 (mov<IMOD4:mode>): New expander.
27248 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27250 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27252 PR rtl-optimization/79405
27253 * fwprop.c (propagations_left): New variable.
27254 (forward_propagate_into): Decrement it.
27255 (fwprop_init): Initialize it.
27256 (fw_prop): If the variable has reached zero, stop propagating.
27257 (fwprop_addr): Ditto.
27259 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27262 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27263 a FUNCTION_DECL, pass it as decl instead of origin to
27266 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27268 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27271 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27274 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27275 TARGET_VSX_SMALL_INTEGER.
27277 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27279 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27280 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27282 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27284 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27285 extraction from odd-numbered MSA register.
27287 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27289 PR middle-end/80173
27290 * expmed.c (store_bit_field_1): Don't attempt to create
27291 a word subreg out of hard registers wider than word if they
27292 have HARD_REGNO_NREGS of 1 for their mode.
27294 PR middle-end/80163
27295 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27296 conversions to integer types wider than word and pointer.
27299 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27300 (rtx_equal_for_cselib_p): Pass 0 to it.
27301 * cselib.c (cselib_hasher::equal): Likewise.
27302 (rtx_equal_for_cselib_1): Add depth argument. If depth
27303 is 128, don't look up VALUE locs and punt. Increment
27304 depth in recursive calls when walking VALUE locs.
27306 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27308 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27309 (make_gcov_file_name): Use the canonical path name for generating
27311 (read_line): Fix handling of files with ascii null bytes.
27313 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27315 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27316 to initialise a vector register instead
27317 of using a const_int.
27319 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27321 PR translation/80189
27322 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27323 diagnostic messages.
27325 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27328 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27329 (dfp_diex_<mode>): Update mode of operand 1.
27330 * doc/extend.texi (dxex, dxexq): Document change to return type.
27331 (diex, diexq): Document change to argument type.
27333 2017-03-30 Martin Jambor <mjambor@suse.cz>
27336 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27337 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27338 it reflects the signature changes performed at the callee side.
27339 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27340 to cgraph_build_function_type_skip_args.
27341 (build_function_decl_skip_args): Adjust call to the above function.
27343 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27346 * config/i386/sse.md
27347 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27348 register as dest whenever it is a MEM not rtx_equal_p to the
27349 corresponding dup operand, and when forcing into reg move the
27350 reg into the memory afterwards.
27351 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27352 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27353 for the force_reg mode.
27354 (avx512vl_vextractf128<mode>): Use register as dest either
27355 always when a MEM, or when it is a MEM not rtx_equal_p to the
27356 corresponding dup operand, or even not when it is a CONST_VECTOR
27357 depending on the mode and lo vs. hi.
27358 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27360 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27361 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27362 Likewise. Require that operands[2] is even.
27363 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27364 Remove extraneous parens. Require that operands[2] is a multiple
27366 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27367 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27368 disallow memory then.
27370 2017-03-30 Richard Biener <rguenther@suse.de>
27372 PR tree-optimization/77498
27373 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27374 to non-constants over backedges.
27376 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27378 PR rtl-optimization/80233
27379 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27380 as last_combined_insn. Do not test for BARRIER_P separately.
27382 2017-03-29 Andreas Schwab <schwab@suse.de>
27385 * calls.c (prepare_call_address): Convert funexp to Pmode before
27386 copying to temp reg.
27388 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27390 PR tree-optimization/80158
27391 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27392 Handle possible future case of more than one alternate
27394 (replace_rhs_if_not_dup): Likewise.
27395 (replace_one_candidate): Likewise.
27397 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27399 PR rtl-optimization/80193
27400 * ira.c (ira): Do not check allocation for LRA.
27402 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27404 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27405 (nvptx_output_simt_exit): Declare.
27406 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27407 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27408 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27409 (nvptx_declare_function_name): ...here. Emit declaration of local
27410 memory space buffer for omp_simt_enter insn.
27411 (nvptx_output_unisimt_switch): New.
27412 (nvptx_output_softstack_switch): New.
27413 (nvptx_output_simt_enter): New.
27414 (nvptx_output_simt_exit): New.
27415 * config/nvptx/nvptx.h (struct machine_function): New fields
27416 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27417 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27418 (UNSPECV_SIMT_EXIT): Ditto.
27419 (omp_simt_enter_insn): New insn.
27420 (omp_simt_enter): New expansion.
27421 (omp_simt_exit): New insn.
27422 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27424 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27425 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27426 (expand_GOMP_SIMT_EXIT): New.
27427 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27428 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27429 (GOMP_SIMT_EXIT): Ditto.
27430 * target-insns.def (omp_simt_enter): New insn.
27431 (omp_simt_exit): Ditto.
27432 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27434 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27435 (lower_rec_input_clauses): Likewise.
27436 (lower_lastprivate_clauses): Handle SIMT privatization.
27438 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27439 (ompdevlow_adjust_simt_enter): New.
27440 (find_simtpriv_var_op): New.
27441 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27442 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27444 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27445 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27446 (copy_decl_for_dup_finish): Ditto.
27448 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27450 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27453 * config/i386/i386.c (ix86_option_override_internal): Always
27454 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27456 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27458 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27460 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27462 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27463 mark new edge's irreducible flag accordign to it.
27464 (vect_do_peeling): Check loop preheader edge's irreducible flag
27465 and pass it to function slpeel_add_loop_guard.
27467 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27469 PR tree-optimization/80218
27470 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27471 Update block frequencies and counts.
27473 2017-03-28 Richard Biener <rguenther@suse.de>
27475 PR tree-optimization/78644
27476 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27477 of a simplification result we may not use it at all.
27479 2017-03-28 Richard Biener <rguenther@suse.de>
27482 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27483 without arguments, generate default definition of a SSA name.
27485 2017-03-28 Richard Biener <rguenther@suse.de>
27487 PR middle-end/80222
27488 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27489 TYPE_REF_CAN_ALIAS_ALL references.
27490 * fold-const.c (fold_indirect_ref_1): Likewise.
27492 2017-03-28 Martin Liska <mliska@suse.cz>
27495 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27496 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27498 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27499 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27501 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27502 (EXTRA_SPECS): Define.
27503 (SUBTARGET_EXTRA_SPECS): Likewise.
27504 (SUBTARGET_CPP_SPEC): Likewise.
27505 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27506 SUBTARGET_EXTRA_SPECS.
27507 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27509 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27511 * config/arc/simdext.md (vst64_insn): Update pattern.
27512 (vld32wh_insn): Likewise.
27513 (vld32wl_insn): Likewise.
27514 (vld64_insn): Likewise.
27515 (vld32_insn): Likewise.
27517 2017-03-28 Marek Polacek <polacek@redhat.com>
27520 * fold-const.c (fold_comparison): Use protected_set_expr_location
27521 instead of SET_EXPR_LOCATION.
27523 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27525 * tree.c (add_expr): Avoid name lookup warning.
27527 2017-03-27 Jeff Law <law@redhat.com>
27529 PR tree-optimization/80216
27530 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27531 function name. Limit recursion depth.
27532 (record_temporary_equivalences): Corresponding changes.
27534 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27536 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27539 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27542 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27544 * cfgcleanup.c (reg_note_cfa_p): New array.
27545 (insns_have_identical_cfa_notes): New function.
27546 (old_insns_match_p): Don't cross-jump in between /f
27547 and non-/f instructions. If both i1 and i2 are frame related,
27548 verify all CFA notes, their order and content.
27550 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27553 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27554 HImode and SImode with zero extend to DImode to one insn.
27555 (bswap<mode>2_extenddi): Likewise.
27556 (bswapsi2_extenddi): Likewise.
27557 (bswaphi2_extendsi): Likewise.
27558 (bswaphi2): Combine bswap HImode and SImode into one insn.
27559 Separate memory insns from swapping register.
27560 (bswapsi2): Likewise.
27561 (bswap<mode>2): Likewise.
27562 (bswaphi2_internal): Delete, no longer used.
27563 (bswapsi2_internal): Likewise.
27564 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27565 store, and gpr<-gpr swap insns.
27566 (bswap<mode>2_store): Likewise.
27567 (bswaphi2_reg): Register only splitter, combine with the splitter.
27568 (bswaphi2 splitter): Likewise.
27569 (bswapsi2_reg): Likewise.
27570 (bswapsi2 splitter): Likewise.
27571 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27572 the insns into load, store, and register/register insns.
27573 (bswapdi2_ldbrx): Likewise.
27574 (bswapdi2_load): Likewise.
27575 (bswapdi2_store): Likewise.
27576 (bswapdi2_reg): Likewise.
27578 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27580 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27581 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27583 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27586 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27588 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27589 special handling for target option conflicts between dform
27590 options (-mpower9-dform, -mpower9-dform-vector,
27591 -mpower9-dform-scalar) and -mno-direct-move.
27593 2017-03-27 Richard Biener <rguenther@suse.de>
27595 PR tree-optimization/80181
27596 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27598 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27600 * config/arc/predicates.md (move_double_src_operand): Replace the
27601 call to move_double_src_operand with a call to address_operand.
27603 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27605 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27606 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27607 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27609 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27611 * config/arc/predicates.md (long_immediate_loadstore_operand):
27612 Consider scaled addresses cases.
27614 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27616 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27617 restored when in interrupt.
27618 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27619 doesn't have delay slot.
27621 2017-03-27 Richard Biener <rguenther@suse.de>
27624 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27625 inlined thunk clones.
27627 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27630 * asan.c (instrument_derefs): Copy over last operand from
27631 original COMPONENT_REF to the new COMPONENT_REF with
27632 DECL_BIT_FIELD_REPRESENTATIVE.
27633 * ubsan.c (instrument_object_size): Likewise.
27635 2017-03-27 Richard Biener <rguenther@suse.de>
27637 PR tree-optimization/80170
27638 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27639 sure DR/SCEV didnt fold in constants we do not see when looking
27640 at the reference base alignment.
27642 2017-03-27 Richard Biener <rguenther@suse.de>
27644 PR middle-end/80171
27645 * gimple-fold.c (fold_ctor_reference): Properly guard against
27646 NULL return value from canonicalize_constructor_val.
27648 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27651 * config/i386/i386.c (ix86_expand_builtin)
27652 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27653 flags reg setting and flags reg using instructions.
27654 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27655 clobbering instructions to zero extend op2.
27657 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27659 * doc/install.texi (Configuration) <--with-aix-soname>:
27660 Update link to AIX ld.
27662 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27664 PR rtl-optimization/80160
27665 PR rtl-optimization/80159
27666 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27667 reg_alternate_class into account.
27669 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27672 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27673 to consider in curr_insn_transform.
27675 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27677 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27678 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27679 and emit_mode_inner.
27681 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27683 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27684 argument to the overloaded builtin variants. Use the new flag to
27685 deprecate certain builtin variants.
27686 * config/s390/s390-builtin-types.def: Add new builtin types.
27687 * config/s390/s390-builtins.h: Support new flags field for
27688 overloaded builtins.
27689 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27690 (s390_macro_to_expand): Enable vector float data type.
27691 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27692 builtins by incrementing the __VEC__ version number.
27693 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27695 (s390_resolve_overloaded_builtin): Emit error messages depending
27696 on the builtin flags.
27697 * config/s390/s390.c (s390_expand_builtin): Support additional
27698 flags argument. Change error message to match the messages
27699 emitted in s390-c.c.
27700 * config/s390/s390.md: New UNSPEC_* constants.
27701 (op_type): Add new instruction types.
27702 * config/s390/vecintrin.h: Add new builtins and test data class
27704 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27705 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27706 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27707 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27708 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27709 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27711 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27712 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27713 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27714 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27716 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27717 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27718 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27719 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27721 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27722 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27723 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27724 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27725 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27726 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27727 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27729 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27730 ("vec_scatter_element<V_HW_4:mode>_DI")
27731 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27732 ("vec_fpint<mode>", "vflls")
27733 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27734 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27735 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27736 ("*vec_cmphe<mode>_cc"): ... these.
27738 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27739 mode constant instead of magic value.
27741 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27743 * config/s390/s390.c (s390_expand_vec_compare): Support other
27744 vector floating point modes than just V2DF.
27745 (s390_expand_vcond): Likewise.
27746 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27747 (s390_cannot_change_mode_class): Prevent mode changes between TF
27748 and V1TF in vector registers.
27749 * config/s390/s390.md (DF, SF): New mode attributes.
27750 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27751 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27752 SFmode support for VRs.
27753 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27755 (VFT, VF_HW): New mode iterators.
27756 (vw, sdx): New mode attributes.
27757 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27758 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27759 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27760 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27761 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27762 also the new vector floating point modes. Renaming to ...
27764 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27765 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27766 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27767 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27768 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27769 ("vec_unordered<mode>"): ... these.
27771 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27772 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27773 ("*vec_extendv2df"): New insn definitions.
27775 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27777 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27778 ("mulditi3_2", "*muldi3_sign"): New patterns.
27779 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27780 rename the pattern definition.
27782 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27784 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27786 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27788 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27790 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27791 instruction if possible.
27792 * config/s390/vector.md (vec_halfnumelts): New mode
27794 ("*vec_vllezlf<mode>"): New pattern.
27796 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27798 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27799 ("popcountv4si2", "popcountv2di2"): Rename to ...
27800 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27801 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27803 ("popcount<mode>2_vxe"): New pattern.
27805 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27807 * common/config/s390/s390-common.c (processor_flags_table): Add
27809 * config.gcc: Add arch12.
27810 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27811 Default to arch12 for unknown CPU model numbers.
27812 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27813 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27814 PROCESSOR_max sanity check.
27815 * config/s390/s390-opts.h (enum processor_type): Add
27817 * config/s390/s390.c (processor_table): Add arch12.
27818 (s390_expand_builtin): Add check for B_VXE flag.
27819 (s390_issue_rate): Add PROCESSOR_ARCH12.
27820 (s390_get_sched_attrmask): Likewise.
27821 (s390_get_unit_mask): Likewise.
27822 (s390_sched_score): Enable z13 scheduling for arch12.
27823 (s390_sched_reorder): Likewise.
27824 (s390_sched_variable_issue): Likewise.
27825 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27827 (s390_tune_attr): Use z13 scheduling also for arch12.
27828 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27829 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27830 (TARGET_VXE_P): New macros.
27831 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27832 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27833 * config/s390/s390.opt: Add arch12 as processor_type.
27835 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27837 * config/s390/s390.md
27838 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27839 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27840 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27842 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27843 Rename expanders to ...
27845 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27846 ("fixuns_truncdddi2_emu"): ... these.
27848 ("fixuns_trunc<mode>si2_emu"): New expander.
27850 ("*fixuns_truncdfdi2_z13"): Rename to ...
27851 ("*fixuns_truncdfdi2_vx"): ... this.
27853 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27855 * config/s390/2964.md: Remove the single element vector compare
27856 instructions which are no longer used.
27857 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27859 (s390_canonicalize_comparison): Remove handling of DFmode
27861 (s390_expand_vec_compare_scalar): Remove function.
27862 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27863 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27865 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27867 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27869 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27871 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27872 will anyway by matched by mov<mode>_64dfp.
27874 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27876 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27877 vlef/vstef. Add missing operand to vleif.
27879 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27881 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27882 pair for all vector types with 64 bit elements.
27883 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27884 * config/s390/vector.md (V_HW_64): ... here.
27885 (V_128_NOSINGLE): New mode iterator.
27886 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27887 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27888 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27889 ("*vec_load_pairv2di"): Change to ...
27890 ("*vec_load_pair<mode>"): ... this one.
27892 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27894 * config/s390/constraints.md: Add comments.
27895 (jKK): Reject element sizes > 8 bytes.
27896 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27898 * config/s390/s390.md: Add the s_operand checks formerly in
27899 s390_split_ok_p to various splitters where they are still
27901 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27902 for 128 bit vectors. Plus two splitters.
27904 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27906 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27909 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27912 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27913 error if the boundary argument is not constant.
27915 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27917 PR rtl-optimization/80112
27918 * loop-doloop.c (doloop_condition_get): Don't check condition
27919 if cmp isn't SET with IF_THEN_ELSE src.
27921 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27923 PR tree-optimization/80158
27924 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27925 replacing a candidate statement, also replace it for the
27926 candidate's alternate interpretation.
27927 (replace_rhs_if_not_dup): Likewise.
27928 (replace_one_candidate): Likewise.
27930 2017-03-24 Richard Biener <rguenther@suse.de>
27932 PR tree-optimization/80167
27933 * graphite-isl-ast-to-gimple.c
27934 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27936 (translate_isl_ast_to_gimple::get_rename): Likewise.
27938 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27940 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27941 handling of certain combinations of target options, including the
27942 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27943 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27945 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27948 * config/arm/arm.md (*load_multiple): Add reload_completed to
27949 matching condition.
27951 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27952 Richard Biener <rguenth@suse.de>
27954 PR tree-optimization/79908
27955 PR tree-optimization/80136
27956 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27957 been cast away, gimplify_and_add suffices.
27959 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27961 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27963 2017-03-23 Richard Biener <rguenther@suse.de>
27965 PR tree-optimization/80032
27966 * gimplify.c (gimple_push_cleanup): Forced unconditional
27967 cleanups still have to go to the conditional_cleanups
27970 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27972 PR tree-optimization/80072
27973 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27975 (next_operand_entry_id): Change type to unsigned int.
27976 (sort_by_operand_rank): Make sure to return the right return value
27977 even if unsigned fields are bigger than INT_MAX.
27978 (struct oecount): Change cnt and id type to unsigned int.
27979 (oecount_hasher::equal): Formatting fix.
27980 (oecount_cmp): Make sure to return the right return value
27981 even if unsigned fields are bigger than INT_MAX.
27982 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27985 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27986 TREE_READONLY on result if writing it more than once.
27989 * doc/invoke.texi (-fsanitize=thread): Document that with
27990 -fnon-call-exceptions atomics are not able to throw
27994 * tsan.c: Include tree-eh.h.
27995 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27996 maybe_clean_or_replace_eh_stmt where needed.
27997 (instrument_memory_accesses): Add cfg_changed argument.
27998 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27999 if it returned true.
28000 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28002 PR rtl-optimization/63191
28003 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28004 wrapper function, moved the whole old content into ...
28005 (ix86_delegitimize_address_1): ... this. New inline function.
28006 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28007 true as last argument instead of ix86_delegitimize_address.
28009 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28011 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28012 cortexa57_branch_cost.
28014 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28016 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28018 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28021 * doc/md.texi (Constraints): Document wA constraint.
28022 * config/rs6000/constraints.md (wA): New.
28023 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28024 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28025 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28026 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28028 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28031 * gimplify.c (is_oacc_declared): New function.
28032 (oacc_default_clause): Use it to set default flags for acc declared
28033 variables inside parallel regions.
28034 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28035 declared variables.
28036 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28037 declare attribute to any decl as necessary.
28039 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28042 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28043 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28044 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28045 (arm_arch_lpae): This.
28046 * config/arm/arm.c (arm_arch7ve): Rename into ...
28047 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28048 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28051 2017-03-22 Martin Liska <mliska@suse.cz>
28054 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28055 error message instead of an ICE.
28057 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28059 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28061 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28063 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28065 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28068 2017-03-21 Martin Sebor <msebor@redhat.com>
28070 * doc/extend.texi: Use "cannot" instead of "can't."
28071 * doc/hostconfig.texi: Same.
28072 * doc/install.texi: Same.
28073 * doc/invoke.texi: Same.
28074 * doc/loop.texi: Same.
28075 * doc/md.texi: Same.
28076 * doc/objc.texi: Same.
28077 * doc/rtl.texi: Same.
28078 * doc/tm.texi: Same.
28079 * doc/tm.texi.in: Same.
28080 * doc/trouble.texi: Same.
28082 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28085 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28086 (collect_checksum_attributes): Set it.
28087 (die_checksum_ordered): Use it.
28089 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28091 PR tree-optimization/79908
28092 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28093 change: For a VA_ARG whose LHS has been cast away, use
28094 force_gimple_operand to construct the side effects.
28096 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28098 PR translation/80001
28099 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28100 more amenable to translation.
28101 (oacc_loop_auto_partitions): Likewise.
28103 2017-03-21 Marek Polacek <polacek@redhat.com>
28104 Martin Sebor <msebor@redhat.com>
28106 PR tree-optimization/80109
28107 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28108 on INTEGRAL_TYPE_P.
28110 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28111 Segher Boessenkool <segher@kernel.crashing.org>
28114 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28115 check reg_used_between_p between insn and one of succ or succ2
28116 depending on if succ is artificial insn not inserted into insn
28119 2017-03-21 Martin Liska <mliska@suse.cz>
28121 PR gcov-profile/80081
28122 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28123 * doc/gcc.texi: Include gcov-dump stuff.
28124 * doc/gcov-dump.texi: New file.
28126 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28128 PR rtl-optimization/79150
28129 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28130 conditional jump, if the jump is the last insn of the loop.
28132 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28133 Richard Biener <rguenth@suse.de>
28135 PR tree-optimization/79908
28136 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28137 been cast away, use force_gimple_operand to construct the side
28140 2017-03-21 Martin Liska <mliska@suse.cz>
28142 PR libfortran/79956
28143 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28146 2017-03-21 Brad Spengler <spender@grsecurity.net>
28149 * plugin.c (htab_hash_plugin): New function.
28150 (add_new_plugin): Use it and adjust.
28151 (parse_plugin_arg_opt): Adjust.
28152 (init_one_plugin): Likewise.
28154 2017-03-21 Richard Biener <rguenther@suse.de>
28156 PR tree-optimization/80032
28157 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28158 if set force the cleanup to happen unconditionally.
28159 (gimplify_target_expr): Push inserted clobbers with force_uncond
28160 to avoid them being removed by control-dependent DCE.
28162 2017-03-21 Richard Biener <rguenther@suse.de>
28164 PR tree-optimization/80122
28165 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28166 va_arg_pack_len when the inlined call stmt requires pack
28168 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28170 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28173 * tsan.c (instrument_builtin_call): If the memory model argument
28174 is not a constant, assume it is valid.
28177 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28180 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28182 PR rtl-optimization/79910
28183 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28184 if its dest is used by an insn before I2 (other than the combined
28185 insns themselves, which are properly handled already).
28187 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28190 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28192 * combine.c (record_used_regs): New static function.
28193 (try_combine): Handle situations where there is an additional
28194 instruction between I2 and I3 which needs to have a LOG_LINK
28198 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28200 * combine.c (try_combine): Delete redundant i1 test. Call
28201 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28203 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28206 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28207 alternatives 13/14.
28209 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28211 PR tree-optimization/80054
28212 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28213 the optimization if a PHI or any of its arguments is not dominated
28214 by the candidate's basis. Use gphi* rather than gimple* as
28216 (replace_profitable_candidates): Clean up a gimple* variable that
28217 should be a gphi* variable.
28219 2017-03-20 Martin Sebor <msebor@redhat.com>
28222 * doc/extend.texi (attribute constructor): Document present limitation.
28224 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28227 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28228 __POWER9_VECTOR__ #ifdef control, change template definition to
28229 use Power9-specific built-in function.
28230 (vec_any_eq): Likewise.
28231 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28232 to control outcomes from this test.
28233 (vector_ae_<mode>p): For VEC_F modes, likewise.
28235 2017-03-20 Ian Lance Taylor <iant@google.com>
28237 * config/i386/i386.c (ix86_function_regparm): Save an extra
28238 register for -fsplit-stack with DECL_STATIC_CHAIN.
28240 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28243 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28244 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28246 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28248 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28250 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28253 2017-03-20 Marek Polacek <polacek@redhat.com>
28256 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28258 2017-03-20 Richard Biener <rguenther@suse.de>
28260 PR tree-optimization/80113
28261 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28262 allocate extra SSA name for PHI def.
28263 (add_close_phis_to_outer_loops): Likewise.
28264 (add_close_phis_to_merge_points): Likewise.
28265 (copy_loop_close_phi_args): Likewise.
28266 (copy_cond_phi_nodes): Likewise.
28268 2017-03-20 Martin Liska <mliska@suse.cz>
28270 PR middle-end/79753
28271 * tree-chkp.c (chkp_build_returned_bound): Do not build
28272 returned bounds for a LHS that's not a BOUNDED_P type.
28274 2017-03-20 Martin Liska <mliska@suse.cz>
28278 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28279 COMPLEX_CST and VECTOR_CST.
28281 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28284 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28285 target operand. A new splitter adds the clobber statement in case
28286 the target operand is dead anyway.
28288 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28290 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28291 to age-old versions of binutils and glibc.
28293 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28295 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28297 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28299 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28301 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28303 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28304 requirement for binutils 2.13.
28306 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28308 * combine.c (try_combine): Delete redundant i1 test. Call
28309 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28311 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28313 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28314 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28316 <riscv64-*-elf>: Re-arrange section
28317 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28318 <riscv32-*-linux>: Likewise.
28319 <riscv64-*-elf>: Likewise
28320 <riscv64-*-linux>: Likewise.
28322 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28325 * aarch64.opt(verbose-cost-dump): Fix typo.
28327 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28330 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28331 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28333 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28335 * reload.c (find_reloads): When reloading a nonoffsettable address,
28336 use RELOAD_OTHER for it and its address reloads.
28338 PR rtl-optimization/79910
28339 * combine.c (record_used_regs): New static function.
28340 (try_combine): Handle situations where there is an additional
28341 instruction between I2 and I3 which needs to have a LOG_LINK
28344 2017-03-17 Jeff Law <law@redhat.com>
28346 PR tree-optimization/71437
28347 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28348 conditional in the hash table first.
28349 (vrp_dom_walker::before_dom_children): Extract condition from
28350 ASSERT_EXPR. Record condition, its inverion and any implied
28351 conditions as well.
28353 2017-03-17 Marek Polacek <polacek@redhat.com>
28354 Markus Trippelsdorf <markus@trippelsdorf.de>
28356 PR tree-optimization/80079
28357 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28360 2017-03-17 Richard Biener <rguenther@suse.de>
28362 PR middle-end/80075
28363 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28364 Properly verify the LHS before the RHS possibly claims to be
28366 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28369 2017-03-17 Martin Jambor <mjambor@suse.cz>
28371 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28372 (List of -O2 options): Likewise.
28373 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28376 2017-03-17 Tom de Vries <tom@codesourcery.com>
28378 * gcov-dump.c (print_usage): Print bug_report_url.
28380 2017-03-17 Richard Biener <rguenther@suse.de>
28382 PR middle-end/80050
28383 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28384 (parser::peek): Likewise.
28386 2017-03-17 Richard Biener <rguenther@suse.de>
28388 PR tree-optimization/80048
28389 * sese.c (free_sese_info): Properly release rename_map and
28390 copied_bb_map elements.
28392 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28394 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28395 Add linked-list forward and backlinks. Insert on
28396 construction, remove on destruction.
28397 (class pass_store_merging): Add m_stores_head field.
28398 (pass_store_merging::terminate_and_process_all_chains):
28399 Iterate over m_stores_head list.
28400 (pass_store_merging::terminate_all_aliasing_chains):
28402 (pass_store_merging::execute): Check for debug stmts first.
28403 Push new chains onto the m_stores_head stack.
28405 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28408 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28409 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28410 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28412 2017-03-16 Jeff Law <law@redhat.com>
28414 PR tree-optimization/71437
28415 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28416 member function. Implementation moved into after_dom_children
28417 member function and into the threader's thread_outgoing_edges
28419 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28420 some code into new thread_outgoing_edges.
28421 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28422 definition. Simplify marker handling (do it here). Assume we always
28423 have the available expression and the const/copies tables.
28424 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28426 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28427 * tree-vrp.c (equiv_stack): No longer file scoped.
28428 (vrp_dom_walker): New class.
28429 (vrp_dom_walker::before_dom_children): New member function.
28430 (vrp_dom_walker::after_dom_children): Likewise.
28431 (identify_jump_threads): Setup domwalker. Use it rather than
28432 walking edges in a random order by hand. Simplify setup/finalization.
28433 (finalize_jump_threads): Remove.
28434 (vrp_finalize): Do not call identify_jump_threads here.
28435 (execute_vrp): Do it here instead and call thread_through_all_blocks
28438 PR tree-optimization/71437
28439 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28441 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28443 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28444 (dom_opt_dom_walker::thread_across_edge): Remove
28445 handle_dominating_asserts argument. All callers changed.
28446 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28447 changes. Remove calls to lhs_of_dominating_assert. Other
28448 uses of handle_dominating_asserts turn into unconditional code
28449 (simplify_control_stmt_condition_1): Likewise.
28450 (simplify_control_stmt_condition): Likewise.
28451 (thread_through_normal_block, thread_across_edge): Likewise.
28452 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28453 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28454 object if it is not an SSA_NAME.
28455 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28456 before calling into the VRP specific simplifiers.
28457 (identify_jump_threads): Remove handle_dominating_asserts
28460 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28463 * tree-diagnostic.c (default_tree_printer): No longer static.
28464 * tree-diagnostic.h (default_tree_printer): New prototype.
28466 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28468 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28469 Change ins into fmov.
28471 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28473 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28474 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28475 Use h_con constraint for operand 1.
28476 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28477 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28479 2017-03-15 Jeff Law <law@redhat.com>
28481 PR tree-optimization/71437
28482 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28483 (record_temporary_equivalences): Use it.
28485 PR tree-optimization/71437
28486 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28487 tree-ssa-scopedtables.
28488 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28489 (record_conditions, record_cond, vuse_eq): Likewise.
28490 (record_edge_info): Adjust to API tweak of record_conditions.
28491 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28492 (record_temporary_equivalences, optimize_stmt): Likewise.
28493 (eliminate_redundant_computations): Likewise.
28494 (record_equivalences_from_stmt): Likewise.
28495 * tree-ssa-scopedtables.c: Include options.h and params.h.
28496 (vuse_eq): New function, moved from tree-ssa-dom.c
28497 (build_and_record_new_cond): Likewise.
28498 (record_conditions): Likewise. Accept vector of conditions rather
28499 than edge_equivalence structure for first argument.
28500 for the first argument.
28501 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28502 from tree-ssa-dom.c.
28503 (avail_exprs_stack::record_cond): Likewise.
28504 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28505 from tree-ssa-dom.c.
28506 (avail_exprs_stack): Add new member functions lookup_avail_expr
28508 (record_conditions): Declare.
28510 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28513 * lra-constraints.c (process_alt_operands): Increase reject for
28514 reloading an input/output operand.
28516 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28519 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28520 insns to convert from signed/unsigned char/short to IEEE 128-bit
28522 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28524 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28527 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28528 subreg of inner mode for values already in registers.
28530 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28532 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28533 iteration reg is used after the loop.
28535 2017-03-14 Martin Sebor <msebor@redhat.com>
28537 PR tree-optimization/79800
28538 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28539 precision in negative-positive range.
28540 (format_floating): Call non-const overload with adjusted precision.
28542 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28545 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28548 2017-03-14 Martin Sebor <msebor@redhat.com>
28550 PR middle-end/80020
28551 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28552 * builtins.def (aligned_alloc): Use it.
28555 * Makefile.in (GTFILES): Add calls.c.
28556 * calls.c: Include "gt-calls.h".
28558 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28560 PR rtl-optimization/79728
28561 * regs.h (struct target_regs): New field
28562 x_contains_allocatable_regs_of_mode.
28563 (contains_allocatable_regs_of_mode): New macro.
28564 * reginfo.c (init_reg_sets_1): Initialize it, and change
28565 contains_reg_of_mode so it includes global regs as well.
28566 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28567 rather than contains_regs_of_mode.
28569 2017-03-14 Martin Liska <mliska@suse.cz>
28571 * doc/invoke.texi: Document options that can't be combined with
28572 -fcheck-pointer-bounds.
28574 2017-03-14 Martin Liska <mliska@suse.cz>
28576 PR middle-end/79831
28577 * doc/invoke.texi (-Wchkp): Document the option.
28579 2017-03-14 Martin Liska <mliska@suse.cz>
28581 * Makefile.in: Install gcov-dump.
28583 2017-03-14 Martin Liska <mliska@suse.cz>
28585 * multiple_target.c (expand_target_clones): Bail out for
28586 an invalid attribute.
28588 2017-03-14 Richard Biener <rguenther@suse.de>
28590 * alias.c (struct alias_set_entry): Pack properly.
28591 * cfgloop.h (struct loop): Likewise.
28592 * cse.c (struct set): Likewise.
28593 * ipa-utils.c (struct searchc_env): Likewise.
28594 * loop-invariant.c (struct invariant): Likewise.
28595 * lra-remat.c (struct cand): Likewise.
28596 * recog.c (struct change_t): Likewise.
28597 * rtl.h (struct address_info): Likewise.
28598 * symbol-summary.h (function_summary): Likewise.
28599 * tree-loop-distribution.c (struct partition): Likewise.
28600 * tree-object-size.c (struct object_size_info): Likewise.
28601 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28602 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28603 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28604 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28605 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28606 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28607 (struct _stmt_vec_info): Likewise.
28609 2017-03-14 Martin Liska <mliska@suse.cz>
28612 * multiple_target.c (create_dispatcher_calls): Check that
28613 a target can create a function dispatcher.
28615 2017-03-14 Martin Liska <mliska@suse.cz>
28618 * multiple_target.c (expand_target_clones): Drop local.local
28619 flag for default implementation.
28621 2017-03-14 Richard Biener <rguenther@suse.de>
28623 PR tree-optimization/80030
28624 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28626 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28628 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28629 gcc_fallthrough() instead of __attribute__((fallthrough));
28631 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28633 * doc/gcc.texi: Remove "up" link to (DIR).
28634 * doc/gccint.texi: Ditto.
28636 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28638 * doc/install.texi (Specific) <avr>: Remove reference to
28641 2017-03-13 Jeff Law <law@redhat.com>
28643 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28644 attribute rather than comments.
28646 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28647 match_scratch operand is highest.
28649 2017-03-13 Martin Liska <mliska@suse.cz>
28651 PR middle-end/78339
28652 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28653 is a CHKP clone, use original declaration.
28655 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28657 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28658 (arc_conditional_register_usage): Use a different allocation order
28659 when optimizing for size.
28660 * common/config/arc/arc-common.c (arc_option_optimization_table):
28661 Section anchors default on when optimizing for size.
28663 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28665 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28667 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28669 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28670 * config/arc/arc.md (cpu_facility): Add cd variant.
28671 (*movqi_insn): Add code density variant.
28672 (*movhi_insn): Likewise.
28673 (*movqi_insn): Likewise.
28674 (*addsi3_mixed): Likewise.
28675 (subsi3_insn): Likewise.
28677 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28679 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28681 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28683 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28684 expressions with MINUS and UNARY ops.
28686 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28689 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28691 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28692 between vec_select and vector argument.
28693 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28694 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28695 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28696 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28697 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28698 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28700 2017-03-13 Richard Biener <rguenther@suse.de>
28703 * params.def (vect-max-peeling-for-alignment): Fix typo.
28705 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28707 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28708 issue that only occurred with binutils below 2.18.
28710 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28712 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28713 refer to binutils 2.11/2.12 minimum.
28715 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28717 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28718 ftp.kernel.org and simplify binutils requirement.
28720 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28722 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28724 (Optimize Options): Ditto. Also remove redundancy.
28726 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28728 PR translation/79848
28729 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28731 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28732 to G_ to avoid double translation.
28734 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28736 PR translation/79923
28737 * auto-profile.c (get_combined_location): Convert leading
28738 character of diagnostics to lower case and remove trailing period.
28739 (read_profile): Likewise for various diagnostics.
28740 * config/arm/arm.c (arm_option_override): Remove trailing period
28741 from various diagnostics.
28742 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28743 (msp430_expand_delay_cycles): Likewise.
28745 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28748 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28749 full command-line argument, rather than just "str".
28750 (aarch64_validate_march): Likewise.
28751 (aarch64_validate_mtune): Likewise.
28753 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28755 PR rtl-optimization/78911
28756 * lra-assigns.c (must_not_spill_p): New function.
28757 (spill_for): Use it.
28759 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28761 PR tree-optimization/79981
28762 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28763 ATOMIC_COMPARE_EXCHANGE ifn result.
28764 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28765 IFN_ATOMIC_COMPARE_EXCHANGE.
28767 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28770 * opts.c (parse_sanitizer_options): Add missing question mark to
28771 "did you mean" message.
28773 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28775 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28777 (VMULEUH_UNS): Likewise.
28778 (VMULOUB_UNS): Likewise.
28779 (VMULOUH_UNS): Likewise.
28780 * config/rs6000/rs6000.c (builtin_function_type): Remove
28781 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28783 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28786 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28787 x with result of extra_parsing_for_operand_code_0.
28788 (function_reader::extra_parsing_for_operand_code_0): Convert
28789 return type from void to rtx, returning x. When reading
28790 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28791 larger size containing struct block_symbol.
28793 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28795 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28796 -mfloat128-hardware without -m64.
28798 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28801 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28802 entries to the case statement that marks unsigned arguments to
28803 overloaded functions.
28805 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28807 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28808 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28810 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28813 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28814 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28816 2017-03-10 Martin Liska <mliska@suse.cz>
28820 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28821 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28824 2017-03-10 Olivier Hainque <hainque@adacore.com>
28826 * tree-switch-conversion (array_value_type): Start by resetting
28827 candidate type to it's main variant.
28829 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28831 PR rtl-optimization/79909
28832 * combine.c (try_combine): Use simplify_replace_rtx on individual
28833 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28834 of the whole CALL_INSN_FUNCTION_USAGE.
28836 PR tree-optimization/79972
28837 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28838 get_range_info on SSA_NAMEs. Formatting fixes.
28840 2017-03-10 Richard Biener <rguenther@suse.de>
28841 Jakub Jelinek <jakub@redhat.com>
28843 PR tree-optimization/77975
28844 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28845 edge to be constant.
28846 (get_val_for): For constant x return it. Formatting fix.
28847 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28848 would use the same bases as the current one.
28850 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28852 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28853 instead of vec_select for V1TImode.
28854 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28856 (VSX_LE_128): Add V1TI to this mode iterator.
28857 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28858 (*vsx_le_perm_store_<mode>): Likewise.
28859 (pre-reload splitter for VSX stores): Likewise.
28860 (post-reload splitter for VSX stores): Likewise.
28861 (*vsx_xxpermdi2_le_<mode>): Likewise.
28862 (*vsx_lxvd2x2_le_<mode>): Likewise.
28863 (*vsx_stxvd2x2_le_<mode>): Likewise.
28865 2017-03-09 Michael Eager <eager@eagercon.com>
28867 Correct failures with --enable-checking=yes,rtl.
28869 * config/microblaze/microblaze.c (microblaze_expand_shift):
28870 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28871 test for const0_rtx.
28872 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28873 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28875 2017-03-09 Richard Biener <rguenther@suse.de>
28877 PR tree-optimization/79977
28878 * graphite-scop-detection.c (scop_detection::merge_sese):
28879 Handle the case of extra exits to blocks dominating the entry.
28881 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28883 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28886 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28888 PR rtl-optimization/79949
28889 * lra-constraints.c (process_alt_operands): Check memory when
28890 trying to predict a cycle. Print about the overall increase.
28892 2017-03-09 Richard Biener <rguenther@suse.de>
28894 PR middle-end/79971
28895 * gimple-expr.c (useless_type_conversion_p): Preserve
28896 TYPE_SATURATING for fixed-point types.
28898 2017-03-09 Richard Biener <rguenther@suse.de>
28901 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28902 alignment of BLKmode params.
28904 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28907 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28908 (VALL_NO_V2Q): Likewise.
28910 * config/aarch64/aarch64-simd.md
28911 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28913 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28914 VALL_NO_V2Q mode iterator.
28915 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28917 2017-03-09 Martin Liska <mliska@suse.cz>
28919 PR tree-optimization/79631
28920 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28921 tree_int_cst_sign_bit just for INTEGER constants.
28923 2017-03-09 Martin Liska <mliska@suse.cz>
28927 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28930 2017-03-09 Marek Polacek <polacek@redhat.com>
28933 * tree.c (inchash::add_expr): Handle TREE_VEC.
28935 2017-03-09 Martin Liska <mliska@suse.cz>
28938 (chkp_narrow_size_and_offset): New function.
28939 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28940 (void chkp_parse_bit_field_ref): New function.
28941 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28942 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28944 2017-03-09 Martin Liska <mliska@suse.cz>
28947 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28948 (chkp_find_bounds_1): Remove gcc_unreachable.
28950 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28953 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28954 BUILT_IN_SYNC*, determine the access type from the size suffix and
28955 always build a MEM_REF with that type. Handle forgotten
28956 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28959 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28960 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28961 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28962 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28963 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28964 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28965 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28966 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28967 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28968 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28969 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28970 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28971 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28972 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28973 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28974 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28975 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28976 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28977 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28978 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28979 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28980 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28981 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28982 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28983 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28984 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28985 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28986 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28987 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28988 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28989 definitions outside of __OPTIMIZE__ guarded section.
28992 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28993 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28994 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28995 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28998 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29000 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29001 ("vfenez<mode>"): Add missing constraints.
29003 2017-03-08 Martin Sebor <msebor@redhat.com>
29006 * config/nds32/nds32.c (nds32_option_override):
29007 Fix misspelled diagnostic.
29009 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29012 * gimplify.c (gimplify_omp_for): Replace index var in outer
29013 taskloop statement with an artificial variable and add
29014 OMP_CLAUSE_PRIVATE clause for it.
29016 2017-03-08 Richard Biener <rguenther@suse.de>
29018 PR tree-optimization/79955
29019 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29020 for accesses that are completely outside of the variable.
29022 2017-03-08 Andrew Haley <aph@redhat.com>
29024 PR tree-optimization/79943
29025 * tree-ssa-loop-split.c (compute_new_first_bound): When
29026 calculating the new upper bound, (END-BEG) should be added, not
29029 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29031 * config/avr/avr.md (setmemhi): Make sure match_dup
29032 operand number comes before match_scratch.
29034 2017-03-08 Richard Biener <rguenther@suse.de>
29036 PR tree-optimization/79920
29037 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29038 with ncopies == 1 to ...
29039 (vect_transform_slp_perm_load): ... here. Properly compute
29040 all element loads by iterating VF times over the group. Do
29041 not handle ncopies (computed in a broken way) in
29042 vect_create_mask_and_perm.
29044 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29047 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29048 is a uniform vector, use uniform_vector_p return value instead of
29049 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29051 2017-03-07 Marek Polacek <polacek@redhat.com>
29053 PR middle-end/79809
29054 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29055 (alloca_call_type): Likewise.
29057 2017-03-07 Martin Liska <mliska@suse.cz>
29059 * gcov.c (process_args): Put comment to correct location.
29061 2017-03-07 Martin Liska <mliska@suse.cz>
29063 PR middle-end/68270
29064 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29065 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29066 (chkp_narrow_bounds_for_field): Likewise.
29067 (chkp_parse_array_and_component_ref): Pass one more argument to
29070 2017-03-07 Richard Biener <rguenther@suse.de>
29072 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29075 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29077 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29078 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29080 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29083 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29084 to end of description.
29085 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29087 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29089 PR rtl-optimization/79901
29090 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29092 (*avx512f_<code><mode>3<mask_name>): ... this.
29093 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29094 iterator instead of VI8_AVX2_AVX512BW.
29096 PR rtl-optimization/79901
29097 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29098 min/max expander, expand it using expand_vec_cond_expr.
29101 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29104 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29107 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29108 to void * for PCH reasons.
29109 * dwarf2out.c (output_loc_operands, output_die): Cast
29110 v.val_vec.array to unsigned char *.
29112 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29115 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29118 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29120 PR rtl-optimization/79571
29121 * lra-constraints.c (process_alt_operands): Calculate static
29122 reject and subtract it from overall when only addresses will be
29125 2017-03-06 Julia Koval <julia.koval@intel.com>
29128 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29129 incoming stack boundary to 128 for 64-bit targets.
29131 2017-03-06 Richard Biener <rguenther@suse.de>
29133 PR tree-optimization/79894
29134 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29135 to NULL after folding it.
29137 2017-03-06 Richard Biener <rguenther@suse.de>
29139 PR tree-optimization/79824
29140 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29141 check disabling peeling for gaps.
29143 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29145 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29146 attributes): Document gettimeofday.
29148 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29150 * config/s390/s390.c (s390_option_override_internal): Set
29151 PARAM_MIN_VECT_LOOP_BOUND
29153 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29155 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29156 * config/s390/s390.md: Likewise.
29158 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29161 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29162 (<avx2_avx512>_perm<mode>): Rename to ...
29163 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29165 (<avx512>_perm<mode>_mask): Rename to ...
29166 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29168 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29169 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29170 instead of VI8F_256_512.
29171 (avx512f_perm<mode>): New define_expand.
29172 (avx512f_perm<mode>_mask): Likewise.
29173 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29174 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29176 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29178 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29179 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29181 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29183 2017-03-06 Martin Liska <mliska@suse.cz>
29186 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29187 when having a SSA NAME w/o VAR_DECL assigned to it.
29189 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29191 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29192 msa_dpsub_<su>_d): Fix MODE for vec_select.
29194 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29196 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29198 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29200 2017-03-06 Richard Biener <rguenther@suse.de>
29202 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29203 * plugin.c (register_plugin_info): Likewise.
29204 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29206 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29208 * config/i386/sse.md (sse_storehps, sse_storelps,
29209 avx_<castmode><avxsizesuffix>_<castmode>,
29210 avx512f_<castmode><avxsizesuffix>_<castmode>,
29211 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29212 in condition that at least one operand is not a MEM.
29214 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29216 PR middle-end/79805
29217 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29218 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29220 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29221 gimple_call_nothrow_p flag based on whether original builtin can throw.
29222 If it can, emit following stmts on the fallthrough edge.
29223 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29224 don't create new bb if inserting just debug stmts on the edge, try to
29225 insert them on the fallthru bb or just reset debug stmts.
29227 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29230 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29231 restore recog_data (including the operand rtxes inside it) around
29232 the call to get_insn_template.
29234 2017-03-03 Martin Sebor <msebor@redhat.com>
29236 PR tree-optimization/79699
29237 * context.c (context::~context): Free MPFR caches to avoid
29238 a memory leak on program exit.
29240 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29242 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29243 Use wide_int::ulow () instead of .elt (0).
29245 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29247 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29248 (*pushxf): Limit oF constraint to 32bit targets and add oC
29249 constraint for 64bit targets.
29250 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29251 (*pushdf): Change rmF constraint to rmC.
29253 2017-03-03 Martin Liska <mliska@suse.cz>
29255 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29256 Remove unused variable.
29258 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29261 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29262 is a memory operand, increase num_memory.
29263 (ix86_expand_args_builtin): Likewise.
29265 2017-03-03 Jan Hubicka <jh@suse.cz>
29268 * ipa-devirt.c (maybe_record_node): Properly handle
29269 __cxa_pure_virtual visibility.
29271 2017-03-03 Martin Liska <mliska@suse.cz>
29273 PR tree-optimization/79803
29274 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29276 (pass_loop_prefetch::execute): Disabled optimization if an
29277 assumption about L1 cache size is not met.
29279 2017-03-03 Martin Liska <mliska@suse.cz>
29281 PR rtl-optimization/79574
29282 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29283 (hash_scan_set): Likewise.
29284 (dump_hash_table): Likewise.
29285 (hoist_code): Likewise.
29287 2017-03-03 Richard Biener <rguenther@suse.de>
29289 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29290 in place of uhigh ().
29291 (fixed_convert_from_real): Likewise.
29293 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29296 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29298 2017-03-03 Richard Biener <rguenther@suse.de>
29300 PR middle-end/79818
29301 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29302 TYPE_OVERFLOW_UNDEFINED check.
29304 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29306 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29308 (vector_ae_<mode>_p): Likewise.
29309 (vector_nez_<mode>_p): Likewise.
29310 (vector_ne_v2di_p): Likewise.
29311 (vector_ae_v2di_p): Likewise.
29312 (vector_ne_<mode>_p): Likewise.
29313 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29315 (vsx_tsqrt<mode>2_fe): Likewise.
29317 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29320 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29322 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29324 PR rtl-optimization/79780
29325 * cprop.c (one_cprop_pass): When second and further conditional trap
29326 in a single basic block is turned into an unconditional trap, turn it
29327 into a deleted note to avoid RTL verification failures.
29329 2017-03-02 Richard Biener <rguenther@suse.de>
29331 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29333 2017-03-02 Richard Biener <rguenther@suse.de>
29335 PR tree-optimization/79345
29337 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29338 param and abort the walk, returning -1 if it is hit.
29339 (walk_aliased_vdefs): Take a limit param and pass it on.
29340 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29341 defaulting to 0 and return a signed int.
29342 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29343 (check_defs): New helper.
29344 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29345 about uninitialized memory.
29346 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29347 bogus uninitialized warning.
29348 (fixed_convert_from_real): Likewise.
29350 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29352 PR tree-optimization/66768
29353 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29354 iv_use if base object can't be determined.
29356 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29358 PR tree-optimization/79345
29359 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29360 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29361 (get_pattern_stats): Initialize it.
29362 * genemit.c (gen_expand): Verify match_scratch numbers come after
29363 match_operand/match_dup numbers.
29364 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29365 match_scratch numbers.
29366 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29368 * config/s390/s390.md (trunctdsd2): Likewise.
29370 2017-03-02 Richard Biener <rguenther@suse.de>
29372 * wide-int.h (wide_int_storage::operator=): Implement in terms
29375 2017-03-02 Richard Biener <rguenther@suse.de>
29377 PR tree-optimization/79777
29378 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29379 the to insert expression to sth existing.
29381 2017-03-01 Martin Sebor <msebor@redhat.com>
29383 PR middle-end/79692
29384 * gimple-ssa-sprintf.c
29385 (directive::known_width_and_precision): New function.
29386 (format_integer): Use it.
29387 (get_mpfr_format_length): Consider the full range of precision
29388 when computing %g output with the # flag. Set the likely byte
29389 count to 3 rather than 1 when precision is indeterminate.
29390 (format_floating): Correct the lower bound of precision.
29392 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29394 * doc/invoke.texi: Document default code model for 64-bit Linux.
29396 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29399 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29400 udiv rather than div since input pattern is unsigned.
29402 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29404 * config/i386/i386.c (print_reg): Warn for values of
29405 unsupported size in integer register.
29407 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29410 * config/rs6000/predicates.md (current_file_function_operand): Do
29411 not allow self calls to be local if the function is replaceable.
29413 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29416 * config/rs6000/altivec.h (vec_ctz and others): Change the
29417 preprocessor macro that controls conditional compilation from
29418 _ARCH_PWR9 to __POWER9_VECTOR__.
29419 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29420 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29421 control (instead of _ARCH_PWR9 control) so that template
29422 definition uses power9-specific function.
29423 (vec_any_eq): Likewise.
29424 (vec_all_ne): Change macro definition to use a power9-specific
29425 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29426 _ARCH_PWR9 control).
29427 (vec_any_eq) Likewise.
29428 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29429 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29430 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29431 support for xvcmpnedp instruction.
29432 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29433 macro expansion so that Power9 implementation of vec_all_ne does
29434 not use the AltiVec predicate framework.
29435 (VCMPNEH_P): Likewise.
29436 (VCMPNEW_P): Likewise.
29437 (VCMPNED_P): Likewise.
29438 (VCMPNEFP_P): Likewise.
29439 (VCMPNEDP_P): Likewise.
29440 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29441 implementation of vec_any_eq to not use AltiVec predicate
29443 (VCMPAEH_P): Likewise.
29444 (VCMPAEW_P): Likewise.
29445 (VCMPAED_P): Likewise.
29446 (VCMPAEFP_P): Likewise.
29447 (VCMPAEDP_P): Likewise.
29448 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29449 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29450 not use the AltiVec predicate framework.
29451 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29452 of vec_any_eq to not use AltiVec predicate framework.
29453 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29454 support for predefined __POWER9_VECTOR__ macro to indicate that
29455 Power9 instruction selection is enabled.
29456 (altivec_overloaded_builtins): Remove extraneous
29457 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29458 function argument types RS6000_BTI_bool_V16QI and
29459 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29460 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29461 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29462 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29463 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29464 Power9 for implementations of vec_cmpne. Change the signature for
29465 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29466 (representing vec_all_ne) to remove the previously described first
29467 argument of type RS6000_BTI_INTSI, as this was an artifact of
29468 reliance on the AltiVec predicate framework, which is no longer
29469 used in the implementation of these functions. Add
29470 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29471 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29472 since, unlike the AltiVec predicate framework implementation, we
29473 do not share function descriptors between vec_alle and vec_anyeq.
29474 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29475 set of modes that receive special treatment even when
29476 TARGET_P9_VECTOR is true. The special treatment emits code that
29477 does not depend on Power9 instructions.
29478 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29479 define_expand to not rely on AltiVec predicate framework.
29480 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29482 (vector_ne_v2di_p): Change this define_expand to not rely on
29483 AltiVec predicate framework.
29484 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29486 (vector_ne_<mode>_p): Change this define_expand to not rely on
29487 AltiVec predicate framework.
29488 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29490 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29491 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29492 define_insn pattern.
29493 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29494 define_insn pattern because the xvcmpne<VSs>. instruction is not
29496 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29497 instruction is not supported.
29499 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29501 * config/nvptx/nvptx.c: Include intl.h.
29503 2017-03-01 Martin Jambor <mjambor@suse.cz>
29506 * ipa-prop.h (ipa_bits): Removed field known.
29507 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29508 to pointers. Adjusted their comments to warn about their sharing.
29509 (ipcp_transformation_summary): Change bits to a vector of pointers.
29510 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29511 (ipa_get_ipa_bits_for_value): Declare.
29512 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29513 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29514 (ipa_bits_hash_table): Likewise.
29515 (ipa_vr_ggc_hash_traits): Likewise.
29516 (ipa_vr_hash_table): Likewise.
29517 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29518 being pointers and vr_known being removed.
29519 (ipa_set_jf_unknown): Likewise.
29520 (ipa_get_ipa_bits_for_value): New function.
29521 (ipa_set_jfunc_bits): Likewise.
29522 (ipa_get_value_range): New overloaded functions.
29523 (ipa_set_jfunc_vr): Likewise.
29524 (ipa_compute_jump_functions_for_edge): Use the above functions to
29525 construct bits and vr parts of jump functions.
29526 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29527 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29529 (ipcp_grow_transformations_if_necessary): Also allocate
29530 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29532 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29533 them. Fix too long lines.
29534 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29535 vr_known being removed.
29536 (ipa_read_jump_function): Use new setter functions to construct bits
29537 and vr parts of jump functions or set them to NULL.
29538 (write_ipcp_transformation_info): Adjust for bits being pointers.
29539 (read_ipcp_transformation_info): Likewise.
29540 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29542 Include gt-ipa-prop.h.
29543 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29545 (ipcp_store_bits_results): Likewise.
29546 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29547 Do not write to existing jump functions but use a temporary instead.
29549 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29552 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29553 attempt to use its first operand as BIT_FIELD_REF base.
29555 2017-03-01 Richard Biener <rguenther@suse.de>
29557 PR middle-end/79721
29558 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29559 interpolating formula in wrapping arithmetic.
29560 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29562 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29564 PR tree-optimization/79734
29565 * tree-vect-generic.c (expand_vector_condition): Optimize
29566 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29567 Handle VEC_COND_EXPR where comparison has different inner width from
29568 type's inner width.
29570 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29572 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29573 markup, and similar issues. Remove @opindex entries for things
29574 that aren't options. Add missing -mmpy-option entries.
29576 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29578 PR tree-optimization/79737
29579 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29580 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29581 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29582 instead of byte_size. Formatting fix.
29583 (shift_bytes_in_array_right): Formatting fix.
29585 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29588 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29589 condition on optimize for the leaf function test.
29591 2017-02-28 Martin Liska <mliska@suse.cz>
29594 * read-rtl-function.c (function_reader::handle_unknown_directive):
29595 Bail out when one uses -flto.
29597 2017-02-28 Martin Liska <mliska@suse.cz>
29599 * common.opt: Replace space with tabular for options of <number>
29601 * config/i386/i386.opt: Show <number> value for
29602 -mlarge-data-threshold.
29603 * opts.c (print_filtered_help): Do not display number in hexadecimal
29606 2017-02-28 Martin Liska <mliska@suse.cz>
29608 * common.opt: Fix --help=option -Q for options which are of
29611 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29613 * config/i386/i386.c (print_reg): Error out for values
29614 of 8-bit size in invalid integer register.
29616 2017-02-28 Martin Sebor <msebor@redhat.com>
29618 PR tree-optimization/79691
29619 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29621 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29624 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29625 gcc_unreachable with output_operand_lossage.
29627 2017-02-28 Richard Biener <rguenther@suse.de>
29629 PR tree-optimization/79740
29630 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29632 (visit_nary_op): Insert the nary into the hashtable if we
29633 pattern-matched sth.
29634 * tree-ssa-pre.c (eliminate_insert): Robustify.
29636 2017-02-28 Richard Biener <rguenther@suse.de>
29638 PR middle-end/79731
29639 * fold-const.c (decode_field_reference): Reject out-of-bound
29642 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29644 * config/i386/i386.c: Include intl.h.
29645 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29646 instead of just cond ? "..." : "...".
29647 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29648 * coverage.c (read_counts_file): Likewise.
29649 * omp-offload.c: Include intl.h.
29650 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29651 of just cond ? "..." : "...".
29652 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29653 of just cond ? "..." : "...".
29655 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29658 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29660 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29661 'tune for' CPU name.
29662 * config/arm/arm-cpu-data.h: Regenerated.
29664 2017-02-28 Richard Biener <rguenther@suse.de>
29666 PR tree-optimization/79732
29667 * tree-inline.c (expand_call_inline): Do not shadow var.
29669 2017-02-28 Richard Biener <rguenther@suse.de>
29671 PR tree-optimization/79723
29672 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29673 address-space properly.
29675 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29677 * doc/optinfo.texi (Optimization groups): Fix option used for
29679 * doc/invoke.texi (-fopt-info): Document "omp".
29680 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29681 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29682 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29683 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29684 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29686 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29688 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29689 OPTGROUP_OMP with "omp".
29691 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29694 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29695 for arithmetic shift of unsigned V2DI.
29697 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29699 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29700 arc/linux.h headers.
29701 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29702 (LINK_SPEC): Likewise.
29703 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29704 (EXTRA_SPECS): Likewise.
29705 (STARTFILE_SPEC): Likewise.
29706 (ENDFILE_SPEC): Likewise.
29707 (LIB_SPEC): Likewise.
29708 (TARGET_SDATA_DEFAULT): Likewise.
29709 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29710 (MULTILIB_DEFAULTS): Likewise.
29711 (DWARF2_UNWIND_INFO): Likewise.
29712 * config/arc/big.h: New file.
29713 * config/arc/elf.h: Likewise.
29714 * config/arc/linux.h: Likewise.
29715 * config/arc/t-uClibc: Remove.
29717 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29719 PR tree-optimization/77536
29720 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29721 (tree_transform_and_unroll_loop): Use above function to compute the
29722 estimated niter of unrolled loop and use it when scaling profile.
29723 Also use count info rather than frequency if it's non-zero.
29724 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29725 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29726 (vect_transform_loop): Call above function.
29728 2017-02-27 Richard Biener <rguenther@suse.de>
29730 PR tree-optimization/45397
29731 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29732 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29733 (visit_nary_op): Add pattern matching for CSEing sign-changed
29734 or truncated operations with wider ones.
29736 2017-02-27 Richard Biener <rguenther@suse.de>
29738 PR tree-optimization/79690
29739 * tree-vect-stmts.c (vectorizable_store): Use vector type
29740 built from the DR with address-space.
29742 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29744 * doc/invoke.texi (Optimize Options): Refine the description
29745 of asan-use-after-return.
29747 2017-02-25 Alan Modra <amodra@gmail.com>
29749 PR rtl-optimization/79584
29750 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29751 base, not ad->base_term, the reg within base. Remove assertion
29752 that ad->base == ad->base_term. Replace gen_int_mode using
29753 bogus mode with const0_rtx.
29755 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29757 PR middle-end/79396
29758 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29759 FMA_EXPR like tcc_binary or tcc_unary.
29761 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29764 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29766 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29767 (output_loc_operands): Handle DW_OP_call_ref and
29768 DW_OP_GNU_variable_value.
29769 (struct variable_value_struct): New type.
29770 (struct variable_value_hasher): Likewise.
29771 (variable_value_hash): New variable.
29772 (string_types): Remove.
29773 (copy_loc_descr): New function.
29774 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29775 (prepend_loc_descr_to_each): New function.
29776 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29777 instead of add_loc_descr_to_each if the first argument is single
29778 location list and the second has multiple.
29779 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29780 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29781 when looking for variable value which doesn't have other location info.
29782 (loc_list_from_tree): Formatting fix.
29783 (gen_array_type_die): Simplify DW_AT_string_length handling.
29784 (adjust_string_types): Remove.
29785 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29786 string_types. Call resolve_variable_values.
29787 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29788 (resolve_addr_in_expr): Likewise. Add A argument.
29789 (copy_deref_exprloc): Remove deref argument. Adjust for the
29790 original expression being DW_OP_GNU_variable_value with optionally
29791 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29792 optionally after it.
29793 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29794 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29795 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29796 (variable_value_hasher::hash, variable_value_hasher::equal): New
29798 (resolve_variable_value_in_expr, resolve_variable_value,
29799 resolve_variable_values, note_variable_value_in_expr,
29800 note_variable_value): New functions.
29801 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29804 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29807 * opts.h (handle_generated_option): Add GENERATED_P argument.
29808 * opts-common.c (handle_option): Adjust function comment.
29809 (handle_generated_option): Add GENERATED_P argument, pass it to
29811 (control_warning_option): Pass false to handle_generated_option
29813 * opts.c (maybe_default_option): Pass true to handle_generated_option
29815 * optc-gen.awk: Likewise.
29817 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29819 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29820 a REG, look at the REG it is a SUBREG of.
29821 (splitter for cmpeqsi_t): Ditto.
29823 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29825 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29826 the special USEs with the pattern of the insn, not the insn itself.
29828 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29831 * doc/invoke.texi: Document -mload-store-pairs.
29833 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29834 Sandra Loosemore <sandra@codesourcery.com>
29836 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29837 argument isn't a CONST_INT.
29838 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29839 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29840 (nios2_validate_compare): Bypass alternate compare logic if *op2
29841 is not a CONST_INT.
29842 (ldstwm_operation_p): Return false if first_base is not a REG or
29843 if first_offset is not a CONST_INT.
29845 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29847 * config/cris/cris.md: Use correct operand in a define_peephole2.
29849 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29851 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29853 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29855 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29856 this_insn if it is an INSN or JUMP_INSN.
29857 (force_offsettable): Look at base, not at addr.
29858 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29859 on things that aren't necessarily CONST_INTs.
29861 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29863 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29864 -mfpmath=sse is the default also for x86-32 targets with SSE2
29865 instruction set when @option{-ffast-math} is enabled
29867 2017-02-24 Jeff Law <law@redhat.com>
29869 PR rtl-optimizatoin/79286
29870 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29873 2017-02-24 Richard Biener <rguenther@suse.de>
29875 PR tree-optimization/79389
29876 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29879 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29881 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29882 function comment to reflect reality.
29883 (loop_exits_before_overflow): Fix typo in function description.
29885 2017-02-24 Richard Biener <rguenther@suse.de>
29887 PR tree-optimization/79389
29888 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29889 properly that a threading opportunity exists. Detect conditional
29890 copy/constant propagation opportunities.
29892 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29894 * config/visium/visium.md (type): Add trap.
29895 (b): New mode attribute.
29896 (*btst): Rename into...
29897 (*btst<mode>): ...this and adjust.
29898 (*cbranchsi4_btst_insn): Rename into...
29899 (*cbranch<mode>4_btst_insn): ...this and adjust.
29900 (trap): New define_insn.
29902 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29904 PR tree-optimization/79389
29905 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29906 (noce_reversed_cond_code): New function.
29907 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29908 reversed_comparison_code. Formatting fix.
29909 (noce_try_store_flag): Test rev_cond != NULL in addition to
29910 reversed_comparison_code.
29911 (noce_try_store_flag_constants): Likewise.
29912 (noce_try_store_flag_mask): Likewise.
29913 (noce_try_addcc): Use rev_cond if non-NULL instead of
29914 reversed_comparison_code.
29915 (noce_try_cmove_arith): Likewise. Formatting fixes.
29916 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29917 (noce_find_if_block): Initialize rev_cond.
29918 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29919 instead of false as last argument never attempt to reverse it
29922 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29924 PR tree-optimization/79663
29925 * tree-predcom.c (combine_chains): Process refs in reverse order
29926 only for ZERO length chains, and add explaining comment.
29928 2017-02-23 Jeff Law <law@redhat.com>
29930 PR tree-optimization/79578
29931 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29932 in call to operand_equal_p.
29934 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29937 * config/i386/cpuid.h: Fix another undefined behavior.
29939 2017-02-23 Richard Biener <rguenther@suse.de>
29941 PR tree-optimization/79683
29942 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29943 vector types for data-refs.
29945 2017-02-23 Martin Liska <mliska@suse.cz>
29947 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29949 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29951 PR middle-end/79665
29952 * internal-fn.c (get_range_pos_neg): Moved to ...
29953 * tree.c (get_range_pos_neg): ... here. No longer static.
29954 * tree.h (get_range_pos_neg): New prototype.
29955 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29956 are known to be in between 0 and signed maximum inclusive, try to
29957 expand both unsigned and signed divmod and use the cheaper one from
29960 2017-02-22 Jeff Law <law@redhat.com>
29962 PR tree-optimization/79578
29963 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29964 to compare base operands.
29966 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29969 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29970 gpc_reg_operand instead of fpr_reg_operand.
29972 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29974 * config/mips/mips.c (mips_return_in_memory): Force FP
29975 vector types to be returned in memory for o32 ABI.
29977 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29979 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29980 instead of DW_TAG_member for static data member declarations and don't
29981 set no_linkage_name for static inline data members.
29982 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29985 2017-02-22 Martin Liska <mliska@suse.cz>
29987 * doc/invoke.texi: Replace inequality signs with square brackets
29990 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29993 * lra-constraints.c (simplify_operand_subreg): Handle
29994 WORD_REGISTER_OPERATIONS targets.
29996 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29999 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30000 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30001 elimination by swapping fld*.
30003 2017-02-22 Richard Biener <rguenther@suse.de>
30005 PR tree-optimization/79673
30006 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30007 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30008 irrelevant address-space qualifiers and avoiding a
30009 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30011 2017-02-22 Richard Biener <rguenther@suse.de>
30013 PR tree-optimization/79666
30014 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30015 to not symbolically negate if that may introduce undefined
30018 2017-02-22 Martin Liska <mliska@suse.cz>
30021 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30022 * data-streamer-out.c (streamer_write_gcov_count_stream):
30024 * value-prof.c (stream_out_histogram_value): Make assert more
30025 precise based on type of counter.
30027 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30030 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30031 Use nonimmediate_operand instead of memory_operand for operand 1.
30032 (float-extend standard_x87sse_constant_load splitter): Ditto.
30034 2017-02-21 Jeff Law <law@redhat.com>
30036 PR tree-optimization/79621
30037 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30038 blocks with edges to themselves.
30040 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30043 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30044 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30045 Use gimple_call_builtin_p.
30048 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30049 on temporarily removed DEBUG_INSNs.
30051 PR tree-optimization/79649
30052 * tree-loop-distribution.c (classify_partition): Give up on
30053 non-generic address space loads/stores.
30055 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30057 * doc/loop.texi (Loop manipulation): Remove nonexistent
30058 tree_ssa_loop_version from the documentation.
30059 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30061 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30064 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30065 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30066 * config/rs6000/rs6000.c: Include except.h.
30067 (rs6000_expand_split_stack_prologue): Call
30068 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30070 2017-02-21 Martin Jambor <mjambor@suse.cz>
30073 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30074 have been analyzed.
30076 2017-02-21 Martin Jambor <mjambor@suse.cz>
30078 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30079 for backward compatibility only.
30080 * doc/invoke.texi (Option Summary): Remove all references to
30081 -fipa-cp-alignment.
30083 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30087 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30089 * lra-constraints.c (curr_insn_transform): Handle
30090 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30092 2017-02-21 Martin Liska <mliska@suse.cz>
30094 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30096 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30098 PR translation/79638
30099 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30101 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30104 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30105 (arm_function_ok_for_sibcall): Return false for an indirect call by
30106 descriptor if all the argument registers are used.
30107 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30108 alignment of the function.
30110 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30112 PR tree-optimization/61441
30113 * simplify-rtx.c (simplify_const_unary_operation): For
30114 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30115 the sNaN unmodified.
30117 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30119 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30120 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30121 instead of SYSTEM_HEADER_DIR.
30123 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30124 Martin Liška <mliska@suse.cz>
30126 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30127 Fix typos and grammar, use active voice, and clarify.
30129 2017-02-20 Marek Polacek <polacek@redhat.com>
30131 PR middle-end/79537
30132 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30135 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30137 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30140 * config/i386/i386.c (ix86_expand_builtin): Handle
30141 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30142 ix86_builtins_isa[fcode].isa as a requirement of those
30143 flags and any other flag in the bitmask.
30144 (ix86_init_mmx_sse_builtins): Use 0 instead of
30145 ~OPTION_MASK_ISA_64BIT as mask.
30146 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30147 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30148 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30149 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30151 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30154 * lra-constraints.c (split_reg): Check requested split mode
30155 is supported by the register.
30157 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30159 * lra-constraints.c (simplify_operand_subreg): Remove early
30162 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30165 * lra-constraints.c (curr_insn_transform): Tighten condition
30166 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30168 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30171 * lra-constraints.c (curr_insn_transform): Handle
30172 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30174 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30177 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30179 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30181 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30184 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30187 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30189 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30192 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30194 * final.c (last_columnnum, override_columnnum): New variables.
30195 (final_start_function): Set last_columnnum, pass it to begin_prologue
30196 hook and pass 0 to dwarf2out_begin_prologue.
30197 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30198 to source_line debug hook.
30199 (notice_source_line): Compute last_columnnum and for debug_column_info
30200 return true on column changes.
30201 * debug.h (struct gcc_debug_hooks): Add column argument to
30202 source_line and begin_prologue hooks.
30203 (debug_nothing_int_charstar_int_bool): Remove prototype.
30204 (debug_nothing_int_int_charstar,
30205 debug_nothing_int_int_charstar_int_bool): New prototypes.
30206 (dwarf2out_begin_prologue): Add column argument.
30207 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30208 begin_prologue hooks.
30209 (debug_nothing_int_charstar_int_bool): Remove.
30210 (debug_nothing_int_int_charstar,
30211 debug_nothing_int_int_charstar_int_bool): New functions.
30212 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30213 through to dwarf2out_source_line.
30214 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30215 (dwarf2out_source_line): Add column argument, emit it if requested.
30216 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30218 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30219 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30220 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30221 through to dwarf2out_begin_prologue.
30222 (vmsdbgout_source_line): Add column argument, pass it through to
30223 dwarf2out_source_line.
30224 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30225 dbxout_source_line caller.
30226 (dbxout_source_line): Add column argument.
30228 * common.opt (gno-column-info, gcolumn-info): New options.
30229 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30230 (check_die): Also test for multiple DW_AT_decl_column attributes.
30231 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30232 DW_AT_decl_column if requested.
30233 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30235 (gen_variable_die): Likewise.
30236 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30237 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30240 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30241 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30242 (ix86_handle_option): Handle OPT_m3dnowa.
30243 * doc/invoke.texi (-m3dnowa): Document.
30244 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30245 -m3dnowa instead of -m3dnow -march=athlon.
30248 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30249 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30251 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30254 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30255 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30256 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30257 generator for vsx_xxpermdi_<mode>_be.
30258 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30259 force big-endian semantics.
30260 (vsx_xxpermdi_<mode>_be): New define_expand with same
30261 implementation as previous version of vsx_xxpermdi_<mode>.
30263 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30265 PR tree-optimization/79327
30266 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30267 variable, its initialization and use.
30269 2017-02-17 Julia Koval <julia.koval@intel.com>
30271 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30272 (OPTION_MASK_ISA_PKU_UNSET): New.
30273 (ix86_handle_option): Handle -mrdpid.
30274 * config/i386/cpuid.h (bit_RDPID): New.
30275 * config/i386/driver-i386.c (host_detect_local_cpu):
30276 Detect RDPID feature.
30277 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30278 * config/i386/i386-c.c (ix86_target_macros_internal):
30280 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30281 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30282 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30283 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30284 * config/i386/i386.md (define_insn "rdpid"): New.
30285 * config/i386/i386.opt Add -mrdpid.
30286 * config/i386/immintrin.h (_rdpid_u32): New.
30288 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30290 PR rtl-optimization/79541
30291 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30292 instead of transforming it into USE.
30294 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30296 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30297 If HONOR_SNANS (SFmode) force the input to a register.
30298 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30299 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30300 an frsp or similar insn.
30302 2017-02-17 Martin Liska <mliska@suse.cz>
30304 PR rtl-optimization/79577
30305 * params.def (selsched-max-sched-times): Increase minimum to 1.
30307 2017-02-17 Martin Liska <mliska@suse.cz>
30309 PR rtl-optimization/79574
30310 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30312 2017-02-17 Martin Liska <mliska@suse.cz>
30314 PR tree-optimization/79529
30315 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30316 ssa_defined_default_def_p to handle cases which are implicitly
30318 * tree-ssa.c (ssa_defined_default_def_p): New function.
30319 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30320 which are implicitly defined.
30321 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30323 2017-02-17 Richard Biener <rguenther@suse.de>
30325 PR middle-end/79576
30326 * params.def (max-ssa-name-query-depth): Limit to 10.
30328 2017-02-17 Richard Biener <rguenther@suse.de>
30330 PR tree-optimization/79552
30331 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30334 2017-02-17 Richard Biener <rguenther@suse.de>
30337 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30339 2017-02-17 Marek Polacek <polacek@redhat.com>
30341 PR middle-end/79536
30342 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30343 (fold_negate_expr): New wrapper.
30345 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30347 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30348 Correct terminology and de-emphasize pre-standard behavior.
30350 2017-02-16 Alan Modra <amodra@gmail.com>
30352 PR rtl-optimization/79286
30353 * ira.c (def_dominates_uses): New function.
30354 (update_equiv_regs): Don't create an equivalence for insns that
30355 may trap where the register def does not dominate the use.
30357 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30359 PR rtl-optimization/78127
30360 * lra.c (lra): Call lra_eliminate before finish the loop after
30363 2017-02-16 Richard Biener <rguenther@suse.de>
30365 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30367 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30368 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30369 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30370 (isl_val_int_from_wi): New function.
30371 (extract_affine_gmp): Rename to ...
30372 (extract_affine_wi): ... this, take a widest_int.
30373 (extract_affine_int): Just wrap extract_affine_wi.
30374 (add_param_constraints): Use isl_val_int_from_wi.
30375 (add_loop_constraints): Likewise, and extract_affine_wi.
30377 2017-02-15 Jeff Law <law@redhat.com>
30379 PR middle-end/79521
30380 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30381 ira_init_register_move_cost_if_necessary.
30383 2017-02-15 Martin Sebor <msebor@redhat.com>
30385 PR middle-end/32003
30386 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30387 removed in a prior commit.
30389 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30391 PR tree-optimization/79347
30392 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30393 counters during peeling.
30395 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30397 * Makefile.in (site.exp): Remove "set ISLVER".
30399 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30402 * real.c (real_from_integer): Call real_convert even for decimal.
30404 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30407 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30409 2017-02-14 Andrew Pinski <apinski@cavium.com>
30411 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30412 cores and change the partno/implementer to be correct.
30413 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30414 the 'B" as the implementer.
30415 * config/aarch64/aarch64-tune.md: Regenerate.
30417 2017-02-14 Carl Love <cel@us.ibm.com>
30419 * config/rs6000/rs6000.c: Add case statement entry to make the
30420 xvcvuxdsp built-in argument unsigned.
30421 * config/rs6000/vsx.md: Fix the source and return operand types so they
30422 match the instruction definitions from the ISA document. Fix typo
30423 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30426 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30429 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30430 member early_clobber_alts.
30431 * lra-lives.c (reg_early_clobber_p): New.
30432 (process_bb_lives): Use it.
30433 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30434 (debug_operand_data): Initialize early_clobber_alts.
30435 (setup_operand_alternative): Set up early_clobber_alts.
30436 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30437 alternatives to new_insn_reg.
30438 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30440 (lra_update_insn_regno_info): Pass the new arg.
30442 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30444 PR middle-end/79505
30445 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30446 (new_oacc_loop_raw): Don't clear already cleared fields.
30449 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30450 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30451 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30453 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30456 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30458 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30461 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30462 the extra instruction to the right place to store 128-bit constant
30465 2017-02-14 Martin Sebor <msebor@redhat.com>
30467 PR middle-end/79448
30468 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30469 warning for strings of unknown length.
30471 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30473 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30475 2017-02-14 Jeff Law <law@redhat.com>
30478 * ira-costs.c (scan_one_insn): Initialize register move costs
30479 for pseudos seen in USE/CLOBBER insns.
30481 PR tree-optimization/79095
30482 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30483 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30484 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30485 if the operands are known to be not equal, then the resulting range
30487 (intersect_ranges): If the new range is ~[0,0] and the old range is
30488 wide, then prefer ~[0,0].
30489 * tree-vrp.c (overflow_comparison_p_1): New function.
30490 (overflow_comparison_p): New function.
30491 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30492 if NAME is used in an overflow test.
30493 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30494 overflow check that can be expressed as an equality test, then adjust
30495 ops to be that equality test.
30497 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30499 * config/s390/s390-builtin-types.def: Remove flags argument.
30500 * config/s390/s390.c (s390_init_builtins): Likewise.
30502 2017-02-14 Martin Liska <mliska@suse.cz>
30504 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30505 vector. Fix trailing white spaces.
30507 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30509 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30512 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30514 PR rtl-optimization/68664
30515 * config/arm/arm.c (arm_sched_can_speculate_insn):
30516 New function. Declare prototype.
30517 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30519 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30521 PR rtl-optimization/68664
30522 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30524 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30526 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30528 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30529 max skip bytes for function, loop and jump.
30531 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30533 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30534 ABS_EXPR for gimple dump.
30536 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30539 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30541 PR tree-optimization/79408
30542 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30543 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30544 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30545 also if rhs1 is INTEGER_CST.
30547 2017-02-14 Richard Biener <rguenther@suse.de>
30549 PR middle-end/79432
30550 * tree-into-ssa.c (insert_phi_nodes): When the function can
30551 have abnormal edges rewrite SSA names with broken use-def
30552 dominance out of SSA and register them for PHI insertion.
30554 2017-02-13 Martin Sebor <msebor@redhat.com>
30556 PR middle-end/79496
30557 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30558 clearing info.nowrite flag when snprintf size argument is a range.
30560 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30562 * cprop.c (cprop_jump): Add missing space in string literal.
30563 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30564 (get_constraint_for_component_ref): Likewise.
30565 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30566 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30567 * lra-constraints.c (process_alt_operands): Likewise.
30568 * ipa-inline.c (inline_small_functions): Likewise.
30569 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30570 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30571 * trans-mem.c (diagnose_tm_1_op): Likewise.
30572 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30573 (grid_parallel_clauses_gridifiable): Likewise.
30575 * config/nvptx/mkoffload.c (process): Add space in between
30578 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30579 "MOD4_SSE_REGS" and "ALL_REGS".
30581 * spellcheck.c (test_data): Add , in between "foo" and "food".
30583 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30586 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30587 boundary crossing check and subsequent code generation agree.
30589 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30591 * config/aarch64/aarch64.c (has_memory_op): Delete.
30592 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30595 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30597 PR rtl-optimization/79388
30598 PR rtl-optimization/79450
30599 * combine.c (distribute_notes): When removing TEM_INSN for which
30600 corresponding dest has last value recorded, invalidate that last
30603 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30605 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30606 of explicit '@'. Add missing assembly comment marker on branch costs
30609 2017-02-13 Nathan Sidwell <nathan@acm.org>
30611 * gengtype-lex.l (<in_struct>): Add '/'.
30613 2017-02-13 Martin Liska <mliska@suse.cz>
30616 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30618 2017-02-13 Richard Biener <rguenther@suse.de>
30620 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30622 * configure: Re-generate.
30623 * config.in: Likewise.
30624 * graphite-dependences.c: Simplify as if
30625 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30626 * graphite-isl-ast-to-gimple.c: Likewise.
30627 * graphite-optimize-isl.c: Likewise.
30628 * graphite-poly.c: Likewise.
30629 * graphite-sese-to-poly.c: Likewise.
30630 * graphite.h: Likewise.
30631 * toplev.c: Include isl/version.h and use isl_version () for
30632 printing the ISL version.
30633 * doc/install.texi: Update ISL requirement.
30635 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30637 * doc/standards.texi (Standards): Update reference to
30640 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30642 * doc/extend.texi (Named Address Spaces): sourceware.org now
30644 * doc/install.texi (Binaries): Ditto.
30647 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30649 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30650 standard terminology "stringize"/"stringizing" throughout.
30651 * doc/cppinternals.texi: Likewise.
30653 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30655 * doc/extend.texi: Fix some spelling mistakes and typos.
30656 * doc/invoke.texi: Likewise.
30658 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30661 * params.def (inline-min-speedup) Change from 10 to 8.
30663 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30665 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30668 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30671 * ipa-inline-analysis.c (get_minimal_bb): New function.
30672 (record_modified): Use it.
30673 (remap_edge_change_prob): Handle also ancestor functions.
30675 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30677 * doc/contrib.texi (Contributors): Remove broken link into
30678 the Mauve CVS repository.
30680 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30682 PR middle-end/79454
30683 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30684 result computation whenever lhs doesn't have vector mode, not
30685 just when it has BLKmode.
30687 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30689 * doc/makefile.texi (profiledbootstrap): Refer to the
30690 installation instructions only in textual form.
30692 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30695 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30697 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30699 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30700 (Specific): Update mingw-w64 reference.
30702 (Specific): Remove broken link to Renesas RX processor.
30704 2017-02-10 Richard Biener <rguenther@suse.de>
30706 * toplev.c (process_options): Do not mention obsolete graphite
30707 options when printing sorry message about missing graphite support.
30708 Mention -floop-nest-optimize.
30710 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30712 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30713 (vtst_p16): Likewise.
30714 (vtstq_p8): Likewise.
30715 (vtstq_p16): Likewise.
30717 (vtstq_p64): Likewise.
30718 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30719 (vset_lane_p64): New.
30720 (vsetq_lane_p64): New.
30722 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30724 PR tree-optimization/79411
30725 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30726 stmt operands are SSA_NAMEs used in abnormal phis.
30727 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30730 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30733 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30736 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30738 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30740 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30742 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30746 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30747 not arbitrary TREE_CONSTANT.
30750 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30751 "omp declare target link" attribute unless is_global_var.
30752 * omp-offload.c (find_link_var_op): Likewise.
30754 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30755 Chung-Lin Tang <cltang@codesourcery.com>
30757 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30759 (gimplify_adjust_omp_clauses): Don't delete TILE.
30760 (gimplify_omp_for): Deal with TILE.
30761 * internal-fn.c (expand_GOACC_TILE): New function.
30762 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30764 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30765 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30767 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30768 avoid DIV for outermost collapse var.
30769 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30770 Remove out of date comments, fix whitespace.
30771 * omp-general.c (omp_extract_for_data): Deal with tiling.
30772 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30773 adjust OLF_DIM_BASE value.
30774 (struct omp_for_data): Add tiling field.
30775 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30776 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30777 for auto loops. Remove default auto determining, moved to
30778 oacc_loop_fixed_partitions.
30779 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30780 stmts, add e_mask field.
30781 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30782 (oacc_thread_numbers): Use oacc_dim_call.
30783 (oacc_xform_tile): New.
30784 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30785 (finish_oacc_loop): Adjust for ifns vector.
30786 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30787 add case for GOACC_TILE fns.
30788 (oacc_loop_xform_loop): Delete.
30789 (oacc_loop_process): Iterate over call list directly, and add
30790 handling for GOACC_TILE fns.
30791 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30793 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30794 vector partitioning to outer loops. Assign 2 partitions to loops
30795 when available. Add TILE handling.
30796 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30797 (execite_oacc_device_lower): Process GOACC_TILE fns,
30798 ignore unknown specs.
30799 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30800 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30801 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30803 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30805 * configure.ac (ACX_BUGURL): Update.
30806 * configure: Regenerate.
30808 2017-02-09 Richard Biener <rguenther@suse.de>
30810 PR tree-optimization/69823
30811 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30812 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30814 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30816 * config/arc/arc-c.def: Add __NPS400__ definition.
30817 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30818 (TARGET_NPS400): Define.
30820 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30822 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30824 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30825 pointer, arch_info.
30826 (arc_cpu_types): Fill the arch_info field with a pointer into the
30827 arc_arch_types table.
30828 (arc_selected_cpu): Declare.
30829 * config/arc/arc.c (arc_selected_cpu): Make global.
30830 (arc_selected_arch): Delete.
30831 (arc_base_cpu): Delete.
30832 (arc_override_options): Remove references to deleted variables,
30833 update access to arch information.
30834 (ARC_OPT): Update access to arch information.
30835 (ARC_OPTX): Likewise.
30836 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30837 (TARGET_ARC600): Update access to arch information.
30838 (TARGET_ARC601): Likewise.
30839 (TARGET_ARC700): Likewise.
30840 (TARGET_EM): Likewise.
30841 (TARGET_HS): Likewise.
30842 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30845 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30848 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30849 condition/operands for integer GE/LE/GEU/LEU operations.
30851 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30853 PR translation/79397
30854 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30857 2017-02-08 Martin Jambor <mjambor@suse.cz>
30860 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30861 whether allocation happened.
30862 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30863 nothing was allocated.
30865 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30867 PR tree-optimization/79408
30868 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30869 constant, but SSA_NAME with a known integer range, use the minimum
30870 of that range instead of op1 to determine if modulo can be replaced
30871 with its first operand.
30873 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30875 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30877 2017-02-08 Richard Biener <rguenther@suse.de>
30879 PR tree-optimization/71824
30880 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30881 Check all loops contained in the merged region.
30883 2017-02-07 Andrew Pinski <apinski@cavium.com>
30885 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30887 2017-02-07 Andrew Pinski <apinski@cavium.com>
30889 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30890 (thunderxt88): Likewise.
30891 (thunderxt81): Disable LSE and change v8.1 to v8.
30892 (thunderxt83): Likewise.
30894 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30895 Richard Biener <rguenther@suse.de>
30897 PR middle-end/79399
30898 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30899 type from int to size_t.
30900 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30902 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30904 PR rtl-optimization/79386
30905 * cprop.c (bypass_conditional_jumps): Initialize
30906 bypass_last_basic_block already before splitting bbs after
30907 unconditional traps...
30908 (bypass_conditional_jumps): ... rather than here.
30911 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30912 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30913 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30914 fix -masm=intel patterns.
30916 2017-02-07 Richard Biener <rguenther@suse.de>
30918 PR tree-optimization/79256
30919 PR middle-end/79278
30920 * builtins.c (get_object_alignment_2): Use min_align_of_type
30921 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30922 and ADJUST_FIELD_ALIGN.
30924 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30926 * doc/tm.texi: Regenerate.
30927 * stor-layout.c (layout_decl): Adjust.
30928 (update_alignment_for_field): Likewise.
30929 (place_field): Likewise.
30930 (min_align_of_type): Likewise.
30931 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30932 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30933 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30934 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30935 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30936 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30937 * config/i386/i386.c (x86_field_alignment): Likewise.
30938 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30939 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30940 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30941 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30942 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30943 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30947 2017-01-30 Richard Biener <rguenther@suse.de>
30949 PR tree-optimization/79256
30950 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30951 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30954 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30956 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30957 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30958 builtins to SImode and emit a zero-extend, if necessary.
30960 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30962 * docs/invoke.texi (RISC-V Options): Alphabetize.
30964 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30966 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30969 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30971 * config/riscv/riscv.c: New file.
30972 * common/config/riscv/riscv-common.c: Likewise.
30973 * config.gcc: Likewise.
30974 * config/riscv/constraints.md: Likewise.
30975 * config/riscv/elf.h: Likewise.
30976 * config/riscv/generic.md: Likewise.
30977 * config/riscv/linux.h: Likewise.
30978 * config/riscv/multilib-generator: Likewise.
30979 * config/riscv/peephole.md: Likewise.
30980 * config/riscv/pic.md: Likewise.
30981 * config/riscv/predicates.md: Likewise.
30982 * config/riscv/riscv-builtins.c: Likewise.
30983 * config/riscv/riscv-c.c: Likewise.
30984 * config/riscv/riscv-ftypes.def: Likewise.
30985 * config/riscv/riscv-modes.def: Likewise.
30986 * config/riscv/riscv-opts.h: Likewise.
30987 * config/riscv/riscv-protos.h: Likewise.
30988 * config/riscv/riscv.h: Likewise.
30989 * config/riscv/riscv.md: Likewise.
30990 * config/riscv/riscv.opt: Likewise.
30991 * config/riscv/sync.md: Likewise.
30992 * config/riscv/t-elf-multilib: Likewise.
30993 * config/riscv/t-linux: Likewise.
30994 * config/riscv/t-linux-multilib: Likewise.
30995 * config/riscv/t-riscv: Likewise.
30996 * configure.ac: Likewise.
30997 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30998 Waterman as RISC-V maintainers.
30999 * doc/install.texi: Add RISC-V entries.
31000 * doc/invoke.texi: Add RISC-V options section.
31001 * doc/md.texi: Add RISC-V constraints section.
31002 * configure: Regenerated.
31004 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31007 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31008 false values to be constant vectors with all 0 or all 1 bits set.
31009 (vcondu<mode><mode>): Likewise.
31010 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31012 (fpmask_comparison_operator): Update comment.
31013 (vecint_comparison_operator): New predicate.
31014 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31015 vector conditionals when the true and false values are constant
31016 vectors with all 0 bits or all 1 bits set.
31018 2017-02-06 Martin Sebor <msebor@redhat.com>
31020 PR tree-optimization/79376
31021 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31023 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31025 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31026 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31027 to simplify split condition.
31029 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31031 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31032 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31035 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31037 PR rtl-optimization/68664
31038 * target.def (can_speculate_insn): New hook.
31039 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31040 * doc/tm.texi: Regenerate.
31041 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31042 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31043 (rs6000_sched_can_speculate_insn): New function.
31045 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31047 PR tree-optimization/79284
31048 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31049 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31050 vectorizable_mask_load_store, vectorizable_operation,
31051 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31052 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31053 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31054 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31055 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31056 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31057 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31058 is_gimple_assign (stmt). Replace another such test with
31059 is_gimple_assign (stmt).
31061 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31064 * config/avr/avr.c (rtl-iter.h): Include it.
31065 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31066 (avr_legitimate_combined_insn): ...and implementation.
31068 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31070 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31071 * config/s390/s390.c (s390_const_operand_ok)
31072 (s390_canonicalize_comparison, s390_extract_part)
31073 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31074 (s390_contiguous_bitmask_p, s390_rtx_costs)
31075 (legitimize_pic_address): Likewise.
31076 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31077 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31078 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31079 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31080 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31082 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31084 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31085 REGNO($0) == REGNO($1).
31087 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31089 * config/s390/linux.h(SIZE_TYPE): Add comment.
31091 2017-02-06 Julian Brown <julian@codesourcery.com>
31092 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31093 Virendra Pathak <virendra.pathak@broadcom.com>
31095 * config/aarch64/aarch64-cores.def: Change the scheduler
31097 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31098 * config/aarch64/thunderx2t99.md: New file.
31100 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31102 * doc/standards.texi (Go Language): Update link to language
31105 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31107 * tree-eh.c (lower_resx): Sanitize profile.
31108 (cleanup_empty_eh_move_lp): Likewise.
31110 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31113 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31115 * cfgloopmanip.h (loop_version): Update prototype.
31116 * modulo-sched.c (sms_schedule): Update call of loop_version.
31117 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31118 * tree-parloops.c (gen_parallel_loop): Likewise.
31119 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31120 * tree-ssa-loop-split.c (split_loop): Likewise.
31121 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31122 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31124 2017-02-05 Martin Liska <mliska@suse.cz>
31127 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31129 (print_operand_address): Initialize a struct to zero.
31131 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31133 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31134 garbage collector only in textual form.
31136 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31138 * doc/extend.texi (x86 specific memory model extensions for
31139 transactional memory): Simplify a phrase.
31141 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31144 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31145 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31146 (atomic_storedi_1): Likewise.
31148 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31150 PR tree-optimization/79338
31151 * tree-parloops.c (gather_scalar_reductions): Don't call
31152 vect_analyze_loop_form for loop->inner before destroying loop's
31155 2017-02-03 Martin Sebor <msebor@redhat.com>
31157 PR tree-optimization/79327
31158 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31159 when precision has resulted in leading zeros.
31160 (format_integer): Adjust the likely counter to assume an unknown
31161 argument that may be zero is non-zero.
31163 2017-02-03 Jason Merrill <jason@redhat.com>
31166 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31167 avoid copying non-taken branch.
31169 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31171 PR tree-optimization/79340
31172 * tree-vect-loop.c (vectorizable_reduction): Release
31173 vec_defs elements after safe_splicing them into other vectors.
31176 PR tree-optimization/79327
31177 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31178 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31180 (format_integer): Use wide_int_to_tree instead of build_int_cst
31181 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31182 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31183 of shortest and longest sequence.
31185 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31187 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31188 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31190 2017-02-03 Walter Lee <walt@tilera.com>
31193 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31194 after initial stackframe link reg save.
31195 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31197 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31200 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31201 wu for stxssp alternative.
31203 2017-02-03 Martin Sebor <msebor@redhat.com>
31205 PR tree-optimization/79352
31206 * gimple-fold.c (get_range_strlen): Add argument.
31207 (get_range_strlen): Change return type to bool.
31208 (get_maxval_strlen): Pass in a dummy argument.
31209 * gimple-fold.h (get_range_strlen): Change return type to bool.
31210 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31211 * tree.h (array_at_struct_end_p): Add argument.
31212 * tree.c (array_at_struct_end_p): Handle it.
31214 2017-02-03 Martin Liska <mliska@suse.cz>
31217 * multiple_target.c (create_dispatcher_calls): Redirect edge
31218 from a caller of a dispatcher.
31219 (expand_target_clones): Make the clones local.
31220 (ipa_target_clone): Do both target clones and resolvers.
31221 (ipa_dispatcher_calls): Remove the pass.
31222 (pass_dispatcher_calls::gate): Likewise.
31223 (make_pass_dispatcher_calls): Likewise.
31224 * passes.def (pass_target_clone): Put as very first IPA early
31227 2017-02-03 Martin Liska <mliska@suse.cz>
31229 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31230 in case of a function with ifunc attribute.
31232 2017-02-03 Martin Liska <mliska@suse.cz>
31234 * cgraph.c (cgraph_node::dump): Dump function version info.
31235 * symtab.c (symtab_node::dump_base): Add missing new line.
31237 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31239 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31240 (ifcombine_ifandif): Use it.
31242 2017-02-03 Martin Liska <mliska@suse.cz>
31244 * doc/invoke.texi: Document default value for
31245 use-after-scope-direct-emission-threshold.
31247 2017-02-03 Martin Liska <mliska@suse.cz>
31249 PR tree-optimization/79339
31250 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31251 (format_floating): Likewise.
31253 2017-02-03 Martin Liska <mliska@suse.cz>
31256 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31258 (ipa_node_params_t::remove): Likewise.
31259 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31260 initialization from removed ipa_node_params_t::insert.
31261 (ipa_node_params::~ipa_node_params): Move from removed
31262 ipa_node_params_t::release.
31263 * symbol-summary.h (symbol_summary::m_released): New member.
31264 Do not release a summary twice. Do not allow to call finalizer
31265 for types of a summary that live in GGC memory.
31267 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31269 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31272 2017-02-02 Martin Sebor <msebor@redhat.com>
31274 PR middle-end/79275
31275 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31276 (format_string): Tighten up the range of output for non-constant
31277 strings and correct the expected range for wide non-constant strings.
31279 2017-02-02 Martin Sebor <msebor@redhat.com>
31281 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31283 PR middle-end/32003
31284 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31286 (-fdump-tree-@var): Add to index and document how to come up
31287 with pass-specific option and dump file names.
31288 (-fdump-passes): Clarify where to look for output.
31290 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31292 PR middle-end/77445
31293 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31294 statistics of the analyzed path; allow threading for speed when
31295 any of BBs along the path are optimized for speed.
31297 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31299 PR middle-end/78468
31300 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31301 settings of the virtual registers.
31304 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31306 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31307 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31310 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31312 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31313 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31315 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31317 * config/s390/s390.md: Add missing comments with the expanded
31319 * config/s390/vector.md: Likewise.
31320 * config/s390/vx-builtins.md: Likewise.
31322 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31325 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31326 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31327 conditions on a single line.
31329 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31331 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31332 __S390_VX__ to __VX__.
31334 2017-02-01 Andrew Pinski <apinski@cavium.com>
31336 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31337 stmt_info to record_stmt_cost.
31338 (vect_get_known_peeling_cost): Pass stmt_info if known to
31340 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31341 cpu_vector_cost field into
31342 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31343 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31344 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31345 splitting of scalar_stmt_cost and vec_stmt_cost.
31346 (thunderx_vector_cost): Likewise.
31347 (cortexa57_vector_cost): LIkewise.
31348 (exynosm1_vector_cost): Likewise.
31349 (xgene1_vector_cost): Likewise.
31350 (thunderx2t99_vector_cost): Improve after the splitting of the two
31352 (aarch64_builtin_vectorization_cost): Update for the splitting of
31353 scalar_stmt_cost and vec_stmt_cost.
31355 2017-02-01 Torvald Riegel <triegel@redhat.com>
31356 Richard Henderson <rth@redhat.com>
31358 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31359 conditional on existance of a fast atomic load.
31360 * optabs-query.c (can_atomic_load_p): New function.
31361 * optabs-query.h (can_atomic_load_p): Declare it.
31362 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31363 no fast atomic load is available for the particular size of access.
31364 (expand_atomic_compare_and_swap): Likewise.
31365 (expand_atomic_load): Likewise.
31366 (expand_atomic_store): Likewise.
31367 (expand_atomic_fetch_op): Likewise.
31368 * testsuite/lib/target-supports.exp
31369 (check_effective_target_sync_int_128): Remove x86 because it provides
31370 no fast atomic load.
31371 (check_effective_target_sync_int_128_runtime): Likewise.
31373 2017-02-01 Richard Biener <rguenther@suse.de>
31375 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31376 (graphite_transform_loops): Provide opt-info for optimized nests.
31377 * tree-parloop.c (parallelize_loops): Provide opt-info for
31378 parallelized loops.
31380 2017-02-01 Richard Biener <rguenther@suse.de>
31382 PR middle-end/79315
31383 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31384 was not set before.
31386 2017-02-01 Richard Biener <rguenther@suse.de>
31388 PR tree-optimization/71824
31389 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31390 Verify the loops are valid in the merged SESE region.
31391 (scop_detection::can_represent_loop_1): Check analyzing the
31392 evolution of the number of iterations in the region succeeds.
31394 2017-01-31 Ian Lance Taylor <iant@golang.org>
31396 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31397 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31399 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31401 PR preprocessor/79210
31402 * input.c (get_substring_ranges_for_loc): Replace line_width
31403 assertion with error-handling.
31405 2017-01-31 Richard Biener <rguenther@suse.de>
31407 PR tree-optimization/77318
31408 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31409 (create_pw_aff_from_tree): Take loop parameter.
31410 (add_condition_to_pbb): Pass loop of the condition to
31411 create_pw_aff_from_tree.
31413 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31415 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31416 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31418 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31422 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31424 (convert_int_to_float128): Likewise.
31425 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31426 (convert_int_to_float128): Likewise.
31427 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31428 (UNSPEC_IEEE128_CONVERT): Likewise.
31429 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31430 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31431 Use local variables for IBM extended format.
31432 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31433 (fix_trunc<mode>si2_fprs): Likewise.
31434 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31435 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31436 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31437 to know that we can now have integers of all sizes in vector
31439 (fix<uns>_<mode>di2_hw): Likewise.
31440 (float<uns>_<mode>si2_hw): Likewise.
31441 (fix_<mode>si2_hw): Likewise.
31442 (fixuns_<mode>si2_hw): Likewise.
31443 (float<uns>_<mode>di2_hw): Likewise.
31444 (float_<mode>di2_hw): Likewise.
31445 (float_<mode>si2_hw): Likewise.
31446 (floatuns_<mode>di2_hw): Likewise.
31447 (floatuns_<mode>si2_hw): Likewise.
31448 (xscvqp<su>wz_<mode>): Delete, no longer used.
31449 (xscvqp<su>dz_<mode>): Likewise.
31450 (xscv<su>dqp_<mode>): Likewise.
31451 (ieee128_mfvsrd_64bit): Likewise.
31452 (ieee128_mfvsrd_32bit): Likewise.
31453 (ieee128_mfvsrwz): Likewise.
31454 (ieee128_mtvsrw): Likewise.
31455 (ieee128_mtvsrd_64bit): Likewise.
31456 (ieee128_mtvsrd_32bit): Likewise.
31458 2017-01-31 Martin Liska <mliska@suse.cz>
31461 * ipa-prop.c (ipa_free_all_node_params): Call release method
31462 instead of ~sumbol_summary to not to trigger double times
31465 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31467 PR tree-optimization/71691
31468 * bitmap.h (class auto_bitmap): New.
31469 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31470 is_maybe_undefined instead of ssa_undefined_value_p.
31472 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31474 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31475 __S390_ARCH_LEVEL__ to __ARCH__.
31477 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31479 PR tree-optimization/79267
31480 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31481 if should_remove_lhs_p is true.
31483 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31486 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31487 (add_alignment_attribute): New.
31488 (base_type_die): Add alignment attribute.
31489 (subrange_type_die): Likewise.
31490 (modified_type_die): Likewise.
31491 (gen_array_type_die): Likewise.
31492 (gen_descr_array_type_die: Likewise.
31493 (gen_enumeration_type_die): Likewise.
31494 (gen_subprogram_die): Likewise.
31495 (gen_variable_die): Likewise.
31496 (gen_field_die): Likewise.
31497 (gen_ptr_to_mbr_type_die): Likewise.
31498 (gen_struct_or_union_type_die): Likewise.
31499 (gen_subroutine_type_die): Likewise.
31500 (gen_typedef_die): Likewise.
31501 (base_type_cmp): Compare alignment attribute.
31503 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31506 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31507 (setb_unsigned) New pattern for setb with CCUNS.
31508 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31509 subfc./subfe sequence to avoid overflow problems. Generate a
31510 shorter sequence with cmpld/setb for power9.
31511 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31512 for generating subfc. instruction.
31513 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31514 now uses this instruction.
31516 2017-01-30 Ian Lance Taylor <iant@google.com>
31519 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31520 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31522 2017-01-30 Martin Sebor <msebor@redhat.com>
31524 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31525 Move constant to the right of a relational operator.
31526 (get_mpfr_format_length, format_character, format_string): Ditto.
31527 (should_warn_p, maybe_warn): Same.
31529 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31531 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31534 * asan.c (get_translation_unit_decl): Remove function.
31535 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31537 2017-01-30 Martin Liska <mliska@suse.cz>
31539 PR gcov-profile/79259
31540 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31541 -fprofile-generate.
31543 2017-01-30 Martin Liska <mliska@suse.cz>
31546 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31547 Initialize variables with NULL value.
31549 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31552 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31554 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31556 2017-01-30 Richard Biener <rguenther@suse.de>
31558 PR tree-optimization/79276
31559 * tree-vrp.c (process_assert_insertions): Properly adjust common
31560 when removing a duplicate.
31562 2017-01-30 Richard Biener <rguenther@suse.de>
31564 PR tree-optimization/79256
31565 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31566 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31568 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31570 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31573 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31574 ("*r<noxa>sbg_<mode>_sll_bitmask")
31575 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31576 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31577 Use contiguous_bitmask_nowrap_operand.
31579 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31582 * config/rs6000/altivec.h (vec_xl): Revise #define.
31583 (vec_xst): Likewise.
31585 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31587 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31589 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31591 PR rtl-optimization/79194
31592 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31593 traps before call to bypass_conditional_jumps.
31595 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31597 PR tree-optimization/71374
31598 * lra-constraints.c (check_conflict_input_operands): New.
31599 (match_reload): Use it.
31601 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31604 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31605 account to calculate conflict_set.
31607 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31609 PR rtl-optimization/78559
31610 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31611 other_insn in combine.
31613 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31615 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31616 uint16_type_node for BT_UINT16.
31618 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31620 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31621 "RTL Tests" to menu.
31622 (GIMPLE Tests): New node.
31623 (RTL Tests): New node.
31625 2017-01-27 Richard Biener <rguenther@suse.de>
31627 PR tree-optimization/79245
31628 * tree-loop-distribution.c (distribute_loop): Apply cost
31629 modeling also to detected patterns.
31631 2017-01-27 Richard Biener <rguenther@suse.de>
31633 PR tree-optimization/71433
31634 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31635 (compare_assert_loc): New function.
31636 (process_assert_insertions): Sort and optimize assert locations
31637 to remove duplicates and push down identical assertions on
31638 edges to their destination block.
31640 2017-01-27 Richard Biener <rguenther@suse.de>
31642 PR tree-optimization/79244
31643 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31644 out SSA names even if abnormal.
31646 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31648 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31649 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31650 instead of MPFR_RNDN.
31652 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31655 * arm.c (arm_option_override): Don't call build_target_option_node
31656 until after doing all option overrides.
31657 (arm_valid_target_attribute_tree): Likewise.
31659 2017-01-27 Martin Liska <mliska@suse.cz>
31661 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31662 for {cd}tors and C++ {cd}tors.
31664 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31666 * config/s390/s390.md ("*setmem_long_and")
31667 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31669 2017-01-26 Martin Sebor <msebor@redhat.com>
31671 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31674 2017-01-26 Martin Sebor <msebor@redhat.com>
31676 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31677 HAVE_DFmode before using XFmode or DFmode.
31678 (parse_directive): Avoid using the z length modifier to avoid
31679 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31681 PR middle-end/78703
31682 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31683 to accept adjustment as an array.
31684 (get_int_range): New function.
31685 (struct directive): Make width and prec arrays.
31686 (directive::set_width, directive::set_precision): Call get_int_range.
31687 (format_integer, format_floating): Handle width and precision ranges.
31688 (format_string, parse_directive): Same.
31690 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31693 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31694 comdat_type_p set, just clone them, but keep the children in the
31698 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31699 which have direct callers with -fvar-tracking-assignments enabled
31701 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31702 inside of type units.
31704 2017-01-26 Martin Sebor <msebor@redhat.com>
31706 PR middle-end/78703
31707 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31709 (struct format_result): Replace number_chars, number_chars_min,
31710 and number_chars_max with a single member of struct result_range.
31712 (format_result::operator+=): Adjust.
31713 (struct fmtresult): Remove bounded. Handle likely and unlikely
31715 (fmtresult::adjust_for_width_or_precision): New function.
31716 (fmtresult:type_max_digits): New function.
31717 (bytes_remaining): Handle likely and unlikely counters.
31718 (min_bytes_remaining): Remove.
31719 (format_percent): Simplify.
31720 (format_integer, format_floating): Set likely and unlikely counters.
31721 (get_string_length, format_character, format_string): Same.
31722 (format_plain, should_warn_p): New function.
31723 (maybe_warn): Call should_warn_p. Update diagnostic messages
31724 and handle those for all directives, including plain strings.
31725 (format_directive): Handle likely and unlikely counters.
31726 Remove unnecessary quoting from diagnostics. Add an informational
31728 (add_bytes): Remove.
31729 (pass_sprintf_length::compute_format_length): Simplify.
31730 (try_substitute_return_value): Handle likely and unlikely counters.
31732 2017-01-26 Carl Love <cel@us.ibm.com>
31734 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31735 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31737 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31740 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31741 endianess for subregs into account.
31742 * lra-constraints.c (lra_constraints): Do risky transformations
31743 always on the first iteration.
31744 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31745 last_call_used_reg_set.
31746 (process_bb_lives): Define and use last_call_used_reg_set.
31747 * lra.c (lra): Always continue after lra_constraints on the first
31750 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31752 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31754 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31756 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31758 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31759 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31760 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31761 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31762 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31763 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31764 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31765 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31766 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31768 2017-01-26 Marek Polacek <polacek@redhat.com>
31771 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31772 for the third operand.
31774 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31776 PR middle-end/79236
31777 * omp-low.c (struct omp_context): Add simt_stmt field.
31778 (scan_omp_for): Return omp_context *.
31779 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31780 context to the _simt_ SIMD stmt.
31781 (lower_omp_for): For combined SIMD with sibling _simt_
31782 SIMD, make sure to use the same decls in _looptemp_
31783 clauses as in the sibling.
31785 2017-01-26 David Sherwood <david.sherwood@arm.com>
31787 PR middle-end/79212
31788 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31791 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31794 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31795 emit fld b; fld a; if possible.
31797 * brig-builtins.def: Update copyright years.
31798 * config/arm/arm_acle_builtins.def: Update copyright years.
31800 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31803 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31804 constraint instead of o for the stxsd instruction.
31806 2017-01-25 Carl Love <cel@us.ibm.com>
31808 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31809 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31811 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31813 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31815 2017-01-25 Richard Biener <rguenther@suse.de>
31817 PR tree-optimization/69264
31818 * target.def (vector_alignment_reachable): Improve documentation.
31819 * doc/tm.texi: Regenerate.
31820 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31822 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31823 earlier changes with respect to TYPE_USER_ALIGN.
31824 (vector_alignment_reachable_p): Likewise. Improve dumping.
31826 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31829 * config/arm/arm.md (xordi3): Force constant operand into a register
31832 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31834 * doc/invoke.texi (-fstore-merging): Correct default optimization
31835 levels at which it is enabled.
31836 (-O): Move -fstore-merging from list to...
31839 2017-01-25 Richard Biener <rguenther@suse.de>
31842 * omp-expand.c: Include debug.h.
31843 (expand_omp_taskreg): Make sure to generate early debug before
31844 outlining anything from a function.
31845 (expand_omp_target): Likewise.
31846 (grid_expand_target_grid_body): Likewise.
31848 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31851 * asan.c (get_translation_unit_decl): New function.
31852 (asan_add_global): Extract modules file name from globals
31853 TRANSLATION_UNIT_DECL name.
31855 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31858 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31859 for long calls with APCS frame and VFP.
31861 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31863 * cfg.c (original_copy_tables_initialized_p): New function.
31864 * cfg.h (original_copy_tables_initialized_p): New decl.
31865 * cfgrtl.c (relink_block_chain): Guard the call to
31866 free_original_copy_tables with a call to
31867 original_copy_tables_initialized_p.
31868 * cgraph.h (symtab_node::native_rtl_p): New decl.
31869 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31870 (symtab_node::needed_p): Don't assert for early assembly output
31871 for __RTL functions.
31872 (cgraph_node::finalize_function): Set "force_output" for __RTL
31874 (cgraph_node::analyze): Bail out early for __RTL functions.
31875 (analyze_functions): Update assertion to support __RTL functions.
31876 (cgraph_node::expand): Bail out early for __RTL functions.
31877 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31879 * function.h (struct function): Update comment for field
31881 * gimple-expr.c: Include "tree-pass.h".
31882 (gimple_has_body_p): Return false for __RTL functions.
31883 * Makefile.in (OBJS): Add run-rtl-passes.o.
31884 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31886 (gcc::pass_manager::get_clean_slate): New accessor.
31887 * passes.c: Include "insn-addr.h".
31888 (should_skip_pass_p): Add logging. Update logic for running
31889 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31890 property-provider override so it is only done for gimple passes.
31892 (skip_pass): New function.
31893 (execute_one_pass): Call skip_pass when skipping passes.
31894 * read-md.c (md_reader::read_char): Support filtering
31895 the input to a subset of line numbers.
31896 (md_reader::md_reader): Initialize fields
31897 m_first_line and m_last_line.
31898 (md_reader::read_file_fragment): New function.
31899 * read-md.h (md_reader::read_file_fragment): New decl.
31900 (md_reader::m_first_line): New field.
31901 (md_reader::m_last_line): New field.
31902 * read-rtl-function.c (function_reader::create_function): Only
31903 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31904 curr_properties. Set DECL_INITIAL to a dummy block.
31905 (read_rtl_function_body_from_file_range): New function.
31906 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31908 * run-rtl-passes.c: New file.
31909 * run-rtl-passes.h: New file.
31911 2017-01-24 Jeff Law <law@redhat.com>
31913 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31916 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31918 PR tree-optimization/79159
31919 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31920 (record_nonwrapping_iv): Improve boundary using above function if no
31921 value range information.
31923 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31924 Martin Jambor <mjambor@suse.cz>
31926 * brig-builtins.def: New file.
31927 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31928 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31929 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31930 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31931 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31932 * builtin-types.def (BT_INT8): New.
31933 (BT_INT16): Likewise.
31934 (BT_UINT8): Likewise.
31935 (BT_UINT16): Likewise.
31936 (BT_FN_ULONG): Likewise.
31937 (BT_FN_UINT_INT): Likewise.
31938 (BT_FN_UINT_ULONG): Likewise.
31939 (BT_FN_UINT_LONG): Likewise.
31940 (BT_FN_UINT_PTR): Likewise.
31941 (BT_FN_ULONG_PTR): Likewise.
31942 (BT_FN_INT8_FLOAT): Likewise.
31943 (BT_FN_INT16_FLOAT): Likewise.
31944 (BT_FN_UINT32_FLOAT): Likewise.
31945 (BT_FN_UINT16_FLOAT): Likewise.
31946 (BT_FN_UINT8_FLOAT): Likewise.
31947 (BT_FN_UINT64_FLOAT): Likewise.
31948 (BT_FN_UINT16_UINT32): Likewise.
31949 (BT_FN_UINT32_UINT16): Likewise.
31950 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31951 (BT_FN_INT_PTR_INT): Likewise.
31952 (BT_FN_UINT_PTR_UINT): Likewise.
31953 (BT_FN_LONG_PTR_LONG): Likewise.
31954 (BT_FN_ULONG_PTR_ULONG): Likewise.
31955 (BT_FN_VOID_UINT64_UINT64): Likewise.
31956 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31957 (BT_FN_INT8_INT8_INT8): Likewise.
31958 (BT_FN_INT16_INT16_INT16): Likewise.
31959 (BT_FN_INT_INT_INT): Likewise.
31960 (BT_FN_UINT_FLOAT_UINT): Likewise.
31961 (BT_FN_FLOAT_UINT_UINT): Likewise.
31962 (BT_FN_ULONG_UINT_UINT): Likewise.
31963 (BT_FN_ULONG_UINT_PTR): Likewise.
31964 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31965 (BT_FN_UINT_UINT_UINT): Likewise.
31966 (BT_FN_VOID_UINT_PTR): Likewise.
31967 (BT_FN_UINT_UINT_PTR: Likewise.
31968 (BT_FN_UINT32_UINT64_PTR): Likewise.
31969 (BT_FN_INT_INT_UINT_UINT): Likewise.
31970 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31971 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31972 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31973 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31974 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31975 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31976 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31977 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31978 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31979 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31980 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31981 * doc/frontends.texi: List BRIG FE.
31982 * doc/install.texi (Testing): Add BRIG tesring requirements.
31983 * doc/invoke.texi (Overall Options): Mention BRIG.
31984 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31986 2017-01-24 Richard Biener <rguenther@suse.de>
31988 PR translation/79208
31989 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31991 2017-01-24 Martin Jambor <mjambor@suse.cz>
31994 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31995 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31996 and known_contexts.
31998 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32000 PR middle-end/79123
32001 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32002 casts from signed to unsigned really don't have a range.
32004 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32006 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32007 GMP_RNDx for compatiblity.
32009 2017-01-24 Martin Liska <mliska@suse.cz>
32012 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32013 that would prevent us to call alloca with -1 as argument.
32015 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32017 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32018 Avoid -Wformat-security warning.
32020 2017-01-23 Andrew Pinski <apinski@cavium.com>
32022 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32025 2017-01-23 Martin Sebor <msebor@redhat.com>
32027 PR middle-end/78703
32028 * gimple-ssa-sprintf.c (warn_level): New global.
32029 (format_integer): Use it here and throughout the rest of the file.
32030 Use the same switch to compute sign as base.
32031 (maybe_warn): New function.
32032 (format_directive): Factor out warnings into maybe_warn.
32033 Add debugging output. Use warn_level.
32034 (add_bytes): Use warn_level.
32035 (pass_sprintf_length::compute_format_length): Add debugging output.
32036 (try_substitute_return_value): Same.
32037 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32039 PR middle-end/78703
32040 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32041 (struct fmtresult, format_integer, format_floating): Adjust.
32042 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32043 (get_string_length, format_string,format_directive): Same.
32044 (pass_sprintf_length::compute_format_length): Same.
32045 (try_substitute_return_value): Simplify slightly.
32047 PR middle-end/78703
32048 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32049 (fmtresult::operator+=): Outlined.
32050 (struct fmtresult): Add ctors.
32051 (struct conversion_spec): Rename...
32052 (struct directive): ...to this. Add and remove data members.
32053 (directive::set_width, directive::set_precision): New functions.
32054 (format_percent): Use fmtresult ctor.
32055 (get_width_and_precision): Remove.
32056 (format_integer): Make naming changes. Avoid computing width and
32058 (format_floating): Same. Adjust indentation.
32059 (format_character, format_none): New functions.
32060 (format_string): Moved character handling to format_character.
32061 (format_directive): Remove arguments, change return type.
32062 (parse_directive): New function.
32063 (pass_sprintf_length::compute_format_length): Move directive
32064 parsing to parse_directive.
32066 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32068 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32069 (assign_assembler_name_if_needed): ... this.
32070 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32071 (assign_assembler_name_if_needed): ... this.
32072 (free_lang_data_in_cgraph): Adjust callers.
32073 * cgraphunit.c (cgraph_node::analyze): Likewise.
32074 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32077 2017-01-23 Richard Biener <rguenther@suse.de>
32079 PR tree-optimization/79088
32080 PR tree-optimization/79188
32081 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32082 resetting loop bounds after last path deletion. Reset loop
32083 bounds of the target loop, make code match the comments.
32084 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32085 Make sure loops need no fixups.
32087 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32089 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32090 exponent support with double type for first argument.
32091 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32092 type returned by __builtin_vec_extract_sig,
32093 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32094 functions from "vector int" to "vector unsigned int" or from
32095 "vector long long int" to "vector unsigned long long int".
32096 Changed type returned by __builtin_vec_extract_exp,
32097 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32098 functions from "vector int" to "vector unsigned int" or from
32099 "vector long long int" to "vector unsigned long long int".
32100 Changed return type of __builtin_vec_test_data_class,
32101 __builtin_vec_test_data_class_sp, and
32102 __builtin_vec_test_data_class_dp from "vector int" to
32103 "vector bool int" or from "vector long long int" to "vector bool
32104 long long int" and changed second argument type from "unsigned
32105 int" to "int". Added new overloaded function forms "vector float
32106 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32107 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32108 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32109 double, vector unsigned long long int)" and "vector double
32110 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32111 long int)". Changed return type of
32112 __builtin_scalar_test_data_class and
32113 __builtin_scalar_test_data_class_sp and
32114 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32115 int" and changed second argument from "unsigned int" to "int".
32116 Changed type returned by __builtin_scalar_test_neg,
32117 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32118 from "int" to "bool int". Added new overloaded function form
32119 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32120 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32121 exponent double-precision with floating point first argument.
32122 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32123 documentation of scalar_test_data_class, scalar_test_neg,
32124 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32125 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32126 vec_test_data_class built-in functions to reflect refinements in
32127 their type signatures.
32129 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32131 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32133 (aarch64_elf_asm_destructor): Likewise.
32135 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32137 PR rtl-optimization/78634
32138 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32139 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32140 * ifcvt.c (noce_try_cmove): Add missing cost check.
32142 PR rtl-optimization/71724
32143 * combine.c (if_then_else_cond): Look for situations where it is
32144 beneficial to undo the work of one of the recursive calls.
32146 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32148 PR tree-optimization/70754
32149 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32150 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32151 combined stmt before it if not NULL.
32152 (combine_chains): Process refs reversely and compute dominance point
32155 2017-01-23 Martin Liska <mliska@suse.cz>
32157 PR tree-optimization/79196
32158 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32159 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32161 (strlen_optimize_stmt): Call the renamed function.
32163 2017-01-23 Michael Matz <matz@suse.de>
32165 PR tree-optimization/78384
32166 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32168 2017-01-23 Richard Biener <rguenther@suse.de>
32170 PR tree-optimization/79186
32171 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32172 both incoming edges before moving an assert.
32174 2017-01-23 Martin Jambor <mjambor@suse.cz>
32176 * ipa-prop.c (load_from_param_1): Removed.
32177 (load_from_unmodified_param): Bits from load_from_param_1 put back
32179 (load_from_param): Removed.
32180 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32181 with stmt. Reverted back to use of load_from_unmodified_param.
32183 2017-01-23 Martin Jambor <mjambor@suse.cz>
32186 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32187 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32188 field a pointer to garbage collected vector, mark lattices and
32189 ipcp_orig_node with GTY((skip)).
32190 (ipa_get_param_count): Adjust to descriptors being a pointer.
32191 (ipa_get_param): Likewise.
32192 (ipa_get_type): Likewise.
32193 (ipa_get_param_move_cost): Likewise.
32194 (ipa_set_param_used): Likewise.
32195 (ipa_get_controlled_uses): Likewise.
32196 (ipa_set_controlled_uses): Likewise.
32197 (ipa_is_param_used): Likewise.
32198 (ipa_node_params_t): Move into garbage collector. New methods insert
32200 (ipa_node_params_sum): Annotate wth GTY(()).
32201 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32203 (ipa_load_from_parm_agg): Adjust declaration.
32204 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32205 * ipa-profile.c (ipa_profile): Likewise.
32206 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32207 (ipa_populate_param_decls): Make descriptors parameter garbage
32209 (ipa_dump_param): Adjust to descriptors being a pointer.
32210 (ipa_alloc_node_params): Likewise.
32211 (ipa_initialize_node_params): Likewise.
32212 (load_from_param_1): Make descriptors parameter garbage collected.
32213 (load_from_unmodified_param): Likewise.
32214 (load_from_param): Likewise.
32215 (ipa_load_from_parm_agg): Likewise.
32216 (ipa_node_params::~ipa_node_params): Removed.
32217 (ipa_free_all_node_params): Remove call to delete operator.
32218 (ipa_node_params_t::insert): New.
32219 (ipa_node_params_t::remove): Likewise.
32220 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32221 copy known_csts and known_contexts vectors.
32222 (ipa_read_node_info): Adjust to descriptors being a pointer.
32223 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32225 (ipcp_transform_function): Make descriptors variable garbage
32228 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32230 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32231 * config/i386/avx512dqintrin.h: Ditto.
32232 * config/i386/avx512fintrin.h: Ditto.
32233 * config/i386/i386.c: Handle new builtins.
32234 * config/i386/i386-builtin.def: Add new builtins.
32235 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32236 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32238 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32239 Martin Liska <mliska@suse.cz>
32241 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32242 * asan.c (asan_expand_poison_ifn): Support stores and use
32243 appropriate ASAN report function.
32244 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32245 * internal-fn.def (ASAN_POISON_USE): Declare.
32246 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32247 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32248 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32249 ASAN_POISON calls w/o LHS.
32250 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32251 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32252 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32253 * gimplify.c (asan_poison_variables): Add attribute
32254 use_after_scope_memory to variables that really needs to live
32256 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32257 having the attribute.
32259 2017-01-23 Martin Liska <mliska@suse.cz>
32261 * asan.c (create_asan_shadow_var): New function.
32262 (asan_expand_poison_ifn): Likewise.
32263 * asan.h (asan_expand_poison_ifn): New declaration.
32264 * internal-fn.c (expand_ASAN_POISON): Likewise.
32265 * internal-fn.def (ASAN_POISON): New builtin.
32266 * sanopt.c (pass_sanopt::execute): Expand
32267 asan_expand_poison_ifn.
32268 * tree-inline.c (copy_decl_for_dup_finish): Make function
32270 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32271 * tree-ssa.c (is_asan_mark_p): New function.
32272 (execute_update_addresses_taken): Rewrite local variables
32273 (identified just by use-after-scope as addressable) into SSA.
32275 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32277 * doc/install.texi (Specific): opensource.apple.com uses https
32278 now. Remove trailing slash.
32280 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32282 * README.Portability: Remove note on an Irix compatibility issue.
32284 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32286 * gcov.c (INCLUDE_ALGORITHM): Define.
32287 (INCLUDE_VECTOR): Define.
32288 No longer include <vector> and <algorithm> directly.
32290 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32292 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32294 * doc/invoke.texi (Code Gen Options): Ditto.
32296 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32299 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32301 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32303 rtl-optimization/79125
32304 * cprop.c (local_cprop_pass): Handle cases where we make an
32305 unconditional trap.
32307 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32311 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32312 read from, for big endian.
32314 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32316 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32317 register pauth builtins for LP64 only.
32319 2017-01-20 Marek Polacek <polacek@redhat.com>
32322 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32325 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32327 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32329 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32330 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32331 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32333 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32336 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32337 in big-endian mode when they are not a single duplicated value.
32339 2017-01-20 Richard Biener <rguenther@suse.de>
32341 * BASE-VER: Bump to 7.0.1.
32343 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32345 * omp-low.c (omplow_simd_context): New struct. Use it...
32346 (lower_rec_simd_input_clauses): ...here and...
32347 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32348 references to idx, lane, max_vf, is_simt.
32350 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32352 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32355 2017-01-20 Martin Jambor <mjambor@suse.cz>
32357 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32358 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32360 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32361 (GTFILES): Rename hsa.c to hsa-common.c.
32362 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32363 * hsa-dump.c: Likewise.
32364 * hsa-gen.c: Likewise.
32365 * hsa-regalloc.c: Likewise.
32366 * ipa-hsa.c: Likewise.
32367 * omp-expand.c: Likewise.
32368 * omp-low.c: Likewise.
32369 * toplev.c: Likewise.
32371 2017-01-20 Marek Polacek <polacek@redhat.com>
32374 * doc/invoke.texi: Document -Wduplicated-branches.
32375 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32376 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32377 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32378 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32379 return 0 only when not OEP_LEXICOGRAPHIC.
32380 (fold_build_cleanup_point_expr): Use the expression
32381 location when building CLEANUP_POINT_EXPR.
32382 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32383 * tree.c (add_expr): Handle error_mark_node.
32385 2017-01-20 Martin Liska <mliska@suse.cz>
32388 * tree-profile.c (init_ic_make_global_vars): Do not call
32390 (gimple_init_gcov_profiler): Likewise.
32392 2017-01-20 Martin Liska <mliska@suse.cz>
32395 * cgraph.h (maybe_create_reference): Remove argument and
32397 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32399 * ipa-cp.c (create_specialized_node): Likewise.
32400 * symtab.c (symtab_node::maybe_create_reference): Handle
32401 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32403 2017-01-20 Martin Liska <mliska@suse.cz>
32405 * read-rtl-function.c (function_reader::create_function): Use
32406 build_decl instread of build_decl_stat.
32408 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32410 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32411 * config/i386/avx512dqintrin.h: Ditto.
32412 * config/i386/avx512fintrin.h: Ditto.
32413 * config/i386/i386-builtin-types.def: Add new types.
32414 * config/i386/i386.c: Handle new types.
32415 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32416 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32417 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32418 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32419 (__builtin_ia32_kshiftridi): New.
32420 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32422 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32426 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32427 define to rs6000_init_stack_protect_guard.
32428 (rs6000_init_stack_protect_guard): New function.
32430 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32431 Yunqiang Su <yunqiang.su@imgtec.com>
32433 * config.gcc (supported_defaults): Add madd4.
32434 (with_madd4): Add validation.
32435 (all_defaults): Add madd4.
32436 * config/mips/mips.opt (mmadd4): New option.
32437 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32439 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32441 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32442 (ISA_HAS_FUSED_MADD4): Likewise.
32443 * doc/invoke.texi (-mmadd4): Document the new option.
32444 * doc/install.texi (--with-madd4): Document the new option.
32446 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32448 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32449 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32450 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32451 (aarch64_init_pauth_hint_builtins): New.
32452 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32453 (aarch64_expand_builtin): Expand new builtins.
32455 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32457 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32458 * combine-stack-adj.c (no_unhandled_cfa): Handle
32459 REG_CFA_TOGGLE_RA_MANGLE.
32460 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32461 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32462 info for return address signing.
32463 (aarch64_expand_epilogue): Likewise.
32465 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32467 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32468 * config/aarch64/aarch64-protos.h
32469 (aarch64_return_address_signing_enabled): New declaration.
32470 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32472 (aarch64_expand_prologue): Sign return address before it's pushed onto
32474 (aarch64_expand_epilogue): Authenticate return address fetched from
32476 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32477 (aarch64_attributes): New function attributes for "sign-return-address".
32478 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32479 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32480 ("*do_return"): Generate combined instructions according to key index.
32481 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32482 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32484 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32485 * config/aarch64/aarch64.opt (msign-return-address=): New.
32486 * doc/extend.texi (AArch64 Function Attributes): Documents
32487 "sign-return-address=".
32488 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32490 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32492 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32493 overall option summary.
32495 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32497 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32498 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32499 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32500 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32502 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32504 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32505 -mpower9-minmax by default for -mcpu=power9.
32506 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32507 128-bit floating point.
32509 2017-01-20 Alan Modra <amodra@gmail.com>
32511 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32512 optimizing for size.
32514 2017-01-20 Alan Modra <amodra@gmail.com>
32517 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32518 for strcmp and strncmp from corresponding builtin decl.
32520 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32522 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32523 instead of i386/rtems-64.h.
32524 * config/i386/rtems-64.h: Remove.
32526 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32530 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32532 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32534 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32536 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32537 Change int to HOST_WIDE_INT.
32538 * config/aarch64/aarch64-protos.h
32539 (aarch64_simd_gen_const_vector_dup): Likewise.
32540 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32542 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32544 * langhooks-def.h (lhd_type_for_size): New decl.
32545 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32546 * langhooks.c (lhd_type_for_size): New function, taken from
32549 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32551 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32552 define_bypass for CR latency.
32553 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32554 (power9-alu2): Add define_bypass for CR latency.
32556 (power9-mul): Update insn latency.
32557 (power9-mul-compare): Update insn latency, bypass latency and remove
32560 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32562 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32564 * config/aarch64/aarch64.md
32565 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32566 aarch64_nopcrelative_literal_loads.
32567 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32569 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32571 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32572 TARGET_LOONGSON_3A.
32573 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32575 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32578 * config.gcc (supported_defaults): Add lxc1-sxc1.
32579 (with_lxc1_sxc1): Add validation.
32580 (all_defaults): Add lxc1-sxc1.
32581 * config/mips/mips.opt (mlxc1-sxc1): New option.
32582 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32584 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32585 __mips_no_lxc1_sxc1.
32586 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32587 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32588 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32590 2017-01-19 Richard Biener <rguenther@suse.de>
32592 PR tree-optimization/72488
32593 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32594 sure to restore SSA info.
32595 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32597 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32599 PR rtl-optimization/79121
32600 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32601 of the inner type when shifting an extended value.
32603 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32606 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32607 interposable aliases.
32609 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32612 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32613 Use the evmergelohi instruction.
32614 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32615 (mov_sitf_e500_subreg8_2_be): Likewise.
32616 (mov_sitf_e500_subreg12_2_le): Likewise.
32617 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32618 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32619 (mov_sitf_e500_subreg8_2_le): Likewise.
32620 (mov_sitf_e500_subreg12_2_be): Likewise.
32622 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32624 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32625 attribute from vecsimple to vecperm.
32626 (altivec_vbpermq2): Likewise.
32628 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32631 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32633 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32634 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32635 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32636 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32637 case where N arg is SIZE_MAX.
32638 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32639 (cmpstrsi): Add pattern.
32641 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32643 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32644 __builtin_vec_revb builtins.
32645 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32646 built-in functions to support generation of the ISA 3.0 XXBR<x>
32647 vector byte reverse instructions.
32648 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32649 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32650 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32651 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32652 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32653 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32654 (P9V_BUILTIN_VEC_REVB): Likewise.
32655 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32656 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32657 (p9_xxbrq_v16qi): Likewise.
32658 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32659 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32660 (p9_xxbrh_v8hi): Likewise.
32661 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32662 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32663 vec_revb built-in functions.
32665 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32667 PR rtl-optimization/78952
32668 * config/i386/i386.md (any_extract): New code iterator.
32669 (*insvqi_2): Use any_extract for source operand.
32670 (*insvqi_3): Use any_shiftrt for source operand.
32672 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32674 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32676 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32678 2017-01-18 Matthias Klose <doko@ubuntu.com>
32680 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32682 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32684 * config/rs6000/altivec.h (vec_bperm): Change #define.
32685 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32686 (altivec_vbpermq2): New define_insn.
32687 (altivec_vbpermd): Likewise.
32688 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32689 function interface.
32690 (VBPERMD): Likewise.
32691 (VBPERM): New polymorphic function interface.
32692 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32693 Add entries for P9V_BUILTIN_VEC_VBPERM.
32694 * doc/extend.texi: Add interfaces for vec_bperm.
32696 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32698 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32699 first letter of error messages.
32700 (s390_resolve_overloaded_builtin): Likewise.
32701 * config/s390/s390.c (s390_expand_builtin): Likewise.
32702 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32703 (s390_valid_target_attribute_inner_p): Likewise.
32704 * config/s390/s390.md ("tabort"): Likewise.
32706 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32708 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32709 (ISA_AVOID_DIV_HILO): New macro.
32710 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32711 (ISA_HAS_DDIV): Likewise.
32713 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32715 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32717 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32719 * doc/invoke.texi (fabi-version): Spelling fix.
32721 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32724 * doc/invoke.texi (fabi-version): Mention mangling fix for
32727 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32730 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32732 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32735 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32736 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32738 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32740 (stack_protect_test): Ditto.
32741 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32742 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32744 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32745 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32746 -mstack-protector-guard-offset=.
32747 (RS/6000 and PowerPC Options): Ditto.
32749 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32751 * config/i386/i386.h (MASK_CLASS_P): New define.
32752 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32753 there are no registers from different register sets also when
32754 mask registers are used. Update function comment.
32755 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32756 to (*k/*r) and (*k/*km) alternatives.
32758 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32760 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32761 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32762 (EH_RETURN_HANDLER_RTX): New define.
32763 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32764 Force frame pointer in EH return functions.
32765 (aarch64_expand_epilogue): Add barrier for eh_return.
32766 (aarch64_final_eh_return_addr): Remove.
32767 (aarch64_eh_return_handler_rtx): New function.
32768 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32770 (aarch64_eh_return_handler_rtx): New prototype.
32772 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32774 * config/rs6000/altivec.h (vec_rlmi): New #define.
32775 (vec_vrlnm): Likewise.
32776 (vec_rlnm): Likewise.
32777 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32778 (UNSPEC_VRLNM): Likewise.
32779 (VIlong): New mode iterator.
32780 (altivec_vrl<VI_char>mi): New define_insn.
32781 (altivec_vrl<VI_char>nm): Likewise.
32782 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32784 (VRLDNM): Likewise.
32785 (RLNM): New polymorphic function entry.
32786 (VRLWMI): New monomorphic function entry.
32787 (VRLDMI): Likewise.
32788 (RLMI): New polymorphic function entry.
32789 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32790 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32791 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32794 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32797 * dwarf2out.c (field_byte_offset): Restore the
32798 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32799 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32800 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32803 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32806 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32808 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32811 * dwarf2out.c (add_data_member_location_attribute): For constant
32812 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32813 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32814 DW_AT_byte_size attributes.
32816 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32818 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32819 after forcing to constant memory when the code model is medium.
32821 2017-01-17 Julia Koval <julia.koval@intel.com>
32824 * config/i386/avx512fintrin.h
32825 (_mm512_i32gather_ps): Change __addr type to void const*.
32826 (_mm512_mask_i32gather_ps): Ditto.
32827 (_mm512_i32gather_pd): Ditto.
32828 (_mm512_mask_i32gather_pd): Ditto.
32829 (_mm512_i64gather_ps): Ditto.
32830 (_mm512_mask_i64gather_ps): Ditto.
32831 (_mm512_i64gather_pd): Ditto.
32832 (_mm512_mask_i64gather_pd): Ditto.
32833 (_mm512_i32gather_epi32): Ditto.
32834 (_mm512_mask_i32gather_epi32): Ditto.
32835 (_mm512_i32gather_epi64): Ditto.
32836 (_mm512_mask_i32gather_epi64): Ditto.
32837 (_mm512_i64gather_epi32): Ditto.
32838 (_mm512_mask_i64gather_epi32): Ditto.
32839 (_mm512_i64gather_epi64): Ditto.
32840 (_mm512_mask_i64gather_epi64): Ditto.
32841 (_mm512_i32scatter_ps): Change __addr type to void*.
32842 (_mm512_mask_i32scatter_ps): Ditto.
32843 (_mm512_i32scatter_pd): Ditto.
32844 (_mm512_mask_i32scatter_pd): Ditto.
32845 (_mm512_i64scatter_ps): Ditto.
32846 (_mm512_mask_i64scatter_ps): Ditto.
32847 (_mm512_i64scatter_pd): Ditto.
32848 (_mm512_mask_i64scatter_pd): Ditto.
32849 (_mm512_i32scatter_epi32): Ditto.
32850 (_mm512_mask_i32scatter_epi32): Ditto.
32851 (_mm512_i32scatter_epi64): Ditto.
32852 (_mm512_mask_i32scatter_epi64): Ditto.
32853 (_mm512_i64scatter_epi32): Ditto.
32854 (_mm512_mask_i64scatter_epi32): Ditto.
32855 (_mm512_i64scatter_epi64): Ditto.
32856 (_mm512_mask_i64scatter_epi64): Ditto.
32857 * config/i386/avx512pfintrin.h
32858 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32859 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32860 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32861 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32862 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32863 (_mm512_prefetch_i32scatter_ps): Ditto.
32864 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32865 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32866 (_mm512_prefetch_i64scatter_pd): Ditto.
32867 (_mm512_prefetch_i64scatter_ps): Ditto.
32868 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32869 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32870 * config/i386/avx512vlintrin.h
32871 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32872 (_mm_mmask_i32gather_ps): Ditto.
32873 (_mm256_mmask_i32gather_pd): Ditto.
32874 (_mm_mmask_i32gather_pd): Ditto.
32875 (_mm256_mmask_i64gather_ps): Ditto.
32876 (_mm_mmask_i64gather_ps): Ditto.
32877 (_mm256_mmask_i64gather_pd): Ditto.
32878 (_mm_mmask_i64gather_pd): Ditto.
32879 (_mm256_mmask_i32gather_epi32): Ditto.
32880 (_mm_mmask_i32gather_epi32): Ditto.
32881 (_mm256_mmask_i32gather_epi64): Ditto.
32882 (_mm_mmask_i32gather_epi64): Ditto.
32883 (_mm256_mmask_i64gather_epi32): Ditto.
32884 (_mm_mmask_i64gather_epi32): Ditto.
32885 (_mm256_mmask_i64gather_epi64): Ditto.
32886 (_mm_mmask_i64gather_epi64): Ditto.
32887 (_mm256_i32scatter_ps): Change __addr type to void*.
32888 (_mm256_mask_i32scatter_ps): Ditto.
32889 (_mm_i32scatter_ps): Ditto.
32890 (_mm_mask_i32scatter_ps): Ditto.
32891 (_mm256_i32scatter_pd): Ditto.
32892 (_mm256_mask_i32scatter_pd): Ditto.
32893 (_mm_i32scatter_pd): Ditto.
32894 (_mm_mask_i32scatter_pd): Ditto.
32895 (_mm256_i64scatter_ps): Ditto.
32896 (_mm256_mask_i64scatter_ps): Ditto.
32897 (_mm_i64scatter_ps): Ditto.
32898 (_mm_mask_i64scatter_ps): Ditto.
32899 (_mm256_i64scatter_pd): Ditto.
32900 (_mm256_mask_i64scatter_pd): Ditto.
32901 (_mm_i64scatter_pd): Ditto.
32902 (_mm_mask_i64scatter_pd): Ditto.
32903 (_mm256_i32scatter_epi32): Ditto.
32904 (_mm256_mask_i32scatter_epi32): Ditto.
32905 (_mm_i32scatter_epi32): Ditto.
32906 (_mm_mask_i32scatter_epi32): Ditto.
32907 (_mm256_i32scatter_epi64): Ditto.
32908 (_mm256_mask_i32scatter_epi64): Ditto.
32909 (_mm_i32scatter_epi64): Ditto.
32910 (_mm_mask_i32scatter_epi64): Ditto.
32911 (_mm256_i64scatter_epi32): Ditto.
32912 (_mm256_mask_i64scatter_epi32): Ditto.
32913 (_mm_i64scatter_epi32): Ditto.
32914 (_mm_mask_i64scatter_epi32): Ditto.
32915 (_mm256_i64scatter_epi64): Ditto.
32916 (_mm256_mask_i64scatter_epi64): Ditto.
32917 (_mm_i64scatter_epi64): Ditto.
32918 (_mm_mask_i64scatter_epi64): Ditto.
32919 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32920 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32921 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32922 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32923 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32924 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32925 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32926 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32927 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32928 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32929 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32930 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32931 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32932 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32933 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32934 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32935 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32936 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32937 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32938 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32939 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32940 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32941 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32942 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32943 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32944 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32945 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32946 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32947 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32948 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32949 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32950 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32951 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32952 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32953 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32954 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32955 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32956 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32957 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32958 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32959 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32960 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32961 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32962 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32963 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32964 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32965 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32966 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32967 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32968 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32969 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32970 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32971 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32972 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32973 definitions accordingly.
32975 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32976 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32979 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32982 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32985 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32986 hard regs for inner regno.
32988 2017-01-17 Martin Liska <mliska@suse.cz>
32991 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32992 assumption and add comment.
32994 2017-01-17 Nathan Sidwell <nathan@acm.org>
32996 * ipa-visibility.c (localize_node): New function, broken out of ...
32997 (function_and_variable_visibility): ... here. Call it.
32999 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33001 PR middle-end/77445
33002 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33003 correctly set frequency of oudgoing edge.
33004 (duplicate_thread_path): Fix profile updating.
33006 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33009 * configure.ac: Add GCC_BASE_VER.
33010 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33011 version from BASE-VER file.
33012 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33013 (gcc.o): Depend on $(BASEVER).
33014 * common.opt (dumpfullversion): New option.
33015 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33016 * doc/invoke.texi: Document -dumpfullversion.
33017 * doc/install.texi: Document --with-gcc-major-version-only.
33018 * configure: Regenerated.
33020 2017-01-17 Richard Biener <rguenther@suse.de>
33022 PR tree-optimization/71433
33023 * tree-vrp.c (register_new_assert_for): Merge same asserts
33024 on all incoming edges.
33025 (process_assert_insertions_for): Handle insertions at the
33028 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33030 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33031 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33033 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33036 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33039 2017-01-17 Alan Modra <amodra@gmail.com>
33042 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33043 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33044 symbolic stack limit when pic.
33046 2017-01-16 Martin Sebor <msebor@redhat.com>
33048 PR tree-optimization/78608
33049 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33051 2017-01-16 Jeff Law <law@redhat.com>
33054 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33055 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33056 for several include directories that may be relative to sysroot.
33057 * config/i386/x-mingw32 (gplus_includedir): Define.
33058 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33059 (native_system_includedir): Likewise.
33060 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33061 override if TARGET_SYSTEM_ROOT is defined.
33062 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33064 PR tree-optimization/79090
33065 PR tree-optimization/33562
33066 PR tree-optimization/61912
33067 PR tree-optimization/77485
33068 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33069 and computed trims into the dump file.
33071 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33073 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33075 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33078 * gimplify.c (gimplify_init_constructor): If want_value and
33079 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33083 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33084 sequence. Formatting fixes.
33085 (doloop_optimize): Formatting fixes.
33088 * gcc.c (debug_level_greater_than_spec_func): New function.
33089 (static_spec_functions): Add debug-level-gt spec function.
33090 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33092 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33093 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33094 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33095 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33096 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33097 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33099 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33101 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33102 QImode fixups to general and mask registers only.
33104 2017-01-16 Carl Love <cel@us.ibm.com>
33106 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33107 for built-in functions
33108 vector signed char vec_nabs (vector signed char)
33109 vector signed short vec_nabs (vector signed short)
33110 vector signed int vec_nabs (vector signed int)
33111 vector signed long long vec_nabs (vector signed long long)
33112 vector float vec_nabs (vector float)
33113 vector double vec_nabs (vector double)
33114 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33116 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33117 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33118 * doc/extend.texi: Update the documentation file for the new built-in
33121 2017-01-16 Martin Sebor <msebor@redhat.com>
33123 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33126 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33128 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33129 UNSPEC_VSX__XXSPLTD to require special splat handling.
33131 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33134 * system.h: Poison strndup.
33136 2017-01-16 Alan Modra <amodra@gmail.com>
33139 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33142 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33144 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33146 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33148 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33149 call recog here. Assert that INSN_CODE (insn) is non-negative.
33151 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33154 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33156 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33157 in the currently scheduled RTL fragment.
33159 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33161 PR rtl-optimization/78751
33162 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33165 2017-01-14 Jeff Law <law@redhat.com>
33167 PR tree-optimization/79090
33168 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33169 variable length stores.
33170 (compute_trims): Delete dead assignment to *trim_tail.
33171 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33174 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33176 PR rtl-optimization/78626
33177 PR rtl-optimization/78727
33178 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33179 of a block, and split such blocks after everything else is finished.
33181 2017-01-14 Alan Modra <amodra@gmail.com>
33184 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33185 target legitimate_combined_insn.
33186 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33187 (rs6000_legitimate_combined_insn): New function.
33188 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33190 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33191 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33192 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33194 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33196 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33198 2017-01-13 Jeff Law <law@redhat.com>
33200 PR tree-optimization/33562
33201 PR tree-optimization/61912
33202 PR tree-optimization/77485
33203 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33205 (delete_dead_assignment): Likewise.
33206 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33207 statement to delete_dead_call and delete_dead_assignment.
33209 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33212 * substring-locations.c (format_warning_va): Strengthen case 1 so
33213 that both endpoints of the substring must be within the format
33214 range for just the substring to be printed.
33216 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33218 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33219 * config/i386/i386.c (ix86_target_string): Add missing options
33220 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33221 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33222 flags_other and ix86_target_other to flags2_other. Display unknown
33224 (ix86_valid_target_attribute_inner_p): Add missing options and
33225 reorder options by implied ISAs, as in ix86_target_string.
33227 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33229 * hash-table.h (hash_table::too_empty_p): New function.
33230 (hash_table::expand): Use it.
33231 (hash_table::traverse): Likewise.
33232 (hash_table::empty_slot): Use sizeof (value_type) instead of
33233 sizeof (PTR) to convert bytes to elements. Shrink the table
33234 if the current size is excessive for the current number of
33237 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33239 * ira-costs.c (record_reg_classes): Break from the inner loop
33240 early once alt_fail is known to be true. Update outer loop
33241 handling accordingly.
33243 2017-01-13 Jeff Law <law@redhat.com>
33245 * tree-ssa-dse.c (decrement_count): New function.
33246 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33247 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33248 when we know the partially dead statement is a mem* function.
33250 PR tree-optimization/61912
33251 PR tree-optimization/77485
33252 * tree-ssa-dse.c: Include expr.h.
33253 (maybe_trim_constructor_store): New function.
33254 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33256 PR tree-optimization/33562
33257 PR tree-optimization/61912
33258 PR tree-optimization/77485
33259 * doc/invoke.texi: Document new dse-max-object-size param.
33260 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33261 * tree-ssa-dse.c: Include params.h.
33262 (dse_store_status): New enum.
33263 (initialize_ao_ref_for_dse): New, partially extracted from
33265 (valid_ao_ref_for_dse, normalize_ref): New.
33266 (setup_live_bytes_from_ref, compute_trims): Likewise.
33267 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33268 (maybe_trim_partially_dead_store): Likewise.
33269 (maybe_trim_complex_store): Likewise.
33270 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33271 Track what bytes live from the original store. Return tri-state
33272 for dead, partially dead or live.
33273 (dse_dom_walker): Add constructor, destructor and new private members.
33274 (delete_dead_call, delete_dead_assignment): New extracted from
33276 (dse_optimize_stmt): Make a member of dse_dom_walker.
33277 Use initialize_ao_ref_for_dse.
33279 PR tree-optimization/33562
33280 PR tree-optimization/61912
33281 PR tree-optimization/77485
33282 * sbitmap.h (bitmap_count_bits): Prototype.
33283 (bitmap_clear_range, bitmap_set_range): Likewise.
33284 * sbitmap.c (bitmap_clear_range): New function.
33285 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33287 2017-01-13 Martin Liska <mliska@suse.cz>
33290 * function.c (set_cfun): Add new argument force.
33291 * function.h (set_cfun): Likewise.
33292 * ipa-inline-transform.c (inline_call): Use the function when
33293 strict alising from is dropped for function we inline to.
33295 2017-01-13 Richard Biener <rguenther@suse.de>
33297 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33298 for dumping GIMPLE INTEGER_CSTs.
33300 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33302 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33303 to 201112L since C++17.
33305 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33308 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33309 if -fsanitize=kernel-address is present.
33311 2017-01-13 Richard Biener <rguenther@suse.de>
33313 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33314 as _Literal ( type ) number in case usual suffixes do not
33315 preserve all information.
33317 2017-01-13 Richard Biener <rguenther@suse.de>
33319 PR tree-optimization/77283
33320 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33321 and ssa-iterators.h.
33322 (is_feasible_trace): Implement a cost model based on joiner
33325 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33328 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33329 char or short to __float128/_Float128 directly.
33331 2017-01-12 Martin Sebor <msebor@redhat.com>
33333 to -Wformat-overflow.
33334 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33335 (min_bytes_remaining): Same.
33336 (get_string_length): Same.
33337 (format_string): Same.
33338 (format_directive): Same.
33340 (pass_sprintf_length::handle_gimple_call): Same.
33342 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33344 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33345 info.nowrite calls with no lhs that can't throw. Return bool
33346 whether gsi_remove has been called or not.
33347 (pass_sprintf_length::handle_gimple_call): Return bool whether
33348 try_substitute_return_value called gsi_remove. Formatting fix.
33349 (pass_sprintf_length::execute): Don't use gsi_remove if
33350 handle_gimple_call returned true.
33353 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33354 be removed due to side-effects, don't remove following barrier nor
33355 turn the successor edge into fallthru edge.
33357 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33360 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33361 element-reversing loads and stores as not swappable.
33363 2017-01-12 Nathan Sidwell <nathan@acm.org>
33364 Nicolai Stange <nicstange@gmail.com>
33366 * combine.c (try_combine): Don't ignore result of overlap checking
33367 loop. Combine overlap & asm check into single loop.
33369 2017-01-12 Richard Biener <rguenther@suse.de>
33371 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33372 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33374 2017-01-12 Richard Biener <rguenther@suse.de>
33376 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33377 and TS_TARGET_OPTION directly derive from TS_BASE.
33378 * tree-core.h (tree_optimization_option): Derive from tree_base.
33379 (tree_target_option): Likewise.
33381 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33383 * config/i386/i386.c (memory_address_length): Increase len
33384 only when rip_relative_addr_p returns false.
33386 2017-01-11 Julia Koval <julia.koval@intel.com>
33388 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33389 (OPTION_MASK_ISA_SGX_SET): New.
33390 (ix86_handle_option): Handle OPT_msgx.
33391 * config.gcc: Added sgxintrin.h.
33392 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33393 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33394 * config/i386/i386.c (ix86_target_string): Add -msgx.
33396 (ix86_option_override_internal): Handle new options.
33397 (ix86_valid_target_attribute_inner_p): Add sgx.
33398 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33399 * config/i386/i386.opt: Add msgx.
33400 * config/i386/sgxintrin.h: New file.
33401 * config/i386/x86intrin.h: Add sgxintrin.h.
33403 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33406 * fold-const.c (maybe_nonzero_address): Return 1 for function
33408 (tree_single_nonzero_warnv_p): Don't handle function local objects
33412 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33415 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33418 * opts.c: Include "spellcheck.h"
33419 (struct string_fragment): New struct.
33420 (struct edit_distance_traits<const string_fragment &>): New
33422 (get_closest_sanitizer_option): New function.
33423 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33425 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33427 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33429 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33430 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33431 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33432 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33433 for initial die_offset if dwarf_split_debug_info.
33434 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33435 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33437 (output_skeleton_debug_sections): Formatting fix. Use
33438 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33439 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33441 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33443 * config/arm/cortex-a53.md: Add bypasses for
33444 cortex_a53_r2f_cvt.
33445 (cortex_a53_r2f): Only use for transfers.
33446 (cortex_a53_f2r): Likewise.
33447 (cortex_a53_r2f_cvt): Add reservation for conversions.
33448 (cortex_a53_f2r_cvt): Likewise.
33450 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33452 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33453 to all inlined functions, change static to extern.
33455 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33458 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33460 (arm_assemble_integer): Likewise.
33462 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33464 * config.gcc: Use new awk script to check CPU, FPU and architecture
33465 parameters for --with-... options.
33466 * config/arm/parsecpu.awk: New file
33467 * config/arm/arm-cpus.in: New file.
33468 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33470 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33472 * config/arm/t-arm: Update dependency rules.
33473 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33474 of processing .def files.
33475 * config/arm/genopt.sh: Deleted.
33476 * config/arm/gentune.sh: Deleted.
33477 * config/arm/arm-cores.def: Deleted.
33478 * config/arm/arm-arches.def: Deleted.
33479 * config/arm/arm-fpus.def: Deleted.
33480 * config/arm/arm-tune.md: Regenerated.
33481 * config/arm/arm-tables.opt: Regenerated.
33482 * config/arm/arm-cpu.h: New generated file.
33483 * config/arm/arm-cpu-data.h: New generated file.
33484 * config/arm/arm-cpu-cdata.h: New generated file.
33486 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33489 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33491 (input_varpool_node): Unpack dynamically_initialized bit.
33493 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33495 PR rtl-optimization/79032
33496 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33497 the alignment of the adjusted memory reference against that of MODE,
33498 instead of the alignment of the original memory reference.
33500 2017-01-11 Martin Jambor <mjambor@suse.cz>
33502 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33504 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33505 decorated functions.
33507 2017-01-11 Richard Biener <rguenther@suse.de>
33509 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33510 set range/nonnull info for PHI results. Do not set it on
33511 stmts marked for removal.
33513 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33515 * expr.c (store_field): In the bitfield case, fetch the return value
33516 from the registers before applying a single big-endian adjustment.
33517 Always do a final load for a BLKmode value not larger than a word.
33519 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33522 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33523 that we correctly handle column numbers greater than
33524 LINE_MAP_MAX_COLUMN_NUMBER.
33526 2017-01-10 Martin Sebor <msebor@redhat.com>
33528 PR middle-end/78245
33529 * gimple-ssa-sprintf.c (get_destination_size): Call
33530 {init,fini}object_sizes.
33531 * tree-object-size.c (addr_object_size): Adjust.
33532 (pass_through_call): Adjust.
33533 (pass_object_sizes::execute): Adjust.
33534 * tree-object-size.h (fini_object_sizes): Declare.
33536 2017-01-10 Martin Sebor <msebor@redhat.com>
33538 PR tree-optimization/78775
33539 * builtins.c (get_size_range): Move...
33540 * calls.c: ...to here.
33541 (alloc_max_size): Accept zero argument.
33542 (operand_signed_p): Remove.
33543 (maybe_warn_alloc_args_overflow): Call get_size_range.
33544 * calls.h (get_size_range): Declare.
33546 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33548 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33549 from TI's devices.csv file as of September 2016.
33550 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33552 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33554 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33555 * doc/invoke.texi: Likewise.
33556 * doc/md.texi: Likewise.
33557 * doc/objc.texi: Likewise.
33559 2017-01-10 Joshua Conner <joshconner@google.com>
33561 * config/arm/fuchsia-elf.h: New file.
33562 * config/fuchsia.h: New file.
33563 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33564 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33566 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33568 2016-01-10 Richard Biener <rguenther@suse.de>
33570 PR tree-optimization/79034
33571 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33572 Propagate out degenerate PHIs in the joiner.
33574 2017-01-10 Martin Liska <mliska@suse.cz>
33576 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33577 (sort_congruence_classes_by_decl_uid): Likewise.
33578 (sort_congruence_class_groups_by_decl_uid): Likewise.
33579 (sem_item_optimizer::merge_classes): Sort class, groups in these
33580 classes and members in the groups by DECL_UID of declarations.
33581 This would make merge operations stable.
33583 2017-01-10 Martin Liska <mliska@suse.cz>
33585 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33586 usage of m_classes_vec.
33587 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33588 (sem_item_optimizer::get_group_by_hash): Likewise.
33589 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33590 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33591 (sem_item_optimizer::verify_classes): Likewise.
33592 (sem_item_optimizer::process_cong_reduction): Likewise.
33593 (sem_item_optimizer::dump_cong_classes): Likewise.
33594 (sem_item_optimizer::merge_classes): Likewise.
33595 * ipa-icf.h (congruence_class_hash): Rename from
33596 congruence_class_group_hash. Remove declaration of m_classes_vec.
33598 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33600 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33601 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33602 * config.gcc: Add avx512vpopcntdqintrin.h.
33603 * config/i386/avx512vpopcntdqintrin.h: New.
33604 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33605 * config/i386/i386-builtin-types.def: Add new types.
33606 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33607 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33608 __builtin_ia32_vpopcountq_v8di_mask): New.
33609 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33610 __AVX512VPOPCNTDQ__.
33611 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33612 (PTA_AVX512VPOPCNTDQ): Define.
33613 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33614 TARGET_AVX512VPOPCNTDQ_P): Define.
33615 * config/i386/i386.opt: Add mavx512vpopcntdq.
33616 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33617 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33619 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33621 PR middle-end/77484
33622 * predict.def (PRED_CALL): Set to 67.
33624 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33626 * expr.c (store_field): In the bitfield case, if the value comes from
33627 a function call and is of an aggregate type returned in registers, do
33628 not modify the field mode; extract the value in all cases if the mode
33629 is BLKmode and the size is not larger than a word.
33631 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33634 * config/i386/cpuid.h: Fix undefined behavior.
33636 2017-01-04 Jeff Law <law@redhat.com>
33638 PR tree-optimization/79007
33639 PR tree-optimization/67955
33640 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33641 conservative for pt.null when flag_non_call_exceptions is on.
33643 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33645 PR translation/79019
33646 PR translation/79020
33647 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33648 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33649 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33651 * config/avr/avr.opt (maccumulate-args): Likewise.
33652 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33653 * common.opt (freport-bug): Likewise.
33654 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33655 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33656 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33657 translatable string.
33658 * config/i386/i386.c (function_value_32): Likewise.
33659 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33660 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33662 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33663 * common/config/msp430/msp430-common.c (msp430_handle_option):
33665 * symtab.c (symtab_node::verify_base): Likewise.
33666 * opts.c (set_debug_level): Likewise.
33667 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33668 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33669 missing whitespace to translatable strings.
33670 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33671 * config/sh/superh.h: Likewise.
33672 * config/i386/xopintrin.h: Likewise.
33673 * config/i386/znver1.md: Likewise.
33674 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33675 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33676 * double-int.h (struct double_int): Likewise.
33677 * double-int.c (div_and_round_double): Likewise.
33678 * wide-int.cc: Likewise.
33679 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33680 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33681 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33682 (crossjumps_occurred): ... this.
33683 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33686 PR tree-optimization/78899
33687 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33688 returning bool return struct loop *, NULL for failure and the new
33690 (versionable_outer_loop_p): Don't version outer loop if it has
33691 dont_vectorized bit set.
33692 (tree_if_conversion): When versioning outer loop, ensure
33693 tree_if_conversion is performed also on the inner loop of the
33694 non-vectorizable outer loop copy.
33695 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33696 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33697 prevent vectorization of it.
33698 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33699 the outer loop vectorization of the non-scalar version is attempted
33700 before vectorization of the inner loop in scalar version. If
33701 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33702 vectorization of its inner loop.
33703 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33704 has 2 inner loops, rename also on edges from bb whose single pred
33705 is outer_loop->header. Fix typo in function comment.
33707 2017-01-09 Martin Sebor <msebor@redhat.com>
33710 * asan.c (asan_emit_stack_protection): Increase local buffer size
33711 to avoid snprintf truncation warning.
33713 2017-01-09 Andrew Pinski <apinski@cavium.com>
33715 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33716 to reference thunderx2t99 for the tuning structure
33717 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33719 (thunderx2t99_extra_costs): This.
33720 * config/aarch64/aarch64-tune.md: Regenerate.
33721 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33722 (vulcan_addrcost_table): This.
33723 (vulcan_regmove_cost): Rename to ...
33724 (thunderx2t99_regmove_cost): This.
33725 (vulcan_vector_cost): Rename to ...
33726 (thunderx2t99_vector_cost): this.
33727 (vulcan_branch_cost): Rename to ...
33728 (thunderx2t99_branch_cost): This.
33729 (vulcan_tunings): Rename to ...
33730 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33731 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33733 2017-01-09 Martin Jambor <mjambor@suse.cz>
33737 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33738 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33739 (propagate_vr_accross_jump_function): Use the above function for all
33740 value range computations for pass-through jump functions and type
33741 converasion from explicit value range values.
33742 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33743 parameters from TYPE_ARG_TYPES.
33744 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33745 (ipa_write_node_info): Stream type of the actual argument.
33746 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33748 2017-01-09 Martin Liska <mliska@suse.cz>
33751 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33752 (lookup_compiler): Do not show error message with have_E.
33754 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33756 PR tree-optimization/78938
33757 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33758 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33759 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33760 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33763 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33765 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33768 2017-01-09 Richard Biener <rguenther@suse.de>
33770 PR tree-optimization/78997
33771 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33772 name condition properly.
33774 2017-01-09 Richard Biener <rguenther@suse.de>
33777 * dwarf2out.c (is_cxx): New overload with context.
33778 (is_naming_typedef_decl): Use it.
33780 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33782 * invoke.texi (Option Summary): Correct spacing in option lists
33783 and add line breaks to fix over-long lines.
33785 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33787 PR middle-end/17660
33789 * extend.texi (Common Variable Attributes): Add xref to GCC
33790 Internals manual to explain mode attribute keywords.
33792 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33795 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33796 and Preprocessor Options.
33797 (Options for Linking): Document -pthread here....
33798 (RS/6000 and PowerPC Options): ...not here.
33799 (Solaris 2 Options): ...or here.
33800 * doc/cppopts.texi: Document -pthread.
33802 2017-01-08 Martin Sebor <msebor@redhat.com>
33804 PR middle-end/77708
33805 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33806 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33807 New member functions.
33808 (format_directive): Used them.
33810 (pass_sprintf_length::handle_gimple_call): Same.
33811 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33812 to avoid truncation for any argument.
33813 (extract_affine_mul): Same.
33814 * tree.c (get_file_function_name): Same.
33816 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33818 PR middle-end/77484
33819 * predict.def (PRED_INDIR_CALL): Set to 86.
33821 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33823 PR preprocessor/54124
33824 * doc/cppopts.texi: Reformat -d subtable to list the full name
33825 of the options. Add cross-reference to the docs for the general
33826 compiler -d options.
33827 * doc/invoke.texi (Developer Options): Add cross-reference to the
33828 preprocessor-specific -d option documentation.
33830 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33832 PR preprocessor/13498
33833 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33834 redudant material, and reflect new command-line options.
33835 (System Headers): Likewise.
33837 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33839 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33840 -isystem, and -idirafter. Copy-edit.
33841 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33842 default for -ftrack-macro-expansion. Delete obsolete and
33843 badly-formatted implementation details about -fdebug-cpp output.
33844 * doc/cppwarnopts.texi: Copy-edit.
33846 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33849 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33850 that the transition from a max line width >= 1<<10 to narrower
33851 lines works correctly.
33853 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33855 * doc/options.texi (PerFunction): New.
33856 * opt-functions.awk (switch_flags): Map both Optimization and
33857 PerFunction to CL_OPTIMIZATION.
33858 * opth-gen.awk: Test for PerFunction flag along with
33860 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33861 it only when the latter is present. Skip those that don't in
33862 the hash function generator.
33863 * common.opt (fvar-tracking): Mark as PerFunction instead of
33865 (fvar-tracking-assignments): Likewise.
33866 (fvar-tracking-assignments-toggle): Likewise.
33867 (fvar-tracking-uninit): Likewise.
33869 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33871 PR translation/79018
33872 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33875 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33878 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33879 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33880 TARGET_LONG_JUMP_TABLE_OFFSETS.
33881 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33882 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33883 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33884 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33885 * config/m68k/m68k.md (tablejump expander): Likewise.
33886 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33887 TARGET_LONG_JUMP_TABLE_OFFSETS.
33888 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33889 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33891 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33892 David Holsgrove <david.holsgrove@xilinx.com>
33894 * common/config/microblaze/microblaze-common.c
33895 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33896 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33898 * config/microblaze/microblaze.c (microblaze_must_save_register)
33899 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33901 (microblaze_eh_return): New function.
33902 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33903 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33904 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33905 * config/microblaze/microblaze.md (eh_return): New pattern.
33907 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33909 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33910 GCC_DIAGNOSTIC_STRINGIFY): Define.
33912 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33914 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33916 * config/arm/arm.md (<mcrr>): New.
33918 * config/arm/arm.c (arm_arch5te): New.
33919 (arm_option_override): Set arm_arch5te.
33920 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33922 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33923 (arm_mcrr_qualifiers): ... this. New.
33924 (MRRC_QUALIFIERS): Define to...
33925 (arm_mrrc_qualifiers): ... this. New.
33926 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33928 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33929 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33930 (MRRCI, mrrc, MRRC): New.
33931 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33932 VUNSPEC_MRRC2): New.
33934 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33936 * config/arm/arm.md (<mcr>): New.
33938 * config/arm/arm.c (arm_coproc_builtin_available): Add
33939 support for mcr, mrc, mcr2 and mrc2.
33940 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33941 (arm_mcr_qualifiers): ... this. New.
33942 (MRC_QUALIFIERS): Define to ...
33943 (arm_mrc_qualifiers): ... this. New.
33944 (MCR_QUALIFIERS): Define to ...
33945 (arm_mcr_qualifiers): ... this. New.
33946 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33948 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33949 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33950 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33951 VUNSPEC_MRC2): New.
33953 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33955 * config/arm/arm.md (*ldc): New.
33959 * config/arm/arm.c (arm_coproc_builtin_available): Add
33960 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33961 (arm_coproc_ldc_stc_legitimate_address): New.
33962 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33963 'qualifier_const_pointer'.
33964 (LDC_QUALIFIERS): Define to...
33965 (arm_ldc_qualifiers): ... this. New.
33966 (STC_QUALIFIERS): Define to...
33967 (arm_stc_qualifiers): ... this. New.
33968 * config/arm/arm-protos.h
33969 (arm_coproc_ldc_stc_legitimate_address): New.
33970 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33971 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33972 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33973 stc2, stcl, stc2l): New.
33974 * config/arm/constraints.md (Uz): New.
33975 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33976 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33977 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33978 VUNSPEC_STC2L): New.
33980 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33982 * config/arm/arm.md (<cdp>): New.
33983 * config/arm/arm.c (neon_const_bounds): Rename this ...
33984 (arm_const_bounds): ... this.
33985 (arm_coproc_builtin_available): New.
33986 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33987 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33988 (CDP_QUALIFIERS): Define to...
33989 (arm_cdp_qualifiers): ... this. New.
33991 (arm_expand_builtin_args): Add case for 6 arguments.
33992 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33993 (arm_const_bounds): ... this.
33994 (arm_coproc_builtin_available): New.
33995 * config/arm/arm_acle.h (__arm_cdp): New.
33997 * config/arm/arm_acle_builtins.def (cdp): New.
33999 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34000 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34001 'arm_const_bounds'.
34002 * config/arm/types.md (coproc): New.
34003 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34004 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34005 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34006 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34008 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34010 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34011 (UBINOP_QUALIFIERS): New.
34013 (acle_builtin_data): New. Change comment.
34014 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34015 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34016 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34017 arm_acle_builtins.def.
34018 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34019 (arm_init_acle_builtins): New.
34020 (CRC32_BUILTIN): Remove.
34021 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34022 crc32cb, crc32ch and crc32cw.
34023 (arm_init_crc32_builtins): Remove.
34024 (arm_init_builtins): Use arm_init_acle_builtins rather
34025 than arm_init_crc32_builtins.
34026 (arm_expand_acle_builtin): New.
34027 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34028 * config/arm/arm_acle_builtins.def: New.
34030 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34032 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34033 (arm_builtin_datum): ... this.
34034 (arm_init_neon_builtin): Rename to ...
34035 (arm_init_builtin): ... this. Add a new parameters PREFIX
34036 and USE_SIG_IN_NAME.
34037 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34038 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34039 'arm_builtin_datum'.
34040 (arm_init_vfp_builtins): Likewise.
34041 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34042 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34043 (arm_expand_neon_args): Rename to ...
34044 (arm_expand_builtin_args): ... this. Rename builtin_arg
34045 enum values and differentiate between ARG_BUILTIN_MEMORY
34046 and ARG_BUILTIN_NEON_MEMORY.
34047 (arm_expand_neon_builtin_1): Rename to ...
34048 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34049 values, arm_expand_builtin_args and add bool parameter NEON.
34050 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34051 (arm_expand_vfp_builtin): Likewise.
34052 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34054 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34056 PR middle-end/77484
34057 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34058 * predict.c (tree_estimate_probability_bb): Reverse direction of
34059 polymorphic call predictor.
34061 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34063 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34064 (determine_pass_name_match): ...this new function and...
34065 (should_skip_pass_p): ...this new function.
34067 2017-01-06 Nathan Sidwell <nathan@acm.org>
34069 * ipa-visibility.c (function_and_variable_visibility): Reformat
34070 comments and long lines. Remove extrneous if.
34071 * symtab.c (symtab_node::make_decl_local): Fix code format.
34072 (symtab_node::set_section_for_node): Fix comment typo.
34074 2017-01-06 Martin Liska <mliska@suse.cz>
34077 * lra-constraints.c: Rename invariant to lra_invariant.
34078 * predict.c (set_even_probabilities): Initialize e to NULL.
34080 2017-01-05 Martin Sebor <msebor@redhat.com>
34082 PR tree-optimization/78910
34083 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34084 (format_integer): Correct off-by-one error in the handling
34085 of precision with negative numbers in signed conversions..
34087 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34089 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34091 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34093 PR tree-optimization/71016
34094 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34095 factor_out_conditional_conversion. Formatting fix.
34096 (factor_out_conditional_conversion): Add cond_stmt argument.
34097 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34098 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34101 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34103 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34104 read-rtl-function.o, and selftest-rtl.o.
34105 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34106 (selftest::aarch64_test_loading_full_dump): New function.
34107 (selftest::aarch64_run_selftests): New function.
34108 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34109 selftest::aarch64_run_selftests.
34110 * config/i386/i386.c
34111 (selftest::ix86_test_loading_dump_fragment_1): New function.
34112 (selftest::ix86_test_loading_call_insn): New function.
34113 (selftest::ix86_test_loading_full_dump): New function.
34114 (selftest::ix86_test_loading_unspec): New function.
34115 (selftest::ix86_run_selftests): Call the new functions.
34116 * emit-rtl.c (maybe_set_max_label_num): New function.
34117 * emit-rtl.h (maybe_set_max_label_num): New decl.
34118 * function.c (instantiate_decls): Guard call to
34119 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34120 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34122 * gensupport.c (gen_reader::gen_reader): Pass "false"
34123 for new "compact" param of rtx_reader.
34124 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34125 rather than an empty string for NULL strings.
34126 * read-md.c: Potentially include config.h rather than bconfig.h.
34127 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34128 (have_error): New global, copied from errors.c.
34129 (md_reader::read_name): Rename to...
34130 (md_reader::read_name_1): ...this, adding "out_loc" param,
34131 and converting "missing name or number" to returning false, rather
34133 (md_reader::read_name): Reimplement in terms of read_name_1.
34134 (md_reader::read_name_or_nil): New function.
34135 (md_reader::read_string): Handle "(nil)" by returning NULL.
34136 (md_reader::md_reader): Add new param "compact".
34137 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34138 (md_reader::read_file): New method.
34139 * read-md.h (md_reader::md_reader): Add new param "compact".
34140 (md_reader::read_file): New method.
34141 (md_reader::is_compact): New accessor.
34142 (md_reader::read_name): Convert return type from void to file_location.
34143 (md_reader::read_name_or_nil): New decl.
34144 (md_reader::read_name_1): New decl.
34145 (md_reader::m_compact): New field.
34146 (noop_reader::noop_reader): Pass "false" for new "compact" param
34148 (rtx_reader::rtx_reader): Add new "compact" param.
34149 (rtx_reader::read_rtx_operand): Make virtual and convert return
34150 type from void to rtx.
34151 (rtx_reader::read_until): New decl.
34152 (rtx_reader::handle_any_trailing_information): New virtual function.
34153 (rtx_reader::postprocess): New virtual function.
34154 (rtx_reader::finalize_string): New virtual function.
34155 (rtx_reader::m_in_call_function_usage): New field.
34156 (rtx_reader::m_reuse_rtx_by_id): New field.
34157 * read-rtl-function.c: New file.
34158 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34159 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34160 (selftest::verify_three_block_rtl_cfg): New decl.
34161 * read-rtl-function.h: New file.
34162 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34163 For host, include function.h, memmodel.h, and emit-rtl.h.
34164 (one_time_initialization): New function.
34165 (struct compact_insn_name): New struct.
34166 (compact_insn_names): New array.
34167 (find_code): Handle insn codes in compact dumps.
34168 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34169 (bind_subst_iter_and_attr): Likewise.
34170 (add_condition_to_string): Likewise.
34171 (add_condition_to_rtx): Likewise.
34172 (apply_attribute_uses): Likewise.
34173 (add_current_iterators): Likewise.
34174 (apply_iterators): Likewise.
34175 (initialize_iterators): Guard usage of apply_subst_iterator with
34176 #ifdef GENERATOR_FILE.
34177 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34178 (md_reader::read_mapping): Likewise.
34179 (add_define_attr_for_define_subst): Likewise.
34180 (add_define_subst_attr): Likewise.
34181 (read_subst_mapping): Likewise.
34182 (check_code_iterator): Likewise.
34183 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34185 (one_time_initialization): New function.
34186 (rtx_reader::read_until): New method.
34187 (read_flags): New function.
34188 (parse_reg_note_name): New function.
34189 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34190 Handle reuse_rtx ids.
34191 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34192 Add parsing support for RTL dumps, mirroring the special-cases in
34193 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34194 values, and calling handle_any_trailing_information.
34195 (rtx_reader::read_rtx_operand): Convert return type from void
34196 to rtx, returning return_rtx. Handle case 'e'. Call
34197 finalize_string on XSTR and XTMPL fields.
34198 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34199 "(nil)" values were omitted. Call the postprocess vfunc on the
34201 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34202 class ctor. Initialize m_in_call_function_usage. Call
34203 one_time_initialization.
34204 * rtl-tests.c (selftest::test_uncond_jump): Call
34205 set_new_first_and_last_insn.
34206 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34207 * selftest-rtl.c: New file.
34208 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34209 (selftest::get_insn_by_uid): New decl.
34210 * selftest-run-tests.c (selftest::run_tests): Call
34211 read_rtl_function_c_tests.
34212 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34213 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34216 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34218 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34219 operands in a special way. Assert that pos+len <= mode precision.
34221 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34223 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34224 3 argument Alias with unlimited for the negative form.
34225 (fno-vect-cost-model): Removed.
34227 2017-01-05 Martin Liska <mliska@suse.cz>
34229 * hsa-gen.c (gen_hsa_divmod): New function.
34230 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34232 2017-01-05 Martin Liska <mliska@suse.cz>
34235 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34238 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34240 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34241 small constant length operands.
34243 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34245 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34246 between loop iterations.
34248 2017-01-05 Martin Liska <mliska@suse.cz>
34251 * gimplify.c (gimplify_decl_expr): Compare to
34252 asan_poisoned_variables instread of checking flags.
34253 (gimplify_target_expr): Likewise.
34254 (gimplify_expr): Likewise.
34255 (gimplify_function_tree): Conditionally initialize
34256 asan_poisoned_variables.
34258 2017-01-04 Jeff Law <law@redhat.com>
34260 PR tree-optimizatin/78812
34261 * rtl.h (contains_mem_rtx_p): Prototype.
34262 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34263 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34264 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34265 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34266 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34268 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34270 * input.c (assert_char_at_range): Default-initialize actual_range.
34272 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34274 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34275 to match the caller.
34277 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34279 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34280 insns after final jump in test to emit dummy move.
34282 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34284 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34285 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34287 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34289 * multiple_target.c (create_dispatcher_calls): Init e_next.
34290 * tree-ssa-loop-split.c (split_loop): Init border.
34291 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34294 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34299 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34300 (altivec_register_operand): Do not return true if the operand
34301 contains a SUBREG mixing SImode and SFmode.
34302 (vsx_register_operand): Likewise.
34303 (vsx_reg_sfsubreg_ok): New predicate.
34304 (vfloat_operand): Do not return true if the operand contains a
34305 SUBREG mixing SImode and SFmode.
34306 (vint_operand): Likewise.
34307 (vlogical_operand): Likewise.
34308 (gpc_reg_operand): Likewise.
34309 (int_reg_operand): Likewise.
34310 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34311 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34312 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34314 (rs6000_emit_move_si_sf_subreg): New helper function.
34315 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34316 fixup SUBREGs involving SImode and SFmode.
34317 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34318 numbers for the new peephole2 optimization.
34319 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34320 the GLIBC math library that do AND/IOR/XOR operations on single
34321 precision floating point.
34322 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34323 target macros to say whether we need to avoid SUBREGs mixing
34325 (TARGET_ALLOW_SF_SUBREG): Likewise.
34326 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34327 (UNSPEC_SI_FROM_SF): Likewise.
34328 (iorxor): Change spacing.
34329 (and_ior_xor): New iterator for AND, IOR, and XOR.
34330 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34331 (movdi_from_sf_zero_ext): Likewise.
34332 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34333 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34334 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34335 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34336 (fms<mode>4): Likewise.
34337 (fnma<mode>4): Likewise.
34338 (fnms<mode>4): Likewise.
34339 (nfma<mode>4): Likewise.
34340 (nfms<mode>4): Likewise.
34342 2017-01-04 Marek Polacek <polacek@redhat.com>
34345 * doc/invoke.texi: Document -Wpointer-compare.
34347 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34349 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34352 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34353 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34356 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34359 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34360 documentation of the powerpc_popcntb_ok attribute.
34361 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34362 code to issue warning messages if a requested CPU configuration is
34363 not supported by the binary (assembler and loader) toolchain.
34364 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34365 made to define a built-in function that has been disabled.
34366 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34367 made to define a built-in function that has been disabled.
34368 (altivec_init_builtins): Add comment explaining why definition
34369 of the DST built-in functions is not preceded by an assertion
34370 check. Add assertions to prevent ICE if attempts are made to
34371 define an altivec predicate or an abs* built-in function that has
34373 (htm_init_builtins): Add comment explaining why definition of the
34374 htm built-in functions is not preceded by an assertion check.
34376 2017-01-04 Jeff Law <law@redhat.com>
34378 PR tree-optimizatin/67955
34379 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34380 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34381 the points-to solution does not include pt_null. Use DECL_PT_UID
34384 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34386 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34387 Use gen_int_mode instead of gen_lopwart for const_int operands.
34389 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34391 PR tree-optimization/71563
34392 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34393 out of range value - has low bits known to be zero.
34395 2017-01-04 Alan Modra <amodra@gmail.com>
34397 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34398 * configure: Regenerate.
34399 * config.in: Regenerate.
34401 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34404 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34405 a substring of the message, but strcmp with the whole message. Ifdef
34406 ENABLE_NLS, translate the message first using dgettext.
34408 2017-01-03 Jeff Law <law@redhat.com>
34410 PR tree-optimizatin/78856
34411 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34412 (mark_threaded_blocks): Remove code to truncate thread paths that
34413 cross multiple loop headers. Instead invalidate the cached loop
34414 iteration information and handle case of a thread path walking
34415 into an irreducible region.
34417 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34420 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34421 assertions. Add support for doing the signbit if the IEEE 128-bit
34422 floating point value is in a GPR.
34423 * config/rs6000/rs6000.md (Fsignbit): Delete.
34424 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34425 Update the length attribute if the value is in a GPR.
34426 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34427 the sign or zero extension instruction, since the value is always 0/1.
34428 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34431 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34432 extracting SImode to a GPR register so that we can generate a
34433 store, limit the vector to be in a traditional Altivec register
34434 for the vextuwrx instruction.
34436 2017-01-03 Ian Lance Taylor <iant@google.com>
34438 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34440 2017-01-03 Martin Sebor <msebor@redhat.com>
34442 PR tree-optimization/78696
34443 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34444 precision. Use MPFR for %f for greater fidelity. Correct handling
34446 (pass_sprintf_length::compute_format_length): Set width and precision
34447 specified by asrerisk to void_node for vararg functions.
34448 (try_substitute_return_value): Adjust dump output.
34450 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34452 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34454 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34456 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34457 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34458 -mlra/-mno-lra was passed to the compiler.
34460 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34462 PR rtl-optimization/65618
34463 * emit-rtl.c (try_split): Move initialization of "before" and
34464 "after" to just before the call to emit_insn_after_setloc.
34466 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34468 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34470 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34472 * dwarf2out.c (gen_enumeration_type_die): When
34473 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34475 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34477 PR tree-optimization/78965
34478 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34479 Change first argument from const call_info & to call_info &. For %n
34480 set info.nowrite to false.
34482 PR middle-end/78901
34483 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34484 possibly throwing calls.
34486 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34487 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34488 and fns handling, rather than in a separate case SSA_NAME.
34490 2017-01-02 Jeff Law <law@redhat.com>
34492 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34493 fixes for first_period and second_period variables.
34495 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34498 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34499 (*insvqi_1): New insn pattern.
34500 (*insvqi_1_mem_rex64): Ditto.
34501 (*insvqi_2): Ditto.
34502 (*insvqi_3): Rename from *insvqi.
34504 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34506 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34508 * doc/cfg.texi (Edges): Remove reference to Java.
34509 (Maintaining the CFG): Ditto.
34511 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34513 PR middle-end/77674
34514 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34515 transparent aliases.
34517 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34519 PR middle-end/77484
34520 * predict.def (PRED_CALL): Update hitrate.
34521 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34522 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34523 into direct/indirect/polymorphic variants.
34525 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34527 Update copyright years.
34529 * gcc.c (process_command): Update copyright notice dates.
34530 * gcov-dump.c (print_version): Ditto.
34531 * gcov.c (print_version): Ditto.
34532 * gcov-tool.c (print_version): Ditto.
34533 * gengtype.c (create_file): Ditto.
34534 * doc/cpp.texi: Bump @copying's copyright year.
34535 * doc/cppinternals.texi: Ditto.
34536 * doc/gcc.texi: Ditto.
34537 * doc/gccint.texi: Ditto.
34538 * doc/gcov.texi: Ditto.
34539 * doc/install.texi: Ditto.
34540 * doc/invoke.texi: Ditto.
34542 Copyright (C) 2017 Free Software Foundation, Inc.
34544 Copying and distribution of this file, with or without modification,
34545 are permitted in any medium without royalty provided the copyright
34546 notice and this notice are preserved.