1 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
3 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
8 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
9 enable generating XXBRH if the value is in a vector register.
10 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
11 value is in a vector register.
12 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
13 register to register bswap64's instead of doing the GPR sequence
14 used on previous machines.
15 (bswapdi2_xxbrd): New insn.
16 (bswapdi2_reg): Disallow on ISA 3.0.
17 (register to register bswap64 splitter): Do not split the insn on
18 ISA 3.0 systems that use XXBRD.
20 2017-11-10 Martin Sebor <msebor@redhat.com>
23 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
25 (find_subframework_file): Same.
27 2017-11-10 Jeff Law <law@redhat.com>
29 * vr-values.h (VR_INITIALIZER): Move #define here.
30 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
31 * Makefile.in (OBJS): Add tree-evrp.o
32 * tree-vrp.h (assert_info): Move structure definition here.
33 (set_value_range_to_varying): Prototype.
34 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
35 (infer_value_range, register_edge_assert_for): Likewise.
36 (stmt_interesting_for_vrp): Likewise.
37 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
38 (set_value_range_to_varying): No longer static.
39 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
40 (infer_value_range, register_edge_assert_for): Likewise.
42 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
44 * auto-profile.c (afdo_indirect_call): Drop frequency.
45 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
46 (cgraph_node::create_edge): Drop frequency argument.
47 (cgraph_node::create_indirect_edge): Drop frequency argument.
48 (cgraph_edge::make_speculative): Drop frequency arguments.
49 (cgraph_edge::resolve_speculation): Do not update frequencies
50 (cgraph_edge::dump_edge_flags): Do not dump frequency.
51 (cgraph_node::dump): Check consistency in IPA mode.
52 (cgraph_edge::maybe_hot_p): Use IPA counter.
53 (cgraph_edge::verify_count_and_frequency): Rename to ...
54 (cgraph_edge::verify_count): ... this one; drop frequency checking.
55 (cgraph_node::verify_node): Update.
56 * cgraph.h (struct cgraph_edge): Drop frequency.
57 (cgraph_edge::frequency): New function.
58 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
60 (cgraph_edge::rebuild_edges): Likewise.
61 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
62 (duplicate_thunk_for_node): Do not pass frequency.
63 (cgraph_node::create_clone): Scale only counts.
64 (cgraph_node::create_virtual_clone): Do not pass frequency.
65 (cgraph_node::create_edge_including_clones): Do not pass frequency.
66 (cgraph_node::create_version_clone): Do not pass frequency.
67 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
68 (cgraph_node::expand_thunk): Do not pass frequency.
69 (cgraph_node::create_wrapper): Do not pass frequency.
70 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
72 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
73 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
74 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
75 (gather_caller_stats): Use frequency function.
76 (ipcp_cloning_candidate_p): Use frequency function.
77 (ipcp_propagate_stage): Use frequency function.
78 (get_info_about_necessary_edges): Use frequency function.
79 (update_profiling_info): Update only IPA profile.
80 (update_specialized_profile): Use frequency functoin.
81 (perhaps_add_new_callers): Update only IPA profile.
82 * ipa-devirt.c (ipa_devirt): Use IPA profile.
83 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
84 (dump_ipa_call_summary): Use frequency function.
85 (estimate_edge_size_and_time): Use frequency function.
86 (ipa_merge_fn_summary_after_inlining): Use frequency function.
87 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
88 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
89 (update_noncloned_counts): ... ths one; scale counts only.
90 (clone_inlined_nodes): Do not scale frequency.
91 (inline_call): Do not pass frequency.
92 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
93 (compute_inlined_call_time): Use IPA profile.
94 (want_inline_small_function_p): Use IPA profile.
95 (want_inline_self_recursive_call_p): Use IPA profile.
96 (edge_badness): Use IPA profile.
97 (lookup_recursive_calls): Use IPA profile.
98 (recursive_inlining): Do not pass frequency.
99 (resolve_noninline_speculation): Do not update frequency.
100 (inline_small_functions): Collect max of IPA profile.
101 (dump_overall_stats): Dump IPA porfile.
102 (dump_inline_stats): Dump IPA porfile.
103 (ipa_inline): Collect IPA stats.
104 * ipa-inline.h (clone_inlined_nodes): Update prototype.
105 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
106 (ipa_propagate_frequency): Use frequency function.
107 (ipa_profile): Cleanup.
108 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
109 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
110 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
111 (input_node): Do not stream frequency.
112 (input_edge): Do not stream frequency.
113 (merge_profile_summaries): Scale only IPA profiles.
114 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
115 * predict.c (drop_profile): Do not recompute frequency.
116 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
117 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
118 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
119 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
120 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
121 * tree-inline.c (copy_bb): Do not scale frequency.
122 (expand_call_inline): Do not scale frequency.
123 (tree_function_versioning): Do not scale frequency.
124 * ubsan.c (ubsan_create_edge): Do not pass frequency.
126 2017-11-10 Julia Koval <julia.koval@intel.com>
128 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8,
129 _mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8,
130 _mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8,
131 _mm256_maskz_gf2p8affine_epi64_epi8,
132 _mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8
133 _mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
134 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi,
135 __builtin_ia32_vgf2p8affineqb_v32qi,
136 __builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
137 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
139 2017-11-10 Tamar Christina <tamar.christina@arm.com>
143 (arm_option_override): Refactor.
144 (arm_option_reconfigure_globals): New.
145 (arm_options_perform_arch_sanity_checks): New.
146 * config/arm/arm-protos.h (arm_option_reconfigure_globals): New prototype.
147 (arm_options_perform_arch_sanity_checks): Likewise
149 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
151 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
153 * rs6000/rs6000.md ("type" attr): Add qmul.
154 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
155 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
156 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
158 2017-11-10 Martin Sebor <msebor@redhat.com>
161 * builtins.c (compute_objsize): Handle arrays that
162 compute_builtin_object_size likes to fail for. Make extern.
163 * builtins.h (compute_objsize): Declare.
164 (check_strncpy_sizes): New function.
165 (expand_builtin_strncpy): Call check_strncpy_sizes.
166 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
167 -Wstringop-truncation.
168 (gimple_fold_builtin_strncat): Same.
169 * gimple.c (gimple_build_call_from_tree): Set call location.
170 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
171 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
172 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
173 (handle_builtin_strlen): Use strlen_to_stridx.
174 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
176 Use strlen_to_stridx.
177 (pass_strlen::execute): Release strlen_to_stridx.
178 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
179 (-Wstringop-truncation): Document new option.
181 2017-11-10 Martin Liska <mliska@suse.cz>
183 PR gcov-profile/82702
184 * gcov.c (main): Handle intermediate files in a different
186 (get_gcov_intermediate_filename): New function.
187 (output_gcov_file): Remove support of intermediate files.
188 (generate_results): Allocate intermediate file.
189 (release_structures): Clean-up properly fn_end.
190 (output_intermediate_file): Start iterating with line 1.
192 2017-11-10 Jakub Jelinek <jakub@redhat.com>
194 PR tree-optimization/82929
195 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
196 ops_swapped_p non-static data member.
197 (store_immediate_info::store_immediate_info): Clear it.
198 (imm_store_chain_info::coalesce_immediate_stores): If swapping
199 ops set ops_swapped_p.
200 (count_multiple_uses): Handle ops_swapped_p.
202 2017-11-10 Martin Liska <mliska@suse.cz>
204 * coverage.c (coverage_init): Stream information about
205 support of has_unexecuted_blocks.
206 * doc/gcov.texi: Document that.
207 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
208 * gcov.c (read_graph_file): Likewise.
209 (output_line_beginning): Fix a small issue with
212 2017-11-10 Bin Cheng <bin.cheng@arm.com>
214 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
215 reference of trivial component.
217 2017-11-10 Jakub Jelinek <jakub@redhat.com>
220 * gimple-ssa-store-merging.c
221 (pass_store_merging::terminate_all_aliasing_chains): For
222 gimple_store_p stmts also call refs_output_dependent_p.
224 PR rtl-optimization/82913
225 * compare-elim.c (try_merge_compare): Punt if def_insn is not
228 2017-11-09 Jeff Law <law@redhat.com>
230 * vr-values.h: New file with vr_values class.
231 * tree-vrp.c: Include vr-values.h
232 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
233 data objects into the vr_values class.
234 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
235 (get_value_range): Make it a member function within vr_values class.
236 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
237 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
238 (op_with_constant_singleton_value_range): Likewise.
239 (extract_range_for_var_from_comparison_expr): Likewise.
240 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
241 (extract_range_from_binary_expr): Likewise.
242 (extract_range_from_unary_expr): Likewise.
243 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
244 (check_for_binary_op_overflow, extract_range_basic): Likewise.
245 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
246 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
247 (compare_name_with_value, compare_names): Likewise.
248 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
249 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
250 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
251 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
252 (extract_range_from_phi_node): Likewise.
253 (simplify_truth_ops_using_ranges): Likewise.
254 (simplify_div_or_mod_using_ranges): Likewise.
255 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
256 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
257 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
258 (simplify_float_conversion_using_ranges): Likewise.
259 (simplify_internal_call_using_ranges): Likewise.
260 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
261 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
262 poitner via x_vr_values for calls into gimple folder.
263 (vrp_initialize_lattice): Make this the vr_values ctor.
264 (vrp_free_lattice): Make this the vr_values dtor.
265 (set_vr_value): New function.
266 (class vrp_prop): Add vr_values data member. Add various member
267 functions as well as member functions that delegate to vr_values.
268 (check_array_ref): Make a member function within vrp_prop class.
269 (search_for_addr_array, vrp_initialize): Likewise.
270 (vrp_finalize): Likewise. Revamp to avoid direct access to
271 vr_value, values_propagated, etc.
272 (check_array_bounds): Extract vrp_prop class instance pointer from
273 walk info structure. Use it to call member functions.
274 (check_all_array_refs): Make a member function within vrp_prop class.
275 Smuggle class instance pointer via walk info structure.
276 (x_vr_values): New local static.
277 (vrp_valueize): Use x_vr_values to get class instance.
278 (vr_valueize_1): Likewise.
279 (class vrp_folder): Add vr_values data member. Add various member
280 functions as well as member functions that delegate to vr_values.
281 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
282 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
283 class instance from vr_values. Use it to call member functions.
284 (vrp_dom_walker): Add vr_values data member.
285 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
286 instance via x_vr_values.
287 (identify_jump_threads): Accept vr_values as argument. Store
288 it into the walker structure.
289 (evrp_dom_walker): Add vr_values class data member. Add various
291 (evrp_dom_walker::try_find_new_range): Use vr_values data
292 member to access the memory allocator.
293 (evrp_dom_walker::before_dom_children): Store vr_values class
294 instance into the vrp_folder class.
295 (evrp_dom_walker::push_value_range): Rework to avoid direct
296 access to num_vr_values and vr_value.
297 (evrp_dom_walker::pop_value_range): Likewise.
298 (execute_early_vrp): Remove call to vrp_initialize_lattice.
299 Use vr_values to get to dump_all_value_ranges member function.
300 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
301 and simplify_cond_using_ranges_2 via vrp_prop class instance.
302 Pass vr_values class instance down to identify_jump_threads.
303 Remove call to vrp_free_lattice.
304 (debug_all_value_ranges): Remove.
306 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
307 (vrp_folder): Likewise.
309 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
310 Get it from the existing bitmap instead.
311 (vrp_intersect_ranges_1): Likewise.
313 2017-11-09 Jakub Jelinek <jakub@redhat.com>
315 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
317 (store_immediate_info::store_immediate_info): Add bitnotp argument,
319 (imm_store_chain_info::coalesce_immediate_stores): Break group
320 if bit_not_p is different.
321 (count_multiple_uses, split_group,
322 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
323 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
324 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
325 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
326 to store_immediate_info ctor.
328 2017-11-09 Jim Wilson <jimw@sifive.com>
330 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
331 (scan_prog_file): Likewise.
333 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
335 * bb-reorder.c (max_entry_frequency): Remove.
336 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
337 connect_traces, push_to_next_round_p): Remove prototypes.
338 (find_traces_1_round): Use counts only.
339 (push_to_next_round_p): Likewise.
340 (find_traces): Likewise.
341 (rotate_loop): Likewise.
342 (find_traces_1_round): Likewise.
343 (connect_traces): Likewise.
344 (edge_order): Likewise.
346 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
348 * config/arm/arm.c (output_return_instruction): Add comments to
349 indicate requirement for cmse_nonsecure_entry return to account
350 for the size of clearing instruction output here.
351 (thumb_exit): Likewise.
352 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
353 return in hardfloat mode.
355 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
357 * config/rs6000/rs6000.c (machine_function): Add a bool,
358 "toc_is_wrapped_separately".
359 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
360 if it wasn't explicitly set or unset, we are optimizing for speed, and
361 doing separate shrink-wrapping.
362 (rs6000_get_separate_components): Enable the TOC component if
363 saving the TOC register in the prologue.
364 (rs6000_components_for_bb): Handle the TOC component.
365 (rs6000_emit_prologue_components): Store the TOC register where needed.
366 (rs6000_set_handled_components): Mark TOC as handled, if handled.
367 (rs6000_emit_prologue): Don't save the TOC if that is already done.
369 2017-11-09 Martin Jambor <mjambor@suse.cz>
371 * ipa-param-manipulation.c: New file.
372 * ipa-param-manipulation.h: Likewise.
373 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
374 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
375 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
376 (ipa_parm_adjustment): Likewise.
377 (ipa_parm_adjustment_vec): Likewise.
378 (ipa_get_vector_of_formal_parms): Moved declaration to
379 ipa-param-manipulation.h.
380 (ipa_get_vector_of_formal_parm_types): Likewise.
381 (ipa_modify_formal_parameters): Likewise.
382 (ipa_modify_call_arguments): Likewise.
383 (ipa_combine_adjustments): Likewise.
384 (ipa_dump_param_adjustments): Likewise.
385 (ipa_modify_expr): Likewise.
386 (ipa_get_adjustment_candidate): Likewise.
387 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
388 ipa-param-manipulation.c.
389 (ipa_get_vector_of_formal_parm_types): Likewise.
390 (ipa_modify_formal_parameters): Likewise.
391 (ipa_modify_call_arguments): Likewise.
392 (ipa_modify_expr): Likewise.
393 (get_ssa_base_param): Likewise.
394 (ipa_get_adjustment_candidate): Likewise.
395 (index_in_adjustments_multiple_times_p): Likewise.
396 (ipa_combine_adjustments): Likewise.
397 (ipa_dump_param_adjustments): Likewise.
398 * tree-sra.c: Also include ipa-param-manipulation.h
399 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
402 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
403 Alan Hayward <alan.hayward@arm.com>
404 David Sherwood <david.sherwood@arm.com>
406 * doc/sourcebuild.texi (vect_masked_store): Document.
408 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
409 Alan Hayward <alan.hayward@arm.com>
410 David Sherwood <david.sherwood@arm.com>
412 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
414 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
415 Alan Hayward <alan.hayward@arm.com>
416 David Sherwood <david.sherwood@arm.com>
418 * doc/sourcebuild.texi (vect_variable_length): Document.
420 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
421 Alan Hayward <alan.hayward@arm.com>
422 David Sherwood <david.sherwood@arm.com>
424 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
426 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
427 Alan Hayward <alan.hayward@arm.com>
428 David Sherwood <david.sherwood@arm.com>
430 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
432 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
433 Alan Hayward <alan.hayward@arm.com>
434 David Sherwood <david.sherwood@arm.com>
436 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
437 previously undocumented selectors.
438 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
440 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
442 * doc/rtl.texi (const_vector): Say that elements can be
444 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
445 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
446 (gen_vec_duplicate): Use it instead of CONSTANT_P.
447 * optabs.c (expand_vector_broadcast): Likewise.
449 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
450 Alan Hayward <alan.hayward@arm.com>
451 David Sherwood <david.sherwood@arm.com>
453 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
454 scaled index even if the unscaled address was invalid.
455 Don't increase the complexity of using a scale in that case.
457 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
458 Alan Hayward <alan.hayward@arm.com>
459 David Sherwood <david.sherwood@arm.com>
461 * doc/rtl.texi: Rewrite the subreg rules so that they partition
462 the inner register into REGMODE_NATURAL_SIZE bytes rather than
463 UNITS_PER_WORD bytes.
464 * emit-rtl.c (validate_subreg): Divide subregs into blocks
465 based on REGMODE_NATURAL_SIZE of the inner mode.
466 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
467 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
468 * expmed.c (lowpart_bit_field_p): Divide the value up into
469 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
470 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
471 whether something is likely to occupy more than one register.
473 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
476 * ipa-inline-transform.c (update_noncloned_frequencies): Use
477 profile_count::adjust_for_ipa_scaling.
478 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
479 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
481 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
483 2017-11-09 Jakub Jelinek <jakub@redhat.com>
485 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
486 (split_group): Add total_orig and total_new arguments, estimate the
487 number of statements related to the store group without store merging
488 and with store merging.
489 (imm_store_chain_info::output_merged_store): Adjust split_group
490 callers, punt if estimated number of statements with store merging
491 is not smaller than estimated number of statements without it.
493 (handled_load): Remove has_single_use checks.
494 (pass_store_merging::process_store): Likewise.
496 2017-11-09 Richard Biener <rguenther@suse.de>
498 PR tree-optimization/82902
499 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
501 2017-11-09 Martin Liska <mliska@suse.cz>
504 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
507 2017-11-09 Martin Liska <mliska@suse.cz>
509 PR tree-optimization/82669
510 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
512 2017-11-09 Martin Liska <mliska@suse.cz>
514 PR gcov-profile/48463
515 * coverage.c (coverage_begin_function): Output also end locus
516 of a function and information whether the function is
518 * gcov-dump.c (tag_function): Parse and print the information.
519 * gcov.c (INCLUDE_MAP): Add include.
520 (INCLUDE_SET): Likewise.
521 (struct line_info): Move earlier in the source file because
522 of vector<line_info> in function_info structure.
523 (line_info::line_info): Likewise.
524 (line_info::has_block): Likewise.
525 (struct source_info): Add new member index.
526 (source_info::get_functions_at_location): New function.
527 (function_info::group_line_p): New function.
528 (output_intermediate_line): New function.
529 (output_intermediate_file): Use the mentioned function.
530 (struct function_start): New.
531 (struct function_start_pair_hash): Likewise.
532 (process_file): Add code that identifies group functions.
533 Assign lines either to global or function scope.
534 (generate_results): Skip artificial functions.
535 (find_source): Assign index for each source file.
536 (read_graph_file): Read new flag artificial and end_line.
537 (add_line_counts): Assign it either to global of function scope.
538 (accumulate_line_counts): Isolate core of the function to
539 accumulate_line_info and call it for both function and global
541 (accumulate_line_info): New function.
542 (output_line_beginning): Fix GNU coding style.
543 (print_source_line): New function.
544 (output_line_details): Likewise.
545 (output_function_details): Likewise.
546 (output_lines): Iterate both source (global) scope and function
548 (struct function_line_start_cmp): New class.
549 * doc/gcov.texi: Reflect changes in documentation.
551 2017-11-09 Jakub Jelinek <jakub@redhat.com>
554 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
555 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
556 and similarly for not instead of neg.
558 2017-11-08 Andi Kleen <ak@linux.intel.com>
560 * config/i386/i386.opt: Add -mforce-indirect-call.
561 * config/i386/predicates.md: Check for flag_force_indirect_call.
562 * doc/invoke.texi: Document -mforce-indirect-call
564 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
566 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
568 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
569 config/riscv/riscv.c (predict.h): New include.
570 (riscv_slow_unaligned_access_p): No longer static.
571 (riscv_block_move_straight): Add require.
572 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
574 2017-11-08 Jakub Jelinek <jakub@redhat.com>
577 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
578 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
579 nonimmediate_operand predicate for operand 1 instead of
582 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
584 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
586 * config/aarch64/constraints.md (Uml): New constraint.
587 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
590 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
592 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
593 of two vec_duplicates into a vec_concat.
595 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
597 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
598 Simplify vec_merge of vec_duplicate and vec_concat.
599 * config/aarch64/constraints.md (Utq): New constraint.
600 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
603 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
605 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
606 Simplify vec_merge of vec_duplicate and const_vector.
607 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
609 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
610 mode iterator. Update predicate on operand 1 to
611 handle non-const_vec constants. Delete constraints.
612 (*aarch64_combinez_be<mode>): Likewise for operand 2.
614 2017-11-08 Jakub Jelinek <jakub@redhat.com>
616 PR tree-optimization/78821
617 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
619 (store_operand_info::store_operand_info): Initialize it to false.
620 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
621 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
622 store in the group, and if chain_info is non-NULL, to ignore altogether
624 (compatible_load_p): Fail if bit_not_p does not match.
625 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
626 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
627 (pass_store_merging::process_store): Adjust
628 terminate_all_aliasing_chains calls to pass NULL in all current spots,
629 call terminate_all_aliasing_chains newly when adding a store into
630 a chain with non-NULL chain_info.
632 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
634 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
636 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
638 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
640 (aarch64_layout_frame): Initialise emit_frame_chain.
641 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
642 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
644 2017-11-08 Martin Liska <mliska@suse.cz>
646 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
647 of gimple_call_internal_p.
649 2017-11-07 Tom de Vries <tom@codesourcery.com>
651 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
653 2017-11-07 Tom de Vries <tom@codesourcery.com>
655 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
658 2017-11-08 Martin Liska <mliska@suse.cz>
661 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
663 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
665 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
667 2017-11-07 Jakub Jelinek <jakub@redhat.com>
670 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
671 (*cmp<mode>_ccz_1): New insn with $k alternative.
674 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
675 RTX_COMM_COMPARE as commutative as well.
676 (ix86_binary_operator_ok): Formatting fix.
677 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
678 *<code><mode>3<mask_name><round_saeonly_name>,
679 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
680 *<s>mul<mode>3_highpart<mask_name>,
681 *vec_widen_umult_even_v16si<mask_name>,
682 *vec_widen_umult_even_v8si<mask_name>,
683 *vec_widen_umult_even_v4si<mask_name>,
684 *vec_widen_smult_even_v16si<mask_name>,
685 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
686 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
687 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
688 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
689 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
690 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
691 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
692 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
693 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
694 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
695 ix86_binary_operator_ok. Formatting fixes.
696 (*<plusminus_insn><mode>3<mask_name><round_name>,
697 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
700 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
702 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
703 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
704 eq and ne if TARGET_ISEL.
705 (cmp): New code_iterator.
706 (UNS, UNSU_, UNSIK): New code_attrs.
707 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
708 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
710 ("eq<mode>3"): ... this.
711 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
713 ("ne<mode>3"): ... this.
715 2017-11-07 Julia Koval <julia.koval@intel.com>
718 * common/config/i386/i386-common.c
719 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
720 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
721 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
722 * config/i386/i386.opt: Ditto.
723 * config/i386/i386.c (ix86_target_string): Ditto.
724 (ix86_option_override_internal): Ditto.
725 (ix86_init_mpx_builtins): Move MPX to args2.
726 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
727 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
728 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
729 __builtin_ia32_vgf2p8affineinvqb_v32qi,
730 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
731 __builtin_ia32_vgf2p8affineinvqb_v16qi,
732 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
734 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
737 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
738 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
739 (zero-extendsidi peephole2): Remove peephole.
741 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
744 * stor-layout.c: Include attribs.h.
745 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
746 explicit lookup of "aligned" attribute.
748 2017-11-07 Andrew Waterman <andrew@sifive.com>
750 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
751 (riscv_expand_block_move): Likewise.
752 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
754 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
755 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
756 * config/riscv/riscv.c (riscv_block_move_straight): New function.
757 (riscv_adjust_block_mem): Likewise.
758 (riscv_block_move_loop): Likewise.
759 (riscv_expand_block_move): Likewise.
760 gcc/config/riscv/riscv.md (movmemsi): New pattern.
762 2017-11-07 Michael Clark <michaeljclark@mac.com>
764 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
765 (MUSL_DYNAMIC_LINKER): Likewise.
767 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
769 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
770 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
773 2017-11-07 Richard Biener <rguenther@suse.de>
775 * match.pd: Fix build.
777 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
778 Jackson Woodruff <jackson.woodruff@arm.com>
780 PR tree-optimization/71026
781 * match.pd: Canonicalize negate in division.
783 2017-11-07 Sudakshina Das <sudi.das@arm.com>
786 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
788 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
790 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
791 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
793 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
795 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
796 non-scalar integral types.
797 * match.pd (negate_expr_p): Handle MINUS_EXPR.
798 (-(A-B), -(~A)): New transformations.
800 2017-11-07 Tom de Vries <tom@codesourcery.com>
802 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
803 semicolon after "do {} while (0)".
804 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
805 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
806 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
807 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
808 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
810 2017-11-07 Tom de Vries <tom@codesourcery.com>
812 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
813 after "do {} while (0)".
814 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
815 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
816 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
817 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
818 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
820 2017-11-07 Tom de Vries <tom@codesourcery.com>
823 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
825 (arm_rtx_costs_internal): Add missing semicolon after
826 HANDLE_NARROW_SHIFT_ARITH call.
828 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
830 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
831 disable isel if it was not set explicitly.
833 2017-11-06 James Bowman <james.bowman@ftdichip.com>
835 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
837 (add_type_attribute) likewise.
839 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
841 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
842 of struct ix86_frame.
843 (ix86_initial_elimination_offset): Likewise.
844 (ix86_expand_split_stack_prologue): Likewise.
846 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
848 * tree-vrp.h (enum value_range_type): Update stale comment.
850 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
851 Alan Hayward <alan.hayward@arm.com>
852 David Sherwood <david.sherwood@arm.com>
854 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
855 (aarch64_expand_vec_perm_const): Take the number of units too.
856 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
857 (aarch64_expand_vec_perm_const): Likewise.
858 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
859 (vec_perm<mode>): Update accordingly.
861 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
862 Alan Hayward <alan.hayward@arm.com>
863 David Sherwood <david.sherwood@arm.com>
865 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
866 Take the number of units too.
867 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
868 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
869 but check for a vector mode before rather than after the call.
870 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
871 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
872 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
873 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
874 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
875 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
876 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
877 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
878 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
879 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
880 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
881 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
882 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
883 (aarch64_sqdmull2_n<mode>): Update accordingly.
885 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
886 Alan Hayward <alan.hayward@arm.com>
887 David Sherwood <david.sherwood@arm.com>
889 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
890 the number of units too.
891 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
892 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
893 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
894 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
895 (vec_store_lanesxi<mode>): Update accordingly.
897 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
898 Alan Hayward <alan.hayward@arm.com>
899 David Sherwood <david.sherwood@arm.com>
901 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
902 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
903 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
904 of units rather than the mode.
905 * config/aarch64/iterators.md (nunits): New mode attribute.
906 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
907 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
908 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
909 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
910 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
911 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
912 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
913 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
914 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
915 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
916 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
917 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
918 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
919 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
920 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
921 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
922 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
923 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
924 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
925 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
926 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
927 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
928 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
929 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
930 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
931 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
932 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
933 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
934 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
935 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
936 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
937 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
938 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
939 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
940 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
941 (aarch64_simd_vec_setv2di): Likewise.
943 2017-11-06 Carl Love <cel@us.ibm.com>
945 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
947 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
948 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
950 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
952 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
953 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
954 Add power 8 macro expansions.
955 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
956 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
957 power 8 instructions. (VSX_XXBR): Add iterator.
959 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
961 * config/arm/arm.md (predicable_short_it): Change default to "no",
962 improve documentation, remove uses that are identical to the default.
963 (enabled_for_depr_it): Rename to enabled_for_short_it.
964 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
965 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
966 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
967 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
968 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
970 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
973 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
974 float128 helper macros, which are no longer used after deleting
975 the old 'q' built-in functions, and moving the round to odd
976 built-in functions to being special built-in functions.
977 (BU_FLOAT128_2): Likewise.
978 (BU_FLOAT128_1_HW): Likewise.
979 (BU_FLOAT128_2_HW): Likewise.
980 (BU_FLOAT128_3_HW): Likewise.
981 (FABSQ): Delete old 'q' built-in functions.
982 (COPYSIGNQ): Likewise.
983 (SQRTF128_ODD): Move round to odd built-in functions to be
984 special built-in functions, so that we can handle
985 -mabi=ieeelongdouble.
986 (TRUNCF128_ODD): Likewise.
987 (ADDF128_ODD): Likewise.
988 (SUBF128_ODD): Likewise.
989 (MULF128_ODD): Likewise.
990 (DIVF128_ODD): Likewise.
991 (FMAF128_ODD): Likewise.
992 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
993 built-in names to 'f128'.
994 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
995 old 'q' built-in functions, as the machine independent code for
996 'f128' built-in functions handles this.
997 (rs6000_expand_builtin): Add expansion for float128 round to odd
998 functions, keying off on -mabi=ieeelongdouble of whether to use
999 the KFmode or TFmode variant.
1000 (rs6000_init_builtins): Initialize the _Float128 round to odd
1002 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1003 _Float128 'q' built-in functions are now mapped into the new
1004 'f128' built-in functions.
1006 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1008 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1009 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1011 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1014 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1015 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1016 * toplev.c: Include "ipa-fnsummary.h".
1017 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1019 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1021 PR tree-optimization/82838
1022 * gimple-ssa-store-merging.c
1023 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1024 on a separate gimple_seq which is then appended to seq.
1026 2017-11-06 Jeff Law <law@redhat.com>
1029 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1030 (get_probe_interval): New functions.
1031 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1032 (ix86_adjust_stack_and_probe): Likewise.
1033 (output_adjust_stack_and_probe): Likewise.
1034 (ix86_emit_probe_stack_range): Likewise.
1035 (ix86_expand_prologue): Likewise.
1037 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1039 PR tree-optimization/82816
1040 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1041 if the modes of the two types are the same.
1042 (convert_plusminus_to_widen): Likewise.
1044 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1046 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1048 (usadv16qi): New define_expand.
1049 (usadv8hi): New define_expand.
1051 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1054 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1056 (inline_transform): Likewise.
1057 * predict.c (counts_to_freqs): Remove useless conditional.
1058 * profile-count.h (profile_count::apply_scale): Move sanity check.
1059 * tree-inline.c (copy_bb): Always scale.
1060 (copy_cfg_body): Likewise.
1062 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1065 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1067 (*cmp_ite0): Add enabled_for_depr_it attribute.
1068 (*cmp_ite1): Likewise.
1070 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1072 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1075 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1077 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1078 return the value by pointer.
1079 (extract_range_from_multiplicative_op_1): Update accordingly.
1080 Return as soon as an operation fails.
1082 2017-11-05 Tom de Vries <tom@codesourcery.com>
1085 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1086 (DEF_SANITIZER_BUILTIN): ... here.
1087 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1088 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1090 2017-11-05 Tom de Vries <tom@codesourcery.com>
1093 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1095 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1096 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1097 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1099 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1100 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1101 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1103 2017-11-05 Tom de Vries <tom@codesourcery.com>
1106 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1109 2017-11-04 Michael Clark <michaeljclark@mac.com>
1111 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1112 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1114 (*addsi3_extended): Likewise.
1115 (*addsi3_extended2): Likewise.
1116 (<optab>si3): Likewise.
1117 (<optab>di3): Likewise.
1118 (<optab><mode>3): Likewise.
1119 (<*optabe>si3_internal): Likewise.
1120 (zero_extendqi<SUPERQI:mode>2): Likewise.
1121 (*add<mode>hi3): Likewise.
1122 (*xor<mode>hi3): Likewise.
1123 (<optab>di3): Likewise.
1124 (*<optab>si3_extend): Likewise.
1125 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1126 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1127 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1129 2017-11-04 Andrew Waterman <andrew@sifive.com>
1131 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1132 TARGET_STRICT_ALIGN based upon -mtune argument.
1134 2017-11-04 Andrew Waterman <andrew@sifive.com>
1136 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1138 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1140 config/i386/i386.c (choose_basereg): Use optional scratch
1141 register and add assertion.
1142 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1143 needed, and don't allocate stack.
1144 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1145 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1146 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1148 2017-11-03 Jeff Law <law@redhat.com>
1150 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1151 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1152 to probe at the start of a noreturn function.
1154 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1156 PR tree-optimization/78821
1157 * gimple-ssa-store-merging.c: Update the file comment.
1158 (MAX_STORE_ALIAS_CHECKS): Define.
1159 (struct store_operand_info): New type.
1160 (store_operand_info::store_operand_info): New constructor.
1161 (struct store_immediate_info): Add rhs_code and ops data members.
1162 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1163 and op1r arguments to the ctor, initialize corresponding data members.
1164 (struct merged_store_group): Add load_align_base and load_align
1166 (merged_store_group::merged_store_group): Initialize them.
1167 (merged_store_group::do_merge): Update them.
1168 (merged_store_group::apply_stores): Pick the constant for
1169 encode_tree_to_bitpos from one of the two operands, or skip
1170 encode_tree_to_bitpos if neither operand is a constant.
1171 (class pass_store_merging): Add process_store method decl. Remove
1172 bool argument from terminate_all_aliasing_chains method decl.
1173 (pass_store_merging::terminate_all_aliasing_chains): Remove
1174 var_offset_p argument and corresponding handling.
1175 (stmts_may_clobber_ref_p): New function.
1176 (compatible_load_p): New function.
1177 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1178 if there is overlap and rhs_code is not INTEGER_CST. For
1179 non-overlapping stores terminate group if rhs is not mergeable.
1180 (get_alias_type_for_stmts): Change first argument from
1181 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1182 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1183 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1185 (get_location_for_stmts): Change first argument from
1186 auto_vec<gimple *> & to vec<gimple *> &.
1187 (struct split_store): Remove orig_stmts data member, add orig_stores.
1188 (split_store::split_store): Create orig_stores rather than orig_stmts.
1189 (find_constituent_stmts): Renamed to ...
1190 (find_constituent_stores): ... this. Change second argument from
1191 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1192 to info structures rather than the statements.
1193 (split_group): Rename ALLOW_UNALIGNED argument to
1194 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1195 it. Adjust find_constituent_stores caller.
1196 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1197 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1198 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1199 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1200 (mem_valid_for_store_merging): New function.
1201 (handled_load): New function.
1202 (pass_store_merging::process_store): New method.
1203 (pass_store_merging::execute): Use process_store method. Adjust
1204 terminate_all_aliasing_chains caller.
1206 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1208 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1209 Return true for more constants, symbols and label references.
1210 (aarch64_valid_floating_const): Remove unused function.
1212 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1215 * config/aarch64/aarch64.c (aarch64_layout_frame):
1216 Undo forcing of LR at bottom of frame.
1218 2017-11-03 Jeff Law <law@redhat.com>
1221 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1222 for int_registers_saved.
1224 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1225 extracted from tree-ssa-dom.c.
1226 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1227 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1228 (record_equivalences_from_incoming_edge): Add additional argument
1229 to single_pred_edge_ignoring_loop_edges call.
1230 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1231 (uncprop_dom_walker::before_dom_children): Add additional argument
1232 to single_pred_edge_ignoring_loop_edges call.
1233 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1234 single_pred_edge_ignoring_loop_edges rather than open coding.
1235 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1237 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1239 * match.pd (-(-A)): Rewrite.
1241 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1243 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1244 (rs6000_emit_int_cmove): New declaration.
1245 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1246 (rs6000_emit_sISEL): Delete.
1247 (rs6000_emit_int_cmove): Make non-static.
1248 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1249 instead of rs6000_emit_sISEL.
1251 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1253 * asan.c (create_cond_insert_point): Maintain profile.
1254 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1256 * basic-block.h (struct basic_block_def): Remove frequency.
1257 (EDGE_FREQUENCY): Use to_frequency
1258 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1260 (find_traces): Update to use to_frequency.
1261 (find_traces_1_round): Likewise; use only IPA counts.
1262 (bb_to_key): Likewise.
1263 (connect_traces): Use IPA counts only.
1264 (copy_bb_p): Update to use to_frequency.
1265 (fix_up_crossing_landing_pad): Likewise.
1266 (sanitize_hot_paths): Likewise.
1267 * bt-load.c (basic_block_freq): Likewise.
1268 * cfg.c (init_flow): Set count_max to uninitialized.
1269 (check_bb_profile): Remove frequencies; check counts.
1270 (dump_bb_info): Do not dump frequencies.
1271 (update_bb_profile_for_threading): Update counts only.
1272 (scale_bbs_frequencies_int): Likewise.
1273 (MAX_SAFE_MULTIPLIER): Remove.
1274 (scale_bbs_frequencies_gcov_type): Update counts only.
1275 (scale_bbs_frequencies_profile_count): Update counts only.
1276 (scale_bbs_frequencies): Update counts only.
1277 * cfg.h (struct control_flow_graph): Add count-max.
1278 (update_bb_profile_for_threading): Update prototype.
1279 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1280 (find_many_sub_basic_blocks): Likewise.
1281 * cfgcleanup.c (try_forward_edges): Likewise.
1282 (try_crossjump_to_edge): Likewise.
1283 * cfgexpand.c (expand_gimple_cond): Likewise.
1284 (expand_gimple_tailcall): Likewise.
1285 (construct_init_block): Likewise.
1286 (construct_exit_block): Likewise.
1287 * cfghooks.c (verify_flow_info): Check consistency of counts.
1288 (dump_bb_for_graph): Do not dump frequencies.
1289 (split_block_1): Do not update frequencies.
1290 (split_edge): Do not update frequencies.
1291 (make_forwarder_block): Do not update frequencies.
1292 (duplicate_block): Do not update frequencies.
1293 (account_profile_record): Do not update frequencies.
1294 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1295 for global heuristics.
1296 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1297 (expected_loop_iterations_unbounded): Use counts only.
1298 * cfgloopmanip.c (scale_loop_profile): Simplify.
1299 (create_empty_loop_on_edge): Simplify
1301 (duplicate_loop_to_header_edge): Simplify
1302 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1303 (update_br_prob_note): Take care of removing note when profile
1305 (relink_block_chain): Do not dump frequency.
1306 (rtl_account_profile_record): Use to_frequency.
1307 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1308 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1309 (cgraph_update_edges_for_call_stmt_node): Likewise.
1310 (cgraph_edge::verify_count_and_frequency): Update.
1311 (cgraph_node::verify_node): Temporarily disable frequency verification.
1312 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1313 to_cgraph_frequency.
1314 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1315 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1317 (cgraph_node::expand_thunk): Update profile.
1318 * except.c (dw2_build_landing_pads): Do not update frequency.
1319 * final.c (compute_alignments): Use to_frequency.
1320 (dump_basic_block_info): Do not dump frequency.
1321 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1322 (dump_gimple_bb_header): Do not dump frequency.
1323 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1325 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1326 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1327 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1328 (init_before_recovery): Do not update frequency.
1329 (sched_create_recovery_edges): Do not update frequency.
1330 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1331 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1332 (ipa_cp_c_finalize): Set max_count to uninitialized.
1333 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1334 (param_change_prob): Use counts.
1335 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1337 * ipa-split.c (consider_split): Use to_frequency.
1338 (split_function): Use to_frequency.
1339 * ira-build.c (loop_compare_func): Likewise.
1340 (mark_loops_for_removal): Likewise.
1341 (mark_all_loops_for_removal): Likewise.
1342 * loop-doloop.c (doloop_modify): Do not update frequency.
1343 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1345 * lto-streamer-in.c (input_function): Update count_max.
1346 * omp-expand.c (expand_omp_taskreg): Update count_max.
1347 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1348 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1349 (maybe_hot_count_p): Use ipa counts only.
1350 (maybe_hot_bb_p): Simplify.
1351 (maybe_hot_edge_p): Simplify.
1352 (probably_never_executed): Do not take frequency argument.
1353 (probably_never_executed_bb_p): Do not pass frequency.
1354 (probably_never_executed_edge_p): Likewise.
1355 (combine_predictions_for_bb): Check that profile is nonzero.
1356 (propagate_freq): Do not set frequency.
1357 (drop_profile): Simplify.
1358 (counts_to_freqs): Simplify.
1359 (expensive_function_p): Use to_frequency.
1360 (propagate_unlikely_bbs_forward): Simplify.
1361 (determine_unlikely_bbs): Simplify.
1362 (estimate_bb_frequencies): Add hack to silence graphite issues.
1363 (compute_function_frequency): Use ipa counts.
1364 (pass_profile::execute): Update.
1365 (rebuild_frequencies): Use counts only.
1366 (force_edge_cold): Use counts only.
1367 * profile-count.c (profile_count::dump): Dump new count types.
1368 (profile_count::differs_from_p): Check compatiblity.
1369 (profile_count::to_frequency): New function.
1370 (profile_count::to_cgraph_frequency): New function.
1371 * profile-count.h (struct function): Declare.
1372 (enum profile_quality): Add profile_guessed_local and
1373 profile_guessed_global0.
1374 (class profile_proability): Decrease number of bits to 29;
1375 update from_reg_br_prob_note and to_reg_br_prob_note.
1376 (class profile_count: Update comment; decrease number of bits
1377 to 61. Check compatibility.
1378 (profile_count::compatible_p): New private member function.
1379 (profile_count::ipa_p): New member function.
1380 (profile_count::operator<): Handle global zero correctly.
1381 (profile_count::operator>): Handle global zero correctly.
1382 (profile_count::operator<=): Handle global zero correctly.
1383 (profile_count::operator>=): Handle global zero correctly.
1384 (profile_count::nonzero_p): New member function.
1385 (profile_count::force_nonzero): New member function.
1386 (profile_count::max): New member function.
1387 (profile_count::apply_scale): Handle IPA scalling.
1388 (profile_count::guessed_local): New member function.
1389 (profile_count::global0): New member function.
1390 (profile_count::ipa): New member function.
1391 (profile_count::to_frequency): Declare.
1392 (profile_count::to_cgraph_frequency): Declare.
1393 * profile.c (OVERLAP_BASE): Delete.
1394 (compute_frequency_overlap): Delete.
1395 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1396 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1397 * sched-ebb.c (rank): Use counts only.
1398 * shrink-wrap.c (handle_simple_exit): Use counts only.
1399 (try_shrink_wrapping): Use counts only.
1400 (place_prologue_for_one_component): Use counts only.
1401 * tracer.c (find_best_predecessor): Use to_frequency.
1402 (find_trace): Use to_frequency.
1403 (tail_duplicate): Use to_frequency.
1404 * trans-mem.c (expand_transaction): Do not update frequency.
1405 * tree-call-cdce.c: Do not update frequency.
1406 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1407 (gimple_merge_blocks): Likewise.
1408 (gimple_split_edge): Likewise.
1409 (gimple_duplicate_sese_region): Likewise.
1410 (gimple_duplicate_sese_tail): Likewise.
1411 (move_sese_region_to_fn): Likewise.
1412 (gimple_account_profile_record): Likewise.
1413 (insert_cond_bb): Likewise.
1414 * tree-complex.c (expand_complex_div_wide): Likewise.
1415 * tree-eh.c (lower_resx): Update profile.
1416 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1418 (initialize_cfun): Do not initialize frequencies
1419 (freqs_to_counts): Delete.
1420 (copy_cfg_body): Ignore count parameter.
1421 (copy_body): Update.
1422 (expand_call_inline): Update count_max.
1423 (optimize_inline_calls): Update count_max.
1424 (tree_function_versioning): Update count_max.
1425 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1426 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1428 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1429 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1430 (try_peel_loop): Likewise.
1431 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1433 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1434 (tree_transform_and_unroll_loop): Do not use frequencies
1435 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1436 Use reliable prediction only.
1437 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1438 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1439 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1440 probability scaling.
1441 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1442 not update frequency
1443 (any_remaining_duplicated_blocks): Likewise.
1444 (update_profile): Likewise.
1445 (estimated_freqs_path): Delete.
1446 (freqs_to_counts_path): Delete.
1447 (clear_counts_path): Delete.
1448 (ssa_fix_duplicate_block_edges): Likewise.
1449 (duplicate_thread_path): Likewise.
1450 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1451 * tree-tailcall.c (decrease_profile): Do not update frequency.
1452 (eliminate_tail_call): Likewise.
1453 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1454 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1455 (optimize_mask_stores): Likewise.
1456 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1457 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1458 (ubsan_expand_ptr_ifn): Update profile.
1459 * value-prof.c (gimple_ic): Simplify.
1460 * value-prof.h (gimple_ic): Update prototype.
1461 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1462 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1464 (want_inline_self_recursive_call_p): Likewise.
1465 (resolve_noninline_speculation): Only cummulate defined counts.
1466 (inline_small_functions): Use nonzero_p.
1467 (ipa_inline): Do not access freed node.
1469 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1471 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1472 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1474 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1476 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1477 non-legitimate address.
1479 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1481 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1482 (*lt0_<mode>di, *lt0_<mode>si): New.
1484 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1486 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1487 TARGET_PAIRED_FLOAT.
1489 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1490 Jim Wilson <jim.wilson@linaro.org>
1492 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1493 * config/aarch64/aarch64-tune.md: Regenerated.
1494 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1495 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1497 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1499 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1500 (arc_expand_prologue): Restore blink for millicode.
1501 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1503 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1506 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1507 gen_vec_duplicate after forcing the scalar into a register.
1509 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1511 * combine (try_combine): Print the insns input to try_combine to the
1514 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1517 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1518 Remove second argument from aarch64_process_target_attr call.
1519 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1521 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1523 (aarch64_handle_attr_arch): Remove second argument.
1524 (aarch64_handle_attr_cpu): Ditto.
1525 (aarch64_handle_attr_tune): Ditto.
1526 (aarch64_handle_attr_isa_flags): Ditto.
1527 (aarch64_process_one_target_attr): Ditto.
1528 (aarch64_process_target_attr): Ditto.
1529 (aarch64_option_valid_attribute_p): Remove second argument.
1530 on aarch64_process_target_attr call.
1532 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1534 * diagnostic.c: Include "selftest-diagnostic.h".
1535 (selftest::assert_location_text): New function.
1536 (selftest::test_diagnostic_get_location_text): New function.
1537 (selftest::diagnostic_c_tests): Call it.
1539 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1541 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1542 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1543 (class selftest::test_diagnostic_context): Move to...
1544 * selftest-diagnostic.c: New file.
1545 * selftest-diagnostic.h: New file.
1547 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1549 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1550 offset range for FT32B.
1551 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1552 * config/ft32/ft32.md: Add TARGET_NOPM.
1553 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1554 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1556 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1558 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1560 2017-11-02 Jeff Law <law@redhat.com>
1562 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1563 virtual keyword on FINAL OVERRIDE members.
1565 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1566 virtuals together. Add virtual destructor.
1567 (substitute_and_fold_engine): Similarly.
1569 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1571 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1573 2017-11-02 Richard Biener <rguenther@suse.de>
1575 PR tree-optimization/82795
1576 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1578 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1580 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1582 (gcc_GAS_CHECK_FEATURE): Remove.
1583 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1584 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1585 * configure: Regenerate.
1587 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1589 * config/arc/arc.c (hwloop_optimize): Account for empty
1592 2017-11-02 Richard Biener <rguenther@suse.de>
1595 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1596 Truncate ARRAY_REF index and element size.
1598 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1600 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1602 2017-11-01 Jeff Law <law@redhat.com>
1604 * tree-ssa-ccp.c (ccp_folder): New class derived from
1605 substitute_and_fold_engine.
1606 (ccp_folder::get_value): New member function.
1607 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1608 (ccp_fold_stmt): Remove prototype.
1609 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1610 * tree-ssa-copy.c (copy_folder): New class derived from
1611 substitute_and_fold_engine.
1612 (copy_folder::get_value): Renamed from get_value.
1613 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1614 * tree-vrp.c (vrp_folder): New class derived from
1615 substitute_and_fold_engine.
1616 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1617 (vrp_folder::get_value): New member function.
1618 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1619 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1620 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1621 provide a class interface to folder/substitute routines.
1622 (ssa_prop_fold_stmt_fn): Remove typedef.
1623 (ssa_prop_get_value_fn): Likewise.
1624 (subsitute_and_fold): Remove prototype.
1625 (replace_uses_in): Likewise.
1626 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1627 Renamed from replace_uses_in. Call the virtual member function
1628 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1629 (substitute_and_fold_dom_walker): Remove initialization of
1630 data member entries for calbacks. Add substitute_and_fold_engine
1631 member and initialize it.
1632 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1633 member functions for get_value, replace_phi_args_in c
1634 replace_uses_in, and fold_stmt calls.
1635 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1636 substitute_and_fold. Remove assert. Update ctor call.
1638 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1639 (ssa_prop_visit_phi_fn): Likewise.
1640 (class ssa_propagation_engine): New class to provide an interface
1642 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1644 (ssa_prop_visit_phi): Likewise.
1645 (ssa_propagation_engine::simulate_stmt): Moved into class.
1646 Call visit_phi/visit_stmt from the class rather than via
1647 file scoped static variables.
1648 (ssa_propagation_engine::simulate_block): Moved into class.
1649 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1650 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1651 set file scoped statics for the visit_stmt/visit_phi callbacks.
1652 * tree-complex.c (complex_propagate): New class derived from
1653 ssa_propagation_engine.
1654 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1655 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1656 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1658 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1659 ssa_propagation_engine.
1660 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1661 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1662 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1663 * tree-ssa-copy.c (copy_prop): New class derived from
1664 ssa_propagation_engine.
1665 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1666 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1667 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1668 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1669 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1670 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1671 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1673 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1675 PR rtl-optimization/82778
1676 PR rtl-optimization/82597
1677 * compare-elim.c (struct comparison): Add in_a_setter field.
1678 (find_comparison_dom_walker::before_dom_children): Remove killed
1679 bitmap and df_simulate_find_defs call, instead walk the defs.
1680 Compute last_setter and initialize in_a_setter. Merge definitions
1681 with first initialization for a few variables.
1682 (try_validate_parallel): Use insn_invalid_p instead of
1683 recog_memoized. Return insn rather than just the pattern.
1684 (try_merge_compare): Fix up comment. Don't uselessly test if
1685 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1687 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1690 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1691 Alan Hayward <alan.hayward@arm.com>
1692 David Sherwood <david.sherwood@arm.com>
1694 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1695 aarch64_hard_regno_nregs to get the number of registers
1698 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1699 Alan Hayward <alan.hayward@arm.com>
1700 David Sherwood <david.sherwood@arm.com>
1702 * config/aarch64/constraints.md (Upl): Rename to...
1704 * config/aarch64/aarch64.md
1705 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1708 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1709 Alan Hayward <alan.hayward@arm.com>
1710 David Sherwood <david.sherwood@arm.com>
1712 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1713 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1716 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1717 Alan Hayward <alan.hayward@arm.com>
1718 David Sherwood <david.sherwood@arm.com>
1720 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1721 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1722 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1724 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1725 of a single element.
1726 * config/aarch64/iterators.md: Add a comment above the permute
1727 unspecs to say that they are generated directly by
1728 aarch64_expand_vec_perm_const.
1729 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1731 2017-11-01 Nathan Sidwell <nathan@acm.org>
1733 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1735 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1737 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1738 and medany code models, and describe what they do.
1740 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1742 Revert accidental duplicate:
1744 * combine.c (can_change_dest_mode): Reject changes in
1745 REGMODE_NATURAL_SIZE.
1747 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1749 PR rtl-optimization/64682
1750 PR rtl-optimization/69567
1751 PR rtl-optimization/69737
1752 PR rtl-optimization/82683
1753 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1754 register mentioned in the note, drop the note, unless it came from I3,
1755 in which case it should go to I3 again.
1757 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1759 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1760 and return false if not.
1761 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1763 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1765 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1766 range is known to be empty.
1768 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1769 Alan Hayward <alan.hayward@arm.com>
1770 David Sherwood <david.sherwood@arm.com>
1772 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1773 and CONST_VECTOR_NUNITS instead of computing the number of units from
1774 the byte sizes of the vector and element.
1775 (simplify_binary_operation_1): Likewise.
1776 (simplify_const_binary_operation): Likewise.
1777 (simplify_ternary_operation): Likewise.
1779 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1780 Alan Hayward <alan.hayward@arm.com>
1781 David Sherwood <david.sherwood@arm.com>
1783 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1784 (int_mem_offset): ...this new function.
1785 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1786 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1787 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1790 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1791 Alan Hayward <alan.hayward@arm.com>
1792 David Sherwood <david.sherwood@arm.com>
1794 * lower-subreg.c (interesting_mode_p): New function.
1795 (compute_costs, find_decomposable_subregs, decompose_register)
1796 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1797 (resolve_clobber, dump_choices): Use it.
1799 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1800 Alan Hayward <alan.hayward@arm.com>
1801 David Sherwood <david.sherwood@arm.com>
1803 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1805 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1806 Alan Hayward <alan.hayward@arm.com>
1807 David Sherwood <david.sherwood@arm.com>
1809 * alias.c (find_base_value, find_base_term): Only process integer
1810 truncations. Check the precision rather than the size.
1812 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1813 Alan Hayward <alan.hayward@arm.com>
1814 David Sherwood <david.sherwood@arm.com>
1816 * machmode.h (is_narrower_int_mode): New function
1817 * optabs.c (expand_float, expand_fix): Use it.
1818 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1820 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1821 Alan Hayward <alan.hayward@arm.com>
1822 David Sherwood <david.sherwood@arm.com>
1824 * rtl.h (narrower_subreg_mode): New function.
1825 * ira-color.c (update_costs_from_allocno): Use it.
1827 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1828 Alan Hayward <alan.hayward@arm.com>
1829 David Sherwood <david.sherwood@arm.com>
1831 * optabs-query.h (convert_optab_p): New function, split out from...
1832 (convert_optab_handler): ...here.
1833 (widening_optab_handler): Delete.
1834 (find_widening_optab_handler): Remove permit_non_widening parameter.
1835 (find_widening_optab_handler_and_mode): Likewise. Provide an
1836 override that operates on mode class wrappers.
1837 * optabs-query.c (widening_optab_handler): Delete.
1838 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1839 parameter. Assert that the two modes are the same class and that
1840 the "from" mode is narrower than the "to" mode. Use
1841 convert_optab_handler instead of widening_optab_handler.
1842 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1843 instead of widening_optab_handler.
1844 * expr.c (expand_expr_real_2): Update calls to
1845 find_widening_optab_handler.
1846 * optabs.c (expand_widen_pattern_expr): Likewise.
1847 (expand_binop_directly): Take the insn_code as a parameter.
1848 (expand_binop): Only call find_widening_optab_handler for
1849 conversion optabs; use optab_handler otherwise. Update calls
1850 to find_widening_optab_handler and expand_binop_directly.
1851 Use convert_optab_handler instead of widening_optab_handler.
1852 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1853 find_widening_optab_handler and use scalar_mode rather than
1855 (convert_plusminus_to_widen): Likewise.
1857 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1858 Alan Hayward <alan.hayward@arm.com>
1859 David Sherwood <david.sherwood@arm.com>
1861 * machmode.h (fixed_size_mode): New class.
1862 * rtl.h (get_pool_mode): Return fixed_size_mode.
1863 * gengtype.c (main): Add fixed_size_mode.
1864 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1865 (get_raw_arg_mode): Likewise.
1866 * doc/tm.texi: Regenerate.
1867 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1868 * targhooks.c (default_get_reg_raw_mode): Likewise.
1869 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1870 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1871 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1872 (msp430_get_raw_result_mode): Likewise.
1873 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1874 * dbxout.c (dbxout_parms): Require fixed-size modes.
1875 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1876 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1877 * omp-low.c (lower_oacc_reductions): Likewise.
1878 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1879 (simplify_subreg): Update accordingly.
1880 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1881 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1882 that aren't fixed-size.
1883 (get_pool_mode): Return a fixed_size_mode.
1884 (output_constant_pool_2): Take a fixed_size_mode.
1886 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1887 Alan Hayward <alan.hayward@arm.com>
1888 David Sherwood <david.sherwood@arm.com>
1890 * doc/rtl.texi (vec_series): Document.
1891 (const): Say that the operand can be a vec_series.
1892 * rtl.def (VEC_SERIES): New rtx code.
1893 * rtl.h (const_vec_series_p_1): Declare.
1894 (const_vec_series_p): New function.
1895 * emit-rtl.h (gen_const_vec_series): Declare.
1896 (gen_vec_series): Likewise.
1897 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1898 (gen_vec_series): Likewise.
1899 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1900 * simplify-rtx.c (simplify_unary_operation): Handle negations
1902 (simplify_binary_operation_series): New function.
1903 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1904 (test_vector_ops_series): New function.
1905 (test_vector_ops): Call it.
1906 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1907 gen_const_vec_series.
1908 (altivec_lvsr): Likewise.
1909 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1911 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1912 Alan Hayward <alan.hayward@arm.com>
1913 David Sherwood <david.sherwood@arm.com>
1915 * doc/rtl.texi (const): Update description of address constants.
1916 Say that vector constants are allowed too.
1917 * common.md (E, F): Use CONSTANT_P instead of checking for
1919 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1920 checking for CONST_VECTOR.
1921 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1923 * expr.c (expand_expr_real_2): Check for vector modes instead
1924 of checking for CONST_VECTOR.
1925 * rtl.h (const_vec_p): New function.
1926 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1927 (unwrap_const_vec_duplicate): Handle them here too.
1929 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1930 David Malcolm <dmalcolm@redhat.com>
1931 Alan Hayward <alan.hayward@arm.com>
1932 David Sherwood <david.sherwood@arm.com>
1934 * rtl.h (vec_duplicate_p): New function.
1935 * selftest-rtl.c (assert_rtx_eq_at): New function.
1936 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1937 (assert_rtx_eq_at): Declare.
1938 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1939 * selftest-run-tests.c (selftest::run_tests): Call it.
1940 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1941 (simplify_unary_operation_1): Recursively handle vector duplicates.
1942 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1944 (simplify_subreg): Handle subregs of vector duplicates.
1945 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1946 (selftest::simplify_rtx_c_tests): New functions.
1948 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1949 Alan Hayward <alan.hayward@arm.com>
1950 David Sherwood <david.sherwood@arm.com>
1952 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1953 (gen_vec_duplicate): Likewise.
1954 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1956 (gen_const_vector): ...here.
1957 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1958 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1959 whose elements are all equal.
1960 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1961 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1962 (simplify_relational_operation): Likewise.
1963 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1965 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1966 (aarch64_expand_vector_init): Likewise.
1967 * config/arm/arm.c (neon_vdup_constant): Likewise.
1968 (neon_expand_vector_init): Likewise.
1969 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1970 (arm_block_set_unaligned_vect): Likewise.
1971 (arm_block_set_aligned_vect): Likewise.
1972 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1973 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1974 (expand_vec_perm_even_odd_pack): Likewise.
1975 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1976 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1977 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1978 gen_const_vec_duplicate.
1979 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1980 * config/mips/mips.c (mips_gen_const_int_vector): Use
1981 gen_const_vec_duplicate.
1982 (mips_expand_vector_init): Use CONST0_RTX.
1983 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1984 (define_split): Use gen_const_vec_duplicate.
1985 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1986 (define_split): Use gen_const_vec_duplicate.
1987 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1988 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1989 * config/spu/spu.c (spu_const): Likewise.
1991 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1992 Alan Hayward <alan.hayward@arm.com>
1993 David Sherwood <david.sherwood@arm.com>
1995 * combine.c (can_change_dest_mode): Reject changes in
1996 REGMODE_NATURAL_SIZE.
1998 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2000 * configure.ac (--enable-libssp): New.
2001 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2002 trying to determine target-specific default. Adjust indentation.
2003 * configure: Regenerated.
2004 * doc/install.texi (Configuration): Expand --disable-libssp
2007 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2009 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2012 2017-10-31 Martin Jambor <mjambor@suse.cz>
2015 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2017 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2019 * auto-profile.c (autofdo_source_profile::read): Use
2020 UNKNOWN_LOCATION rather than 0.
2021 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2022 (warning_at): ...this overload.
2023 (warning_at_rich_loc_n): Rename to...
2024 (warning_n): ...this overload.
2025 (error_at_rich_loc): Rename to...
2026 (error_at): ...this overload.
2027 (pedwarn_at_rich_loc): Rename to...
2028 (pedwarn): ...this overload.
2029 (permerror_at_rich_loc): Rename to...
2030 (permerror): ...this overload.
2031 (inform_at_rich_loc): Rename to...
2032 (inform): ...this overload.
2033 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2034 (diagnostic_n_impl_richloc): Rename to...
2035 (diagnostic_n_impl): ...this rich_location *-based decl.
2036 (inform_at_rich_loc): Rename to...
2037 (inform): ...this, and add an assertion.
2038 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2039 (warning_at_rich_loc): Rename to...
2040 (warning_at): ...this, and add an assertion.
2041 (warning_at_rich_loc_n): Rename to...
2042 (warning_n): ...this, and add an assertion.
2043 (warning_n): Update location_t-based implementation for removal of
2044 location_t-based diagnostic_n_impl.
2045 (pedwarn_at_rich_loc): Rename to...
2046 (pedwarn): ...this, and add an assertion.
2047 (permerror_at_rich_loc): Rename to...
2048 (permerror): ...this, and add an assertion.
2049 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2050 (error_at_rich_loc): Rename to...
2051 (error_at): ...this, and add an assertion.
2052 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2053 (driver::do_spec_on_infiles): Likewise.
2054 * substring-locations.c (format_warning_va): Update for renaming
2055 of inform_at_rich_loc.
2057 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2059 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2060 _Float<N>X built-in functions so that the variant without the
2061 "__builtin_" prefix is only enabled for the GNU C and Objective C
2062 languages when they are in non-strict ANSI/ISO mode.
2063 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2064 * target.def (floatn_builtin_p): Add a target hook to control
2065 whether _Float<N> and _Float<N>X built-in functions without the
2066 "__builtin_" prefix are enabled, and return true for C and
2067 Objective C in the default hook. Include langhooks.h in
2069 * targhooks.h (default_floatn_builtin_p): Likewise.
2070 * targhooks.c (default_floatn_builtin_p): Likewise.
2071 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2072 floatn_builtin_p target hook.
2073 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2075 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2076 Eric Botcazou <ebotcazou@adacore.com>
2078 PR rtl-optimization/81803
2079 * lra-constraints.c (curr_insn_transform): Also reload the whole
2080 register for a strict subreg no wider than a word if this is for
2081 a WORD_REGISTER_OPERATIONS target.
2083 2017-10-31 Jason Merrill <jason@redhat.com>
2085 * gdbinit.in: Skip over inlines from timevar.h.
2087 2017-10-31 Martin Liska <mliska@suse.cz>
2089 * doc/gcov.texi: Document new option.
2090 * gcov.c (print_usage): Likewise print it.
2091 (process_args): Support the argument.
2092 (format_count): New function.
2093 (format_gcov): Use the function.
2095 2017-10-31 Martin Liska <mliska@suse.cz>
2097 * gcov.c (struct name_map): do not use typedef.
2098 Define operator== and operator<.
2099 (name_search): Remove.
2100 (name_sort): Remove.
2101 (main): Do not allocate names.
2102 (process_file): Add vertical space.
2103 (generate_results): Use std::find.
2104 (release_structures): Do not release memory.
2105 (find_source): Use std::find.
2107 2017-10-31 Martin Liska <mliska@suse.cz>
2109 * gcov.c (struct line_info): Remove it's typedef.
2110 (line_info::line_info): Add proper ctor.
2111 (line_info::has_block): Do not use a typedef.
2112 (struct source_info): Do not use typedef.
2113 (circuit): Likewise.
2114 (get_cycles_count): Likewise.
2115 (output_intermediate_file): Iterate via vector iterator.
2116 (add_line_counts): Use std::vector methods.
2117 (accumulate_line_counts): Likewise.
2118 (output_lines): Likewise.
2120 2017-10-31 Martin Liska <mliska@suse.cz>
2122 * gcov.c (struct source_info): Remove typedef.
2123 (source_info::source_info): Add proper ctor.
2124 (accumulate_line_counts): Use struct, not it's typedef.
2125 (output_gcov_file): Likewise.
2126 (output_lines): Likewise.
2127 (main): Do not allocate an array.
2128 (output_intermediate_file): Use size of vector container.
2129 (process_file): Resize the vector.
2130 (generate_results): Do not preallocate, use newly added vector
2132 (release_structures): Do not release sources.
2133 (find_source): Use vector methods.
2134 (add_line_counts): Do not use typedef.
2136 2017-10-31 Martin Liska <mliska@suse.cz>
2138 * doc/gcov.texi: Document that.
2139 * gcov.c (add_line_counts): Mark lines with a non-executed
2141 (output_line_beginning): Handle such lines.
2142 (output_lines): Pass new argument.
2143 (output_intermediate_file): Print it in intermediate format.
2145 2017-10-31 Martin Liska <mliska@suse.cz>
2147 * color-macros.h: New file.
2148 * diagnostic-color.c: Factor out color related to macros to
2150 * doc/gcov.texi: Document -k option.
2151 * gcov.c (INCLUDE_STRING): Include string.h.
2152 (print_usage): Add -k option.
2153 (process_args): Parse it.
2154 (pad_count_string): New function.
2155 (output_line_beginning): Likewise.
2156 (DEFAULT_LINE_START): New macro.
2157 (output_lines): Support color output.
2159 2017-10-31 Martin Liska <mliska@suse.cz>
2161 PR gcov-profile/82633
2162 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2163 their interaction with GCOV infrastructure.
2164 * configure.ac: Add -fkeep-{inline,static}-functions to
2166 * configure: Regenerate.
2168 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2171 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2173 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2176 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2177 into a register if it does not fit into an immediate offset field.
2179 2017-10-31 Olivier Hainque <hainque@adacore.com>
2181 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2183 2017-10-31 Julia Koval <julia.koval@intel.com>
2185 * config.gcc: Add gfniintrin.h.
2186 * config/i386/gfniintrin.h: New.
2187 * config/i386/i386-builtin-types.def
2188 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2189 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2190 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2191 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2192 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2193 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2194 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2195 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2196 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2197 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2198 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2199 * config/i386/immintrin.h: Include gfniintrin.h.
2200 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2202 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2204 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2206 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2208 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2209 (arm_ashldi3_1bit): Remove pattern.
2210 (ashrdi3): Remove shift by 1 expansion.
2211 (arm_ashrdi3_1bit): Remove pattern.
2212 (lshrdi3): Remove shift by 1 expansion.
2213 (arm_lshrdi3_1bit): Remove pattern.
2214 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2215 cost of ashldi3 by 1.
2216 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2217 (<shift>di3_neon): Likewise.
2219 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2221 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2222 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2223 and (*aarch64_simd_mov<VQ:mode>).
2224 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2225 pattern alternative.
2226 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2227 attributes to match pattern alternative.
2229 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2231 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2232 * config/rs6000/emmintrin.h: New file.
2233 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2235 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2237 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2238 * (movdi_vfp_cortexa8): Remove pattern.
2240 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2242 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2244 (Specific, alpha*-dec-osf5.1): Remove.
2245 (Specific, mips-sgi-irix5): Remove.
2246 (Specific, mips-sgi-irix6): Remove.
2248 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2251 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2252 arguments to clear_bit_region_be.
2254 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2256 * gimplify.c: Include memmodel.h.
2258 2017-10-30 Martin Jambor <mjambor@suse.cz>
2260 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2261 condition whether loop should be executed at all.
2263 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2265 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2266 gimple folding of vec_madd() intrinsics.
2267 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2268 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2269 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2271 2017-10-30 Richard Biener <rguenther@suse.de>
2273 PR tree-optimization/82762
2275 2017-10-23 Richard Biener <rguenther@suse.de>
2277 PR tree-optimization/82129
2279 2017-08-01 Richard Biener <rguenther@suse.de>
2281 PR tree-optimization/81181
2282 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2283 (compute_antic): ... end of iteration here.
2285 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2287 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2288 -std=iso9899:2017 and -std=gnu17.
2289 * doc/standards.texi (C Language): Document C17 support.
2290 * doc/cpp.texi (Overview): Mention -std=c17.
2291 (Standard Predefined Macros): Document C11 and C17 values of
2292 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2293 * doc/extend.texi (Inline): Do not list individual options for
2294 standards newer than C99.
2295 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2297 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2300 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2302 * asan.c (asan_finish_file): Align asan globals array by shadow
2305 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2308 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2309 (struct store_immediate_info): Add bitregion_start and bitregion_end
2311 (store_immediate_info::store_immediate_info): Add brs and bre
2312 arguments and initialize bitregion_{start,end} from those.
2313 (struct merged_store_group): Add bitregion_start, bitregion_end,
2314 align_base and mask fields. Drop unnecessary struct keyword from
2315 struct store_immediate_info. Add do_merge method.
2316 (clear_bit_region_be): Use memset instead of loop storing zeros.
2317 (merged_store_group::do_merge): New method.
2318 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2319 stores as long as the surrounding bitregions have no gaps.
2320 (merged_store_group::merge_overlapping): Use do_merge.
2321 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2322 is byte aligned, rather than requiring that start and width are
2323 byte aligned. Drop unnecessary struct keyword from
2324 struct store_immediate_info. Allocate and populate also mask array.
2325 Make start of the arrays relative to bitregion_start rather than
2326 start and size them according to bitregion_{end,start} difference.
2327 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2328 struct store_immediate_info.
2329 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2330 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2331 struct keyword from struct store_immediate_info.
2332 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2333 between stores as long as the surrounding bitregions have no gaps.
2335 (struct split_store): Add orig non-static data member.
2336 (split_store::split_store): Initialize orig to false.
2337 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2338 if there is exactly a single original stmt. Change stmts argument
2339 to pointer from reference, if NULL, don't push anything to it. Add
2340 first argument, use it to optimize skipping over orig stmts that
2341 are known to be before bitpos already. Simplify.
2342 (split_group): Return unsigned int count how many stores are or
2343 would be needed rather than a bool. Add allow_unaligned argument.
2344 Change split_stores argument from reference to pointer, if NULL,
2345 only do a dry run computing how many stores would be produced.
2346 Rewritten algorithm to use both alignment and misalign if
2347 !allow_unaligned and handle bitfield stores with gaps.
2348 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2349 from bitregion_start instead of start. Compute allow_unaligned
2350 here, if true, do 2 split_group dry runs to compute which one
2351 produces fewer stores and prefer aligned if equal. Punt if
2352 new count is bigger or equal than original before emitting any
2353 statements, rather than during that. Remove no longer needed
2354 new_ssa_names tracking. Replace num_stmts with
2355 split_stores.length (). Use 32-bit stack allocated entries
2356 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2357 if possible. Handle bitfields with gaps.
2358 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2359 Compute bitregion_{start,end} for the stores and construct
2360 store_immediate_info with that. Formatting fixes.
2362 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2365 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2366 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2368 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2370 * gimplify.c: Include tm_p.h.
2372 * common.opt (gcoff): Re-add as ignored option.
2373 (gcoff1, gcoff2, gcoff3): Likewise.
2375 * Makefile.in (OBJS): Delete sdbout.o.
2376 (GTFILES): Delete $(srcdir)/sdbout.c.
2377 * debug.h: Delete sdb_debug_hooks.
2378 * final.c: Delete sdbout.h include.
2379 (final_scan_insn): Delete SDB_DEBUG check.
2380 (rest_of_clean_state): Likewise.
2381 * output.h: Delete sdb_begin_function_line.
2384 * toplev.c: Delete sdbout.h include.
2385 (process_options): Delete SDB_DEBUG check.
2386 * tree-core.h (tree_type_common): Delete pointer field of
2388 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2389 TYPE_SYMTAB_POINTER.
2390 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2391 (TYPE_SYMTAB_IS_POINTER): Delete.
2392 (TYPE_SYMTAB_IS_DIE): Renumber.
2393 * xcoffout.c: Refer to former sdbout.c file.
2394 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2396 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2397 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2398 (Debugging Options): Delete -gcoff.
2399 (-gstabs): Delete SDB reference.
2401 (-gcoff@var{level}): Delete.
2402 * doc/passes.texi (Debugging information output): Delete SDB and
2403 sdbout.c references.
2404 * doc/tm.texi: Regenerate.
2405 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2406 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2408 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2409 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2410 (SDB_DEBUGGING_INFO): Delete.
2411 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2412 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2413 * target.def (output_source_filename): Delete COFF reference.
2415 * common.opt (gcoff): Delete.
2416 (gxcoff+): Update Negative chain.
2417 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2419 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2420 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2421 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2423 (expand_function_start): Change sdb reference to past tense.
2424 (expand_function_end): Change sdb reference to past tense.
2425 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2426 * opts.c (debug_type_names): Delete coff entry.
2427 (common_handle_option): Delete OPT_gcoff case.
2428 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2430 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2431 * config/cris/cris.h: Delete SDB reference in comment.
2432 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2433 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2434 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2435 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2437 (ix86_expand_prologue): Likewise.
2438 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2439 * config/ia64/ia64.h: Likewise.
2440 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2441 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2443 * config/mmix/mmix.h: Likewise.
2444 * config/nds32/nds32.c: Likewise.
2445 * config/stormy/storym16.h: Likewise.
2446 * config/visium/visium.h: Likewise.
2447 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2449 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2451 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2452 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2453 FRAME_POINTER_REGNUM point at high end of local var area.
2455 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2457 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2458 Move comment around. Do not reset best_edge for a copiable
2459 destination if the copy would cause a partition change.
2460 (better_edge_p): Remove redundant check.
2462 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2464 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2466 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2468 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2469 for math functions that have _Float<N> and _Float<N>X variants.
2470 (mathfn_built_in_2): Add support for math functions that have
2471 _Float<N> and _Float<N>X variants.
2472 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2473 (expand_builtin_mathfn_ternary): Add support for fma with
2474 _Float<N> and _Float<N>X variants.
2475 (expand_builtin): Likewise.
2476 (fold_builtin_3): Likewise.
2477 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2478 create math function _Float<N> and _Float<N>X variants as external
2480 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2481 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2482 the __builtin_ prefix and if not strict ansi, without the prefix.
2483 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2484 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2485 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2486 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2487 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2488 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2489 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2490 function signatures for fma _Float<N> and _Float<N>X variants.
2491 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2492 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2493 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2494 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2495 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2496 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2497 * gencfn-macros.c (print_case_cfn): Add support for math functions
2498 that have _Float<N> and _Float<N>X variants.
2499 (print_define_operator_list): Likewise.
2500 (fltfn_suffixes): Likewise.
2502 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2503 for math functions that have _Float<N> and _Float<N>X variants.
2504 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2505 and _Float<N>X variants.
2506 (COPYSIGN): Likewise.
2509 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2510 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2512 (integer_valued_read_call_p): Likewise.
2513 * fold-const-call.c (fold_const_call_ss): Likewise.
2514 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2515 _Float<N> and _Float<N>X variants.
2516 (fold_const_call_ssss): Add support for fma _Float<N> and
2517 _Float<N>X variants.
2518 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2519 support for copysign and fma _Float<N> and _Float<N>X variants.
2520 (backprop::process_builtin_call_use): Likewise.
2521 * tree-call-cdce.c (can_test_argument_range); Add support for
2522 sqrt _Float<N> and _Float<N>X variants.
2523 (edom_only_function): Likewise.
2524 (get_no_error_domain): Likewise.
2525 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2526 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2527 copysign _Float<N> and _Float<N>X variants.
2528 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2529 handled by machine independent code.
2530 (FMAF128): Likewise.
2531 * doc/cpp.texi (Common Predefined Macros): Document defining
2532 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2533 fma _Float<N> and _Float<N>X variants.
2535 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2538 * config/i386/i386-modes.def (CCFPU): Remove definition.
2539 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2540 (ix86_cc_modes_compatible): Ditto.
2541 (ix86_expand_carry_flag_compare): Ditto.
2542 (ix86_expand_int_movcc): Ditto.
2543 (ix86_expand_int_addcc): Ditto.
2544 (ix86_reverse_condition): Ditto.
2545 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2546 Return true/false for unordered/ordered fp comparisons.
2547 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2548 (ix86_prepare_fp_compare_args): Update for rename.
2549 (ix86_expand_fp_compare): Update for rename. Generate unordered
2550 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2551 (ix86_expand_sse_compare_and_jump): Ditto.
2552 * config/i386/predicates.md (fcmov_comparison_operator):
2553 Remove CCFPU mode handling.
2554 (ix86_comparison_operator): Ditto.
2555 (ix86_carry_flag_operator): Ditto.
2556 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2557 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2558 (*cmpu<mode>_cc_i387): Ditto.
2559 (FPCMP): Remove mode iterator.
2560 (unord): Remove mode attribute.
2561 (unord_subst): New define_subst transformation
2562 (unord): New define_subst attribute.
2564 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2565 (*cmpi<unord>xf_i387): Ditto.
2566 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2567 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2568 using unord_subst transformation.
2569 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2570 (round_saeonly): Also handle CCFP mode.
2571 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2572 Remove UNSPEC_SAHF unspec handling.
2574 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2576 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2578 2017-10-27 Jeff Law <law@redhat.com>
2580 * gimple-ssa-sprintf.c: Include domwalk.h.
2581 (class sprintf_dom_walker): New class, derived from dom_walker.
2582 (sprintf_dom_walker::before_dom_children): New function.
2583 (struct call_info): Moved into sprintf_dom_walker class
2584 (compute_formath_length, handle_gimple_call): Likewise.
2585 (sprintf_length::execute): Call the dominator walker rather
2586 than walking the statements.
2588 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2589 gimple statement locations.
2590 (check_array_bounds): Corresponding changes. Get the statement's
2591 location directly from wi->stmt.
2593 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2596 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2598 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2600 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2601 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2603 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2606 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2607 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2608 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2609 maybe_get_pool_constant.
2610 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2613 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2615 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2616 2.26 caveat. Update gas and gld versions.
2617 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2620 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2622 * cgraph.h (set_malloc_flag): Declare.
2623 * cgraph.c (set_malloc_flag_1): New function.
2624 (set_malloc_flag): Likewise.
2625 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2626 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2628 (read_ipa_call_summary): Add support for reading is_return_callee.
2629 (write_ipa_call_summary): Stream is_return_callee.
2630 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2631 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2632 ipa-prop.h, ipa-fnsummary.h.
2633 (pure_const_names): Change to static.
2634 (malloc_state_e): Define.
2635 (malloc_state_names): Define.
2636 (funct_state_d): Add field malloc_state.
2637 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2638 (check_retval_uses): New function.
2639 (malloc_candidate_p): Likewise.
2640 (analyze_function): Add support for malloc attribute.
2641 (pure_const_write_summary): Stream malloc_state.
2642 (pure_const_read_summary): Add support for reading malloc_state.
2643 (dump_malloc_lattice): New function.
2644 (propagate_malloc): New function.
2645 (warn_function_malloc): New function.
2646 (ipa_pure_const::execute): Call propagate_malloc and
2647 ipa_free_fn_summary.
2648 (pass_local_pure_const::execute): Add support for malloc attribute.
2649 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2650 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2652 2017-10-27 Martin Liska <mliska@suse.cz>
2654 PR gcov-profile/82457
2655 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2656 for fork-like functions to be properly instrumented.
2658 2017-10-27 Richard Biener <rguenther@suse.de>
2661 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2662 info when we redirected EH.
2664 2017-10-26 Michael Collison <michael.collison@arm.com>
2666 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2668 (<optab>_trunchf<GPI:mode>2: New pattern.
2669 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2670 * config/aarch64/iterators.md (wv): New mode attribute.
2671 (vf, VF): New mode attributes.
2672 (vgp, VGP): New mode attributes.
2673 (s): Update attribute with SImode and DImode prefixes.
2675 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2677 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2678 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2679 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2680 (nios2_option_overide): Initialize it. Don't allow R0-relative
2681 addressing with PIC.
2682 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2683 (nios2_symbolic_constant_p): Likewise.
2684 (nios2_legitimate_address_p): Likewise.
2685 (nios2_r0rel_section_name_p): New.
2686 (nios2_symbol_ref_in_r0rel_data_p): New.
2687 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2688 (r0rel_constant_p): New.
2689 (nios2_print_operand_address): Handle r0rel_constant_p.
2690 (nios2_cdx_narrow_form_p): Likewise.
2691 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2692 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2693 (Nios II Options): Document -mr0rel-sec.
2695 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2697 * config/nios2/nios2.c: Include xregex.h.
2698 (nios2_gprel_sec_regex): New.
2699 (nios2_option_overide): Initialize it. Don't allow GP-relative
2700 addressing with PIC.
2701 (nios2_small_section_name_p): Check for regex match.
2702 * config/nios2/nios2.opt (mgprel-sec=): New option.
2703 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2704 (Nios II Options): Document -mgprel-sec.
2706 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2708 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2710 2017-10-26 Tom de Vries <tom@codesourcery.com>
2712 PR tree-optimization/82707
2713 * gimple.c (gimple_copy): Fix unsharing of
2714 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2716 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2718 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2719 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2720 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2721 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2722 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2723 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2724 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2725 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2726 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2727 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2728 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2729 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2730 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2731 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2732 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2733 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2734 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2736 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2738 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2740 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2741 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2742 warning to rs6000.c. Remove the Undocumented flag, since it has
2744 (-mabi=ibmlongdouble): Likewise.
2745 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2746 already set, set the default format for long double.
2747 (rs6000_debug_reg_global): Print whether long double is IBM or
2749 (rs6000_option_override_internal): Rework setting long double
2750 format. Only warn if the user is changing the long double default
2751 and they did not use -Wno-psabi.
2752 * doc/invoke.texi (PowerPC options): Update the documentation for
2753 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2755 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2756 Alan Hayward <alan.hayward@arm.com>
2757 David Sherwood <david.sherwood@arm.com>
2759 * rtl.h (wider_subreg_mode): New function.
2760 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2761 rather than an unsigned int *.
2762 * ira-color.c (regno_max_ref_width): Replace with...
2763 (regno_max_ref_mode): ...this new variable.
2764 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2765 Use wider_subreg_mode.
2766 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2767 rather than an unsigned int *.
2768 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2769 (process_alt_operands): Likewise.
2770 (invariant_p): Likewise.
2771 * lra-spills.c (assign_mem_slot): Likewise.
2772 (add_pseudo_to_slot): Likewise.
2773 * lra.c (collect_non_operand_hard_regs): Likewise.
2774 (add_regs_to_insn_regno_info): Likewise.
2775 * reload1.c (regno_max_ref_width): Replace with...
2776 (regno_max_ref_mode): ...this new variable.
2777 (reload): Update accordingly. Update call to
2778 ira_sort_regnos_for_alter_reg.
2779 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2780 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2781 (scan_paradoxical_subregs): Likewise.
2783 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2785 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2786 (aarch64_frame): Add emit_frame_chain boolean.
2787 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2788 Move eh_return case to aarch64_layout_frame.
2789 (aarch64_layout_frame): Initialize emit_frame_chain.
2790 (aarch64_expand_prologue): Use emit_frame_chain.
2792 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2794 * config/aarch64/aarch64.c (aarch64_layout_frame):
2795 Ensure LR is always stored at the bottom of the callee-saves.
2796 Remove rarely used frame layout which saves callee-saves at top of
2797 frame, so the store of LR can be used as a valid probe in all cases.
2799 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2801 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2802 Improve unaligned TImode/TFmode base/offset split.
2804 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2805 Alan Hayward <alan.hayward@arm.com>
2806 David Sherwood <david.sherwood@arm.com>
2808 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2809 * combine.c (find_single_use_1): Likewise.
2810 (expand_field_assignment): Likewise.
2811 (move_deaths): Likewise.
2812 * lra-constraints.c (simplify_operand_subreg): Likewise.
2813 (curr_insn_transform): Likewise.
2814 * lra.c (collect_non_operand_hard_regs): Likewise.
2815 (add_regs_to_insn_regno_info): Likewise.
2816 * rtlanal.c (reg_referenced_p): Likewise.
2817 (covers_regno_no_parallel_p): Likewise.
2819 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2821 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2822 Don't print any bits outside the precision of the value.
2823 * wide-int.cc (test_printing): Add some new tests.
2825 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2827 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2828 supports -xbrace_comment option.
2829 * configure: Regenerate.
2830 * config.in: Regenerate.
2831 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2832 (ASM_CPU_SPEC): Use it.
2834 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2836 * target.def (static_rtx_alignment): New hook.
2837 * targhooks.h (default_static_rtx_alignment): Declare.
2838 * targhooks.c (default_static_rtx_alignment): New function.
2839 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2840 * doc/tm.texi: Regenerate.
2841 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2842 instead of targetm.constant_alignment. Remove call to
2844 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2845 (cris_preferred_mininum_alignment): New function, split out from...
2846 (cris_constant_alignment): ...here.
2847 (cris_static_rtx_alignment): New function.
2848 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2850 (ix86_constant_alignment): ...here.
2851 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2852 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2853 (mmix_static_rtx_alignment): New function.
2854 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2855 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2857 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2860 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2861 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2863 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2865 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2866 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2867 mark as a sign-extending load.
2868 (local_pic_load_u): Define.
2870 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2873 * fold-const.c (operand_equal_for_comparison_p): Also return true
2874 if ARG0 is a simple variant of ARG1 with narrower precision.
2875 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2877 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2879 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2881 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2882 scatter_static, scatter_per_elt.
2883 * x86-tune-costs.h: Add new cost entries.
2885 2017-10-25 Richard Biener <rguenther@suse.de>
2887 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2888 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2889 class pass_fre): Move to ...
2890 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2891 class pass_fre): ... here and adjust for statistics.
2893 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2896 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2897 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2898 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2899 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2902 2017-10-25 Richard Biener <rguenther@suse.de>
2904 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2905 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2906 eliminate_push_avail, eliminate_insert): Move inside...
2907 (class eliminate_dom_walker): ... this class in preparation
2909 (fini_eliminate): Remove by merging with ...
2910 (eliminate): ... this function. Adjust for class changes.
2911 (pass_pre::execute): Remove fini_eliminate call.
2912 (pass_fre::execute): Likewise.
2914 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2917 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2918 (VPERMI2, VPERMI2I): New mode iterators.
2919 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2920 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2922 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2923 mode iterator. Remove 3 old define_insn patterns.
2924 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2925 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2926 VPERMI2 mode iterator, remove the other two expanders.
2927 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2928 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2929 instructions, remove the other two patterns.
2930 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2931 mode iterator, remove the other two patterns.
2932 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2933 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2934 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2935 and adjust argument order accordingly.
2936 (ix86_expand_vec_perm): Adjust caller.
2937 (expand_vec_perm_1): Likewise.
2938 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2939 (expand_vec_perm_vpermt2_vpshub2): ... this.
2940 (ix86_expand_vec_perm_const_1): Adjust caller.
2941 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2944 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2945 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2946 (vec_shl_<mode>): Remove unused expander.
2947 (avx512bw_<shift_insn><mode>3): New define_insn.
2948 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2949 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2951 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2954 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2957 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2959 PR rtl-optimization/82396
2960 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2961 (autopref_multipass_init): Simplify initialization.
2962 (autopref_rank_data): Simplify sort order.
2963 * gcc/sched-int.h (autopref_multipass_data_): Remove
2964 multi_mem_insn_p, min_offset and max_offset.
2966 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2969 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2970 Check special value of flag_omit_frame_pointer.
2971 (aarch64_can_eliminate): Likewise.
2972 (aarch64_override_options_after_change_1): Simplify handling of
2973 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2975 2017-10-24 Richard Biener <rguenther@suse.de>
2977 PR tree-optimization/82697
2978 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2979 zero for conditional load and unconditional store.
2981 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2983 * doc/install.texi: Document bootstrap-cet.
2985 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2988 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2989 ENDBR instruction at function entrance if function is only
2992 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2995 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2996 patterns to better describe from which operation the CF is computed.
2997 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2998 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2999 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3000 is 0, use _0 suffixed expanders instead of emitting a comparison
3003 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3005 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3006 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3008 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3011 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3012 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3013 * loop-iv.c (iv_get_reaching_def): Likewise.
3014 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3015 variable is promoted and the partition contains undefined values.
3017 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3019 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3021 (nios2_address_cost): Define.
3022 (nios2_legitimize_address): Recognize (exp + constant) directly.
3023 (TARGET_ADDRESS_COST): Define.
3025 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3027 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3028 (nios2_symbolic_memory_operand_p): Declare.
3029 (nios2_split_large_constant): Declare.
3030 (nios2_split_symbolic_memory_operand): Declare.
3031 * config/nios2/nios2.c: Adjust includes.
3032 (nios2_symbolic_constant_allowed): New.
3033 (nios2_symbolic_constant_p): New.
3034 (nios2_plus_symbolic_constant_p): New.
3035 (nios2_valid_addr_expr_p): Recognize addresses involving
3037 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3038 (nios2_symbolic_memory_operand_p): New.
3039 (nios2_large_constant_p): New.
3040 (nios2_split_large_constant): New.
3041 (nios2_split_plus_large_constant): New.
3042 (nios2_split_symbolic_memory_operand): New.
3043 (nios2_legitimize_address): Code refactoring. Handle addresses
3044 involving symbolic constants.
3045 (nios2_emit_move_sequence): Likewise.
3046 (nios2_print_operand): Improve error output.
3047 (nios2_print_operand_address): Handle LO_SUM.
3048 (nios2_cdx_narrow_form_p): Likewise.
3049 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3050 operands involving symbolic constants.
3051 (movhi_internal, movsi_internal): Likewise.
3052 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3053 (extendhisi2, extendqi<mode>2): Likewise.
3055 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3057 * tree-pass.h (PROP_rtl_split_insns): Define.
3058 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3060 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3062 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3064 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3067 * target.def (const_not_ok_for_debug_p): Default to
3068 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3069 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3070 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3071 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3072 which targetm.const_not_ok_for_debug_p returned true.
3073 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3075 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3077 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3078 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3079 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3081 (ix86_const_not_ok_for_debug_p): New function.
3082 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3083 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3085 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3088 * system.h: Conditionally include "unique-ptr.h" if
3089 INCLUDE_UNIQUE_PTR is defined.
3090 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3091 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3093 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3095 * config/rl78/rl78.md: New define_expand "subdi3".
3097 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3100 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3101 DF_REF_INSN if DF_REF_INSN_INFO is false.
3103 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3105 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3106 xmm_move instead of sse_move.
3107 (sse_store_index): New function.
3108 (ix86_register_move_cost): Be more sensible about mismatch stall;
3109 model AVX moves correctly; make difference between sse->integer and
3111 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3112 moves; make difference between SSE and AVX.
3113 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3114 and zmm_move. Increase size of sse load and store tables;
3115 add unaligned load and store tables; add ssemmx_to_integer.
3116 * x86-tune-costs.h: Update all entries according to real
3117 move latencies from Agner Fog's manual and chip documentation.
3119 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3122 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3123 * config/i386/constraints.md (Wf): New constraint.
3124 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3125 (cmp<dwi>_doubleword): Removed.
3126 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3127 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3128 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3129 expand with cmp<dwi>_doubleword. For LTU and GEU use
3130 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3132 * common.opt (gcolumn-info): Enable by default.
3133 * doc/invoke.texi (gcolumn-info): Document new default.
3135 2017-10-23 Richard Biener <rguenther@suse.de>
3137 PR tree-optimization/82672
3138 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3139 Fold the stmt if we propagated into it.
3141 2017-10-23 Richard Biener <rguenther@suse.de>
3143 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3144 (bitmap_remove_expr_from_set): ... this. All callers call this
3145 for non-constant values.
3146 (bitmap_set_subtract): Rename to...
3147 (bitmap_set_subtract_expressions): ... this. Adjust and
3149 (bitmap_set_contains_value): Remove superfluous check.
3150 (bitmap_set_replace_value): Inline into single caller ...
3151 (bitmap_value_replace_in_set): ... here and simplify.
3152 (dependent_clean): Merge into ...
3153 (clean): ... this using an overload. Adjust.
3154 (prune_clobbered_mems): Adjust.
3155 (compute_antic_aux): Likewise.
3156 (compute_partial_antic_aux): Likewise.
3158 2017-10-23 Richard Biener <rguenther@suse.de>
3160 PR tree-optimization/82129
3162 2017-08-01 Richard Biener <rguenther@suse.de>
3164 PR tree-optimization/81181
3165 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3166 (compute_antic): ... end of iteration here.
3168 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3170 * target.def (starting_frame_offset): New hook.
3171 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3172 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3173 * doc/tm.texi.in: Regenerate.
3174 * hooks.h (hook_hwi_void_0): Declare.
3175 * hooks.c (hook_hwi_void_0): New function.
3176 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3177 STARTING_FRAME_OFFSET.
3178 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3179 * reload1.c (reload): Likewise.
3180 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3181 instead of STARTING_FRAME_OFFSET.
3182 * function.c (try_fit_stack_local): Likewise.
3183 (assign_stack_local_1): Likewise
3184 (instantiate_virtual_regs): Likewise.
3185 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3186 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3187 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3188 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3189 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3190 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3191 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3192 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3193 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3194 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3195 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3196 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3197 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3198 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3199 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3200 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3201 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3202 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3203 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3204 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3205 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3206 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3207 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3208 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3209 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3210 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3211 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3212 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3213 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3214 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3215 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3216 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3217 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3218 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3219 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3220 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3221 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3222 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3223 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3224 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3225 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3226 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3227 return a HOST_WIDE_INT.
3228 (avr_builtin_setjmp_frame_value): Use it instead of
3229 STARTING_FRAME_OFFSET.
3230 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3231 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3232 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3234 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3235 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3236 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3237 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3238 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3239 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3240 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3241 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3242 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3243 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3244 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3245 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3247 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3248 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3249 * config/mips/mips.c (mips_compute_frame_info): Refer to
3250 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3251 (mips_starting_frame_offset): New function.
3252 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3253 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3254 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3255 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3256 and return a HOST_WIDE_INT.
3257 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3258 (mmix_initial_elimination_offset): Refer to
3259 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3260 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3261 * config/pa/pa.c (pa_starting_frame_offset): New function.
3262 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3263 (pa_expand_prologue): Likewise.
3264 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3265 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3266 !FRAME_GROWS_DOWNWARD handling to...
3267 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3268 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3269 !FRAME_GROWS_DOWNWARD handling to...
3270 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3271 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3272 !FRAME_GROWS_DOWNWARD handling to...
3273 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3274 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3276 (rs6000_starting_frame_offset): New function.
3277 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3278 !FRAME_GROWS_DOWNWARD handling to...
3279 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3280 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3281 !FRAME_GROWS_DOWNWARD handling to...
3282 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3283 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3284 !FRAME_GROWS_DOWNWARD handling to...
3285 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3286 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3287 (rs6000_starting_frame_offset): New function.
3288 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3289 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3290 * config/vax/vax.c (vax_starting_frame_offset): New function.
3291 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3292 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3293 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3294 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3295 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3296 * system.h (STARTING_FRAME_OFFSET): Poison.
3298 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3300 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3301 SCALAR_TYPE_MODE instead of TYPE_MODE.
3303 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3304 Alan Hayward <alan.hayward@arm.com>
3305 David Sherwood <david.sherwood@arm.com>
3307 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3309 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3310 Alan Hayward <alan.hayward@arm.com>
3311 David Sherwood <david.sherwood@arm.com>
3313 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3315 2017-10-23 Richard Biener <rguenther@suse.de>
3317 PR tree-optimization/82129
3318 * tree-ssa-pre.c (bitmap_set_and): Remove.
3319 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3320 canonicalizing expressions in the set to those with lowest
3321 ID rather than taking that from the first edge.
3323 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3325 * combine.c (rtx_equal_for_field_assignment_p): Use
3326 byte_lowpart_offset.
3328 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3329 Alan Hayward <alan.hayward@arm.com>
3330 David Sherwood <david.sherwood@arm.com>
3332 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3333 to a SUBREG_PROMOTED_VAR.
3335 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3336 Alan Hayward <alan.hayward@arm.com>
3337 David Sherwood <david.sherwood@arm.com>
3339 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3340 (expand_debug_source_expr): Likewise.
3341 * combine.c (combine_simplify_rtx): Likewise.
3342 * cse.c (fold_rtx): Likewise.
3343 * optabs.c (expand_float): Likewise.
3344 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3345 (simplify_binary_operation_1): Likewise.
3347 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3348 Alan Hayward <alan.hayward@arm.com>
3349 David Sherwood <david.sherwood@arm.com>
3351 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3352 (record_promoted_value): Likewise.
3353 * expr.c (expand_expr_real_2): Likewise.
3354 * ree.c (update_reg_equal_equiv_notes): Likewise.
3355 (combine_set_extension): Likewise.
3356 * rtlanal.c (low_bitmask_len): Likewise.
3357 * simplify-rtx.c (neg_const_int): Likewise.
3358 (simplify_binary_operation_1): Likewise.
3360 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3361 Alan Hayward <alan.hayward@arm.com>
3362 David Sherwood <david.sherwood@arm.com>
3364 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3365 * regcprop.c (maybe_mode_change): Likewise.
3366 * reload1.c (alter_reg): Likewise.
3368 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3370 * inchash.h (inchash::hash::add_wide_int): New function.
3371 * lto-streamer-out.c (hash_tree): Use it.
3373 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3375 * inchash.h (inchash::hash::add_wide_int): Rename to...
3376 (inchash::hash::add_hwi): ...this.
3377 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3378 (polymorphic_call_target_hasher::hash): Likewise.
3379 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3380 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3381 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3382 * lto-streamer-out.c (hash_tree): Likewise.
3383 * optc-save-gen.awk: Likewise.
3384 * tree.c (add_expr): Likewise.
3386 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3389 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3390 for ordered inequality comparisons even with TARGET_IEEE_FP.
3392 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3395 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3396 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3397 Expand with cmp<dwi>_doubleword.
3399 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3401 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3402 List CET intrinsics.
3403 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3404 specific to -fcf-protection option.
3406 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3408 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3409 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3410 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3411 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3412 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3413 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3414 (extra_objs): Add cet.o for Linux/x86 targets.
3415 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3416 * config/i386/cet.c: New file.
3417 * config/i386/cetintrin.h: Likewise.
3418 * config/i386/t-cet: Likewise.
3419 * config/i386/cpuid.h (bit_SHSTK): New.
3420 (bit_IBT): Likewise.
3421 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3422 pass IBT and SHSTK bits.
3423 * config/i386/i386-builtin-types.def
3424 (VOID_FTYPE_UNSIGNED_PVOID): New.
3425 (VOID_FTYPE_UINT64_PVOID): Likewise.
3426 * config/i386/i386-builtin.def: Add CET intrinsics.
3427 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3428 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3429 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3430 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3432 * config/i386/i386.c (rest_of_insert_endbranch): New.
3433 (pass_data_insert_endbranch): Likewise.
3434 (pass_insert_endbranch): Likewise.
3435 (make_pass_insert_endbranch): Likewise.
3436 (ix86_notrack_prefixed_insn_p): Likewise.
3437 (ix86_target_string): Add -mibt, -mshstk flags.
3438 (ix86_option_override_internal): Add flag_cf_protection
3440 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3441 (ix86_print_operand): Add 'notrack' prefix output.
3442 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3443 (ix86_expand_builtin): Expand CET intrinsics.
3444 (x86_output_mi_thunk): Add 'endbranch' instruction.
3445 * config/i386/i386.h (TARGET_IBT): New.
3446 (TARGET_IBT_P): Likewise.
3447 (TARGET_SHSTK): Likewise.
3448 (TARGET_SHSTK_P): Likewise.
3449 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3450 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3451 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3452 (builtin_setjmp_setup): New pattern.
3453 (builtin_longjmp): Likewise.
3454 (rdssp<mode>): Likewise.
3455 (incssp<mode>): Likewise.
3456 (saveprevssp): Likewise.
3457 (rstorssp): Likewise.
3458 (wrss<mode>): Likewise.
3459 (wruss<mode>): Likewise.
3460 (setssbsy): Likewise.
3461 (clrssbsy): Likewise.
3462 (nop_endbr): Likewise.
3463 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3465 * config/i386/immintrin.h: Include <cetintrin.h>.
3466 * config/i386/linux-common.h
3467 (file_end_indicate_exec_stack_and_cet): New prototype.
3468 (TARGET_ASM_FILE_END): New.
3470 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3472 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3473 latencies instead of having separate table; make difference between
3474 integer and float costs.
3475 * i386.h (processor_costs): Remove scalar_stmt_cost,
3476 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3477 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3479 * x86-tune-costs.h: Remove entries which has been removed in
3480 procesor_costs from all tables; make cond_taken_branch_cost
3481 and cond_not_taken_branch_cost COST_N_INSNS based.
3483 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3485 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3487 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3489 * config/i386/i386.md (isa): Remove fma_avx512f.
3490 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3491 <avx512>_fmadd_<mode>_mask3<round_name>,
3492 <avx512>_fmsub_<mode>_mask<round_name>,
3493 <avx512>_fmsub_<mode>_mask3<round_name>,
3494 <avx512>_fnmadd_<mode>_mask<round_name>,
3495 <avx512>_fnmadd_<mode>_mask3<round_name>,
3496 <avx512>_fnmsub_<mode>_mask<round_name>,
3497 <avx512>_fnmsub_<mode>_mask3<round_name>,
3498 <avx512>_fmaddsub_<mode>_mask<round_name>,
3499 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3500 <avx512>_fmsubadd_<mode>_mask<round_name>,
3501 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3502 (*vec_widen_umult_even_v16si<mask_name>,
3503 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3504 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3506 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3508 * extend.texi: Add 'nocf_check' documentation.
3509 * gimple.texi: Add second parameter to
3510 gimple_build_call_from_tree.
3511 * invoke.texi: Add -fcf-protection documentation.
3512 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3514 2017-10-20 Richard Biener <rguenther@suse.de>
3516 PR tree-optimization/82473
3517 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3518 the largest input type.
3520 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3522 * c-attribs.c (handle_nocf_check_attribute): New function.
3523 (c_common_attribute_table): Add 'nocf_check' handling.
3524 * gimple-parser.c: Add second argument NULL to
3525 gimple_build_call_from_tree.
3526 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3527 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3529 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3530 * common.opt: Add fcf-protection flag.
3531 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3532 * flag-types.h: Add enum cf_protection_level.
3533 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3534 Add 'nocf_check' attribute propagation to gimple call.
3535 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3536 (gimple_build_call_from_tree): Update prototype.
3537 (gimple_call_nocf_check_p): New function.
3538 (gimple_call_set_nocf_check): Likewise.
3539 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3540 * ipa-icf.c: Add nocf_check attribute in statement hash.
3541 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3542 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3543 * toplev.c (process_options): Add flag_cf_protection handling.
3545 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3547 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3549 2017-10-20 Richard Biener <rguenther@suse.de>
3551 PR tree-optimization/82603
3552 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3553 remove false predicated stores.
3555 2017-10-20 Richard Biener <rguenther@suse.de>
3557 * graphite-isl-ast-to-gimple.c
3558 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3559 Remove return value and simplify, dump copied stmt after lhs
3561 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3562 Reduce dump verbosity.
3563 (gsi_insert_earliest): Likewise.
3564 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3565 * graphite.c (print_global_statistics): Adjust dumping.
3566 (print_graphite_scop_statistics): Likewise.
3567 (print_graphite_statistics): Do not dump loops here.
3568 (graphite_transform_loops): But here.
3570 2017-10-20 Nicolas Roche <roche@adacore.com>
3572 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3573 * configure: Regenerate.
3575 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3578 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3579 functions when optimizing replace GIMPLE_RETURN stmts with
3580 calls to __builtin_unreachable ().
3583 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3584 for -fsanitize=thread link of executables.
3585 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3586 link of executables.
3589 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3591 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3592 define_insns for logical vector shifts to use VI248_AVX512BW
3593 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3594 condition, useless isa and prefix attributes. Change the first
3595 2 of these define_insns to ...
3596 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3597 define_insn for avx512vl.
3598 (<shift_insn><mode>3): ... and this, new define_insn without
3599 masking for non-avx512vl.
3602 * config/i386/sse.md (*andnot<mode>3,
3603 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3604 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3605 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3606 not applied use empty suffix even for TARGET_AVX512VL.
3607 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3608 is applied, supply evex,evex or evex,evex,evex instead of just
3611 2017-10-20 Julia Koval <julia.koval@intel.com>
3613 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3614 (OPTION_MASK_ISA_GFNI_UNSET): New.
3615 (ix86_handle_option): Handle OPT_mgfni.
3616 * config/i386/cpuid.h (bit_GFNI): New.
3617 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3618 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3619 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3620 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3621 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3622 * config/i386/i386.opt: Add mgfni.
3624 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3626 * config/msp430/msp430.c (msp430_option_override): Disable
3627 -fdelete-null-pointer-checks.
3628 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3630 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3632 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3633 of x87 and SSE instructions.
3635 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3637 * asan.c (create_cond_insert_point): Do not update edge count.
3638 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3639 (afdo_propagate_circuit): Likewise.
3640 (afdo_calculate_branch_prob): Likewise.
3641 (afdo_annotate_cfg): Likewise.
3642 * basic-block.h (struct edge_def): Remove count.
3643 (edge_def::count): New accessor.
3644 * bb-reorder.c (rotate_loop): Update.
3645 (find_traces_1_round): Update.
3646 (connect_traces): Update.
3647 (sanitize_hot_paths): Update.
3648 * cfg.c (unchecked_make_edge): Update.
3649 (make_single_succ_edge): Update.
3650 (check_bb_profile): Update.
3651 (dump_edge_info): Update.
3652 (update_bb_profile_for_threading): Update.
3653 (scale_bbs_frequencies_int): Update.
3654 (scale_bbs_frequencies_gcov_type): Update.
3655 (scale_bbs_frequencies_profile_count): Update.
3656 (scale_bbs_frequencies): Update.
3657 * cfganal.c (connect_infinite_loops_to_exit): Update.
3658 * cfgbuild.c (compute_outgoing_frequencies): Update.
3659 (find_many_sub_basic_blocks): Update.
3660 * cfgcleanup.c (try_forward_edges): Update.
3661 (try_crossjump_to_edge): Update
3662 * cfgexpand.c (expand_gimple_cond): Update
3663 (expand_gimple_tailcall): Update
3664 (construct_exit_block): Update
3665 * cfghooks.c (verify_flow_info): Update
3666 (redirect_edge_succ_nodup): Update
3667 (split_edge): Update
3668 (make_forwarder_block): Update
3669 (duplicate_block): Update
3670 (account_profile_record): Update
3671 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3672 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3673 * cfgloopmanip.c (scale_loop_profile): Update.
3675 (lv_adjust_loop_entry_edge): Update.
3676 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3677 (force_nonfallthru_and_redirect): Update.
3678 (purge_dead_edges): Update.
3679 (rtl_flow_call_edges_add): Update.
3680 * cgraphunit.c (init_lowered_empty_function): Update.
3681 (cgraph_node::expand_thunk): Update.
3682 * gimple-pretty-print.c (dump_probability): Update.
3683 (dump_edge_probability): Update.
3684 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3685 * haifa-sched.c (sched_create_recovery_edges): Update.
3686 * hsa-gen.c (convert_switch_statements): Update.
3687 * ifcvt.c (dead_or_predicable): Update.
3688 * ipa-inline-transform.c (inline_transform): Update.
3689 * ipa-split.c (split_function): Update.
3690 * ipa-utils.c (ipa_merge_profiles): Update.
3691 * loop-doloop.c (add_test): Update.
3692 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3693 * lto-streamer-in.c (input_cfg): Update.
3694 (input_function): Update.
3695 * lto-streamer-out.c (output_cfg): Update.
3696 * modulo-sched.c (sms_schedule): Update.
3697 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3698 * predict.c (maybe_hot_edge_p): Update.
3699 (unlikely_executed_edge_p): Update.
3700 (probably_never_executed_edge_p): Update.
3701 (dump_prediction): Update.
3702 (drop_profile): Update.
3703 (propagate_unlikely_bbs_forward): Update.
3704 (determine_unlikely_bbs): Update.
3705 (force_edge_cold): Update.
3706 * profile.c (compute_branch_probabilities): Update.
3707 * reg-stack.c (better_edge): Update.
3708 * shrink-wrap.c (handle_simple_exit): Update.
3709 * tracer.c (better_p): Update.
3710 * trans-mem.c (expand_transaction): Update.
3711 (split_bb_make_tm_edge): Update.
3712 * tree-call-cdce.c: Update.
3713 * tree-cfg.c (gimple_find_sub_bbs): Update.
3714 (gimple_split_edge): Update.
3715 (gimple_duplicate_sese_region): Update.
3716 (gimple_duplicate_sese_tail): Update.
3717 (gimple_flow_call_edges_add): Update.
3718 (insert_cond_bb): Update.
3719 (execute_fixup_cfg): Update.
3720 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3721 * tree-complex.c (expand_complex_div_wide): Update.
3722 * tree-eh.c (lower_resx): Update.
3723 (unsplit_eh): Update.
3724 (cleanup_empty_eh_move_lp): Update.
3725 * tree-inline.c (copy_edges_for_bb): Update.
3726 (freqs_to_counts): Update.
3727 (copy_cfg_body): Update.
3728 * tree-ssa-dce.c (remove_dead_stmt): Update.
3729 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3730 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3731 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3732 (unloop_loops): Update.
3733 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3734 * tree-ssa-loop-split.c (connect_loops): Update.
3735 (split_loop): Update.
3736 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3737 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3738 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3739 * tree-ssa-reassoc.c (branch_fixup): Update.
3740 * tree-ssa-tail-merge.c (replace_block_by): Update.
3741 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3742 (compute_path_counts): Update.
3743 (update_profile): Update.
3744 (recompute_probabilities): Update.
3745 (update_joiner_offpath_counts): Update.
3746 (estimated_freqs_path): Update.
3747 (freqs_to_counts_path): Update.
3748 (clear_counts_path): Update.
3749 (ssa_fix_duplicate_block_edges): Update.
3750 (duplicate_thread_path): Update.
3751 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3752 (case_bit_test_cmp): Update.
3753 (collect_switch_conv_info): Update.
3754 (gen_inbound_check): Update.
3755 (do_jump_if_equal): Update.
3756 (emit_cmp_and_jump_insns): Update.
3757 * tree-tailcall.c (decrease_profile): Update.
3758 (eliminate_tail_call): Update.
3759 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3760 (vect_do_peeling): Update.
3761 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3762 * ubsan.c (ubsan_expand_null_ifn): Update.
3763 (ubsan_expand_ptr_ifn): Update.
3764 * value-prof.c (gimple_divmod_fixed_value): Update.
3765 (gimple_mod_pow2): Update.
3766 (gimple_mod_subtract): Update.
3767 (gimple_ic): Update.
3768 (gimple_stringop_fixed_value): Update.
3770 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3773 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3775 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3777 PR rtl-optimization/82395
3778 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3779 based on non_spilled_static_chain_regno_p.
3781 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3783 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3784 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3785 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3787 2017-10-19 Martin Sebor <msebor@redhat.com>
3789 PR tree-optimization/82596
3790 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3792 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3794 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3795 (get_mem_refs_of_builtin_call): Likewise.
3796 * builtins.c (expand_builtin_apply): Adjust call to
3797 allocate_dynamic_stack_space.
3798 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3799 the third argument to allocate_dynamic_stack_space, otherwise -1.
3800 (expand_builtin): Deal with all alloca variants.
3801 (is_inexpensive_builtin): Likewise.
3802 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3803 * calls.c (special_function_p): Deal with all alloca variants.
3804 (initialize_argument_information): Adjust call to
3805 allocate_dynamic_stack_space.
3806 (expand_call): Likewise.
3807 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3808 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3809 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3810 use it for the stack usage computation.
3811 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3812 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3813 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3814 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3815 (in_loop_p): Remove first argument and useless check.
3816 (pass_walloca::execute): Remove useless test and adjust call to above.
3817 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3818 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3819 (gimplify_call_expr): Deal with all alloca variants.
3820 * hsa-gen.c (gen_hsa_alloca): Likewise.
3821 (gen_hsa_insns_for_call): Likewise.
3822 * ipa-pure-const.c (special_builtin_state): Likewise.
3823 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3824 * tree-object-size.c (alloc_object_size): Likewise.
3825 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3826 (call_may_clobber_ref_p_1): Likewise.
3827 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3828 (ccp_fold_stmt): Likewise.
3829 (optimize_stack_restore): Likewise.
3830 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3831 (mark_all_reaching_defs_necessary_1): Likewise.
3832 (propagate_necessity): Likewise.
3833 (eliminate_unnecessary_stmts): Likewise.
3834 * tree.c (build_common_builtin_nodes): Build
3835 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3836 (build_alloca_call_expr): New function.
3837 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3838 (CASE_BUILT_IN_ALLOCA): Likewise.
3839 (build_alloca_call_expr): Declare.
3840 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3842 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3845 * dwarf2out.c (new_die_raw): New static inline function.
3846 (new_die): Use it to create the DIE.
3847 (add_AT_external_die_ref): Likewise.
3848 (clone_die): Likewise.
3849 (clone_as_declaration): Likewise.
3850 (dwarf2out_vms_debug_main_pointer): Likewise.
3851 (base_type_die): Likewise. Remove early return for corner cases.
3852 Do not call add_pubtype on the DIE here.
3853 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3854 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3855 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3856 native order exists for base types, attach the DIE manually and call
3857 add_pubtype on it. Do not equate a reverse order DIE to the type.
3859 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3861 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3862 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3864 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3866 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3868 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3869 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3870 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3871 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3872 * opts.c (sanitizer_opts): Add builtin.
3873 * ubsan.c (instrument_builtin): New function.
3874 (pass_ubsan::execute): Call it.
3875 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3876 * doc/invoke.texi: Document -fsanitize=builtin.
3878 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3879 builtins, store max (log2 (align), 0) into uchar field instead of
3880 align into uptr field.
3881 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3882 store uchar 0 field instead of uptr 0 field.
3883 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3884 instead of passing one address of struct with 2 locations pass
3885 two addresses of structs with 1 location each.
3886 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3887 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3888 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3889 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3890 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3891 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3892 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3893 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3895 2017-10-19 Martin Liska <mliska@suse.cz>
3898 * file-find.c (remove_prefix): Remove.
3899 * file-find.h (remove_prefix): Likewise.
3900 * gcc-ar.c: Remove smartness of lookup.
3902 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3904 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3905 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3906 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3908 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3911 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3912 (setcc + and to xor + setcc): New peephole2.
3914 2017-10-19 Tom de Vries <tom@codesourcery.com>
3916 * doc/sourcebuild.texi (Test Directives, Variants of
3917 dg-require-support): Add dg-require-stack-size.
3919 2017-10-19 Martin Liska <mliska@suse.cz>
3922 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3923 that have a large alignment.
3924 (gimplify_target_expr): Likewise.
3926 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3928 PR rtl-optimization/82602
3929 * ira.c (rtx_moveable_p): Return false for volatile asm.
3931 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3934 * config/i386/i386-modes.def (CCGZ): New CC mode.
3935 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3936 * config/i386/predicates.md (ix86_comparison_operator):
3938 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3939 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3940 with double-word subtraction.
3941 (put_condition_code): Handle CCGZmode.
3943 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3945 * wide-int.cc (debug (const wide_int &)): New.
3946 (debug (const wide_int *)): New.
3947 (debug (const widest_int &)): New.
3948 (debug (const widest_int *)): New.
3950 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3953 * lra-constraints.c (curr_insn_transform): Use non-input operand
3954 instead of output one for matched reload.
3956 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3958 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3959 (tree-ssa-loop-ivopts.h): New header file.
3960 (struct builtin_info): New fields.
3961 (classify_builtin_1): Compute and record base and offset parts for
3962 memset builtin partition by calling strip_offset.
3963 (offset_cmp, fuse_memset_builtins): New functions.
3964 (finalize_partitions): Fuse adjacent memset partitions by calling
3966 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3967 Expose the interface.
3968 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3970 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3972 PR tree-optimization/82574
3973 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3974 that data reference must be executed exactly once per iteration
3975 against the outermost loop in nest.
3976 (classify_partition): Update call to above function.
3978 2017-10-18 Richard Biener <rguenther@suse.de>
3980 PR tree-optimization/82591
3981 * graphite.c (graphite_transform_loops): Move code gen message
3983 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3984 Here. Handle scop_to_isl_ast failing.
3985 (scop_to_isl_ast): Limit the number of ISL operations.
3987 2017-10-18 Richard Biener <rguenther@suse.de>
3989 * graphite-isl-ast-to-gimple.c
3990 (translate_isl_ast_to_gimple::set_rename): Simplify.
3991 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3992 (graphite_copy_stmts_from_block): ... here.
3993 (copy_bb_and_scalar_dependences): Simplify.
3994 (add_parameters_to_ivs_params): Canonicalize.
3995 (generate_entry_out_of_ssa_copies): Simplify.
3996 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3997 by passing in ISL dimension.
3998 (parameter_index_in_region_1): Rename to ...
3999 (parameter_index_in_region): ... this.
4000 (extract_affine): Adjust assert, pass down parameter index.
4001 (add_param_constraints): Use range-info when available.
4002 (build_scop_context): Adjust.
4003 * sese.c (new_sese_info): Adjust.
4004 (free_sese_info): Likewise.
4005 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4006 Remove unused typedefs.
4007 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4009 2017-10-18 Martin Liska <mliska@suse.cz>
4011 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4013 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4015 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4016 (s390_sched_init): Do not reset s390_sched_state if we entered the
4017 current basic block via a fallthru edge and all others are unlikely.
4019 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4021 * config/s390/s390.c (NUM_SIDES): New variable.
4022 (LONGRUNNING_THRESHOLD): New variable.
4023 (LATENCY_FACTOR): New variable.
4024 (s390_sched_score): Decrease score for long-running instructions on
4026 (s390_sched_variable_issue): Perform bookkeeping for long-running
4029 2017-10-18 Richard Biener <rguenther@suse.de>
4031 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4032 Simplify with removal of the parameter rename map.
4033 (set_rename): Likewise.
4034 (should_copy_to_new_region): Likewise.
4035 (graphite_copy_stmts_from_block): Likewise.
4036 (copy_bb_and_scalar_dependences): Remove initialization of
4037 unused copied_bb_map.
4039 (copy_internal_parameters): Likewise.
4040 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4041 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4042 Use INTEGRAL_TYPE_P.
4043 (parameter_index_in_region_1): Rename to ...
4044 (assign_parameter_index_in_region): ... this. Assert we have
4045 a parameter we handle.
4046 (scan_tree_for_params): Adjust.
4047 * sese.h (parameter_rename_map_t): Remove.
4048 (struct sese_info_t): Remove unused parameter_rename_map and
4049 copied_bb_map members.
4050 * sese.c (new_sese_info): Adjust.
4051 (free_sese_info): Likewise.
4053 2017-10-18 Martin Liska <mliska@suse.cz>
4056 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4057 on an abnormal edge.
4059 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4061 * doc/invoke.texi (ffunction-sections and fdata-sections):
4064 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4066 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4067 the use statement can throw internally.
4069 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4071 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4072 any RTX present on the RHS of a SET.
4073 * compare-elim.c (try_eliminate_compare): Restore comment.
4075 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4077 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4078 may be also called on tcc_type nodes.
4079 * langhooks.c (lhd_tree_size): Likewise.
4081 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4083 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4084 format_warning_at_substring.
4085 (maybe_warn): Convert source_range * param to a location_t. Pass
4086 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4087 (format_directive): Remove code to extract source_ranges and
4088 source_range * in favor of just a location_t.
4089 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4091 * substring-locations.c (format_warning_va): Convert
4092 source_range * param to a location_t.
4093 (format_warning_at_substring): Likewise.
4094 * substring-locations.h (format_warning_va): Likewise.
4095 (format_warning_at_substring): Likewise.
4097 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4099 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4101 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4102 and scatter/gather ops.
4104 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4105 vec_gather_load and vec_scatter_store.
4106 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4107 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4108 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4109 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4110 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4111 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4113 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4115 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4116 Detect FTST instruction and handle its register pops. Only pop
4117 second operand if can_pop_second_op is true.
4118 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4119 set can_pop_second_op to false in the compare_for_stack_reg call.
4121 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4122 output_fp_compare for stack register operands.
4123 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4124 instructions here. Do not emit stack register pops here. Assert
4125 that FCOMPP pops next to top stack register. Rewrite function.
4127 2017-10-17 Nathan Sidwell <nathan@acm.org>
4130 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4131 use DECL_ASSEMBLER_NAME_RAW.
4134 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4137 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4138 Wilco Dijkstra <wilco.dijkstra@arm.com>
4140 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4141 converstion to Pmode from the buf_addr.
4143 2017-10-17 Richard Biener <rguenther@suse.de>
4145 * graphite-dependences.c (scop_get_reads_and_writes): Change
4146 output parameters to references.
4148 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4150 PR 71026/tree-optimization
4151 * fold-const.c (distribute_real_division): Removed.
4152 (fold_binary_loc): Remove calls to distribute_real_divison.
4154 2017-10-17 Richard Biener <rguenther@suse.de>
4156 * graphite-scop-detection.c
4157 (scop_detection::stmt_has_simple_data_refs_p): Always use
4158 the full nest as region.
4159 (try_generate_gimple_bb): Likewise.
4160 * sese.c (scalar_evolution_in_region): Simplify now that
4161 SCEV can handle instantiation in regions.
4162 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4163 in the non-loop part of a function if requested.
4165 2017-10-17 Richard Biener <rguenther@suse.de>
4167 PR tree-optimization/82563
4168 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4170 (graphite_regenerate_ast_isl): Call it.
4171 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4173 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4175 PR tree-optimization/82549
4176 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4177 Formatting fixes. Instead of calling make_bit_field_ref with negative
4180 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4182 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4183 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4184 _mm_maskz_reduce_ss): New.
4185 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4186 __builtin_ia32_reducess_mask): Ditto..
4187 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4188 * config/i386/sse.md (reduces<mode>): Renamed to ...
4189 (reduces<mode><mask_scalar_name>): ... this.
4190 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4192 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4193 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4195 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4197 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4198 * selftest-run-tests.c (selftest::run_tests): Call
4199 selftest::unique_ptr_tests_cc_tests.
4200 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4201 * unique-ptr-tests.cc: New file.
4203 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4206 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4208 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4209 (make_hard_regno_born, make_hard_regno_dead): Update
4210 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4212 2017-10-16 Jeff Law <law@redhat.com>
4214 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4216 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4218 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4220 2017-10-16 Olivier Hainque <hainque@adacore.com>
4222 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4223 with_cpu if we were configured for an e500v2 target cpu name.
4225 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4227 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4228 * doc/invoke.texi: Document +nodsp as a valid extension for
4231 2017-10-16 Martin Liska <mliska@suse.cz>
4233 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4234 (test_set_range): Likewise.
4235 (test_range_functions): Rename to ...
4236 (test_bit_in_range): ... this.
4237 (sbitmap_c_tests): Add new test.
4239 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4241 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4243 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4244 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4246 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4248 * config/aarch64/aarch64-builtins.c
4249 (aarch64_types_quadopu_lane_qualifiers): New.
4250 (TYPES_QUADOPU_LANE): New.
4251 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4252 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4253 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4254 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4255 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4256 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4257 (Vdottype, DOTPROD): New.
4258 (sur): Add SDOT and UDOT.
4260 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4262 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4263 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4264 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4266 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4267 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4268 Enable TARGET_DOTPROD.
4269 (cortex-a75.cortex-a55): Likewise.
4270 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4272 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4274 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4275 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4277 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4279 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4280 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4281 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4282 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4283 * config/arm/types.md (neon_dot, neon_dot_q): New.
4284 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4286 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4288 * config/arm/arm.h (TARGET_DOTPROD): New.
4289 * config/arm/arm.c (arm_arch_dotprod): New.
4290 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4291 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4292 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4293 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4294 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4295 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4296 * doc/invoke.texi (armv8.2-a): Document dotprod
4298 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4300 * i386.c (ix86_vec_cost): New function.
4301 (ix86_rtx_costs): Handle vector operations better.
4302 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4303 * x86-tune-costs.h: Add new costs to all tables.
4305 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4307 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4309 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4310 divsd, sqrtss and sqrtsd
4311 * x86-tune-costs.h: Add new entries to all costs.
4312 (znver1_cost): Fix to match real instruction latencies.
4314 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4315 Michael Collison <michael.collison@arm.com>
4317 * compare-elim.c: Include emit-rtl.h.
4318 (can_merge_compare_into_arith): New function.
4319 (try_validate_parallel): Likewise.
4320 (try_merge_compare): Likewise.
4321 (try_eliminate_compare): Call the above when no previous clobber
4323 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4326 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4330 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4331 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4335 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4336 to be any operand_equal_p operands. For & (B - 1) require
4337 B to be power of 2. Recognize
4338 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4340 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4343 * optabs.c (expand_memory_blockage): Fix call of
4344 targetm.have_memory_blockage.
4346 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4349 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4350 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4351 objects to extra_objs instead of overwriting it.
4353 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4355 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4356 Use any_fp_register_operand as operand[3] predicate. Simplify
4357 equality test for operands[2] and operands[4] memory location.
4358 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4359 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4360 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4361 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4362 any_fp_register_operand as operand[1] predicate. Simplify
4363 equality test for operands[0] and operands[3] memory location.
4364 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4365 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4366 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4368 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4370 * target-insns.def: Add memory_blockage.
4371 * optabs.c (expand_memory_blockage): New function.
4372 (expand_asm_memory_barrier): Rename ...
4373 (expand_asm_memory_blockage): ... to this.
4374 (expand_mem_thread_fence): Call expand_memory_blockage
4375 instead of expand_asm_memory_barrier.
4376 (expand_mem_singnal_fence): Ditto.
4377 (expand_atomic_load): Ditto.
4378 (expand_atomic_store): Ditto.
4379 * doc/md.texi (Standard Pattern Names For Generation):
4380 Document memory_blockage instruction pattern.
4382 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4384 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4385 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4386 * config/rl78/rl78.md: New define_expand "adddi3".
4388 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4390 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4393 2017-10-13 Jeff Law <law@redhat.com>
4395 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4397 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4400 * internal-fn.c (expand_mul_overflow): If both operands have
4401 the same highpart of -1 or 0 and the topmost bit of lowpart
4402 is different, overflow is if res <= 0 rather than res < 0.
4404 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4406 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4407 TARGET_P9_VECTOR code for unaligned_load case.
4409 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4411 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4413 2017-10-13 Nathan Sidwell <nathan@acm.org>
4415 * tree-core.h (tree_contains_struct): Make bool.
4416 * tree.c (tree_contains_struct): Likewise.
4417 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4418 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4419 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4420 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4421 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4423 2017-10-13 Richard Biener <rguenther@suse.de>
4425 * graphite-isl-ast-to-gimple.c
4426 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4427 parameters and dominance check.
4428 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4429 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4430 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4431 Do not update SSA form here or do intermediate IL verification.
4432 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4433 (graphite_initialize): Remove check on the number of loops in
4434 the function and inline into graphite_transform_loops.
4435 (graphite_finalize): Inline into graphite_transform_loops.
4436 (graphite_transform_loops): Perform SSA update and IL verification
4438 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4440 2017-10-13 Richard Biener <rguenther@suse.de>
4442 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4443 graphite_expression_type_precision): Avoid global constructor
4445 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4446 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4447 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4448 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4449 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4450 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4452 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4455 * config/i386/i386.h (ix86_red_zone_size): New.
4456 * config/i386/i386.md (push peephole2s): Replace
4457 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4459 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4460 Alan Hayward <alan.hayward@arm.com>
4461 David Sherwood <david.sherwood@arm.com>
4463 * combine.c (can_change_dest_mode): Reject changes in
4464 REGMODE_NATURAL_SIZE.
4466 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4467 Alan Hayward <alan.hayward@arm.com>
4468 David Sherwood <david.sherwood@arm.com>
4470 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4471 (expand_debug_source_expr): Likewise.
4472 * combine.c (combine_simplify_rtx): Likewise.
4473 * cse.c (fold_rtx): Likewise.
4474 * fwprop.c (canonicalize_address): Likewise.
4475 * targhooks.c (default_shift_truncation_mask): Likewise.
4477 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4478 Alan Hayward <alan.hayward@arm.com>
4479 David Sherwood <david.sherwood@arm.com>
4481 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4482 (widened_mode): Likewise.
4483 (expand_unop): Likewise.
4484 * ree.c (transform_ifelse): Likewise.
4485 (merge_def_and_ext): Likewise.
4486 (combine_reaching_defs): Likewise.
4487 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4489 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4490 Alan Hayward <alan.hayward@arm.com>
4491 David Sherwood <david.sherwood@arm.com>
4493 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4494 * combine.c (gen_lowpart_for_combine): Likewise.
4495 * dwarf2out.c (rtl_for_decl_location): Likewise.
4496 * final.c (alter_subreg): Likewise.
4497 * rtlhooks.c (gen_lowpart_general): Likewise.
4498 (gen_lowpart_if_possible): Likewise.
4500 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4501 Alan Hayward <alan.hayward@arm.com>
4502 David Sherwood <david.sherwood@arm.com>
4504 * calls.c (expand_call): Use subreg_lowpart_offset.
4505 * cse.c (cse_insn): Likewise.
4506 * regcprop.c (copy_value): Likewise.
4507 (copyprop_hardreg_forward_1): Likewise.
4509 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4512 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4513 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4514 =Q constraints to +Q and into insn condition add check
4515 that operands[0] and operands[1] are equal.
4516 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4517 =Q constraints to +Q and into insn condition add check
4518 that operands[0] is equal to either operands[1] or operands[2].
4521 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4522 instead of handling MINUS_EXPR twice (once for each argument),
4523 canonicalize operand order and handle just once, use rtype where
4524 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4527 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4528 any values of __C while still being pattern recognizable as a simple
4531 2017-10-13 Richard Biener <rguenther@suse.de>
4533 PR tree-optimization/82451
4535 2017-10-02 Richard Biener <rguenther@suse.de>
4537 PR tree-optimization/82355
4538 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4539 a mapping for the enclosing loop but avoid generating one for
4541 (copy_bb_and_scalar_dependences): Remove premature codegen
4542 error on PHIs in blocks duplicated into multiple places.
4543 * graphite-scop-detection.c
4544 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4545 in the region use it as loop and nest to analyze the DR in.
4546 (try_generate_gimple_bb): Likewise.
4547 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4548 (add_loop_constraints): For blocks in a loop not in the region
4549 create a dimension with a single iteration.
4550 * sese.h (gbb_loop_at_index): Remove assert.
4552 * cfgloop.c (loop_preheader_edge): For the loop tree root
4553 return the single successor of the entry block.
4554 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4555 Reset the SCEV hashtable and niters.
4556 * graphite-scop-detection.c
4557 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4558 assert that we only have POLYNOMIAL_CHREC that vary in loops
4559 contained in the region.
4560 (scop_detection::graphite_can_represent_expr): Adjust.
4561 (scop_detection::stmt_has_simple_data_refs_p): For loops
4562 not in the region set loop to NULL. The nest is now the
4563 entry edge to the region.
4564 (try_generate_gimple_bb): Likewise.
4565 * sese.c (scalar_evolution_in_region): Adjust for
4566 instantiate_scev change.
4567 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4568 Make nest parameter the edge into the region.
4569 (create_data_ref): Likewise.
4570 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4571 entry edge into a region and adjust instantiate_scev calls.
4572 (create_data_ref): Likewise.
4573 (graphite_find_data_references_in_stmt): Likewise.
4574 (find_data_references_in_stmt): Pass the loop preheader edge
4575 from the nest argument.
4576 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4577 parameter the edge into the region.
4578 (instantiate_parameters): Use the loop preheader edge as entry.
4579 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4581 (get_instantiated_value_entry): Make instantiate_below parameter
4582 the edge into the region.
4583 (instantiate_scev_name): Likewise. Adjust dominance checks,
4584 when we cannot use loop-based instantiation instantiate by
4585 walking use-def chains.
4586 (instantiate_scev_poly): Adjust.
4587 (instantiate_scev_binary): Likewise.
4588 (instantiate_scev_convert): Likewise.
4589 (instantiate_scev_not): Likewise.
4590 (instantiate_array_ref): Remove.
4591 (instantiate_scev_3): Likewise.
4592 (instantiate_scev_2): Likewise.
4593 (instantiate_scev_1): Likewise.
4594 (instantiate_scev_r): Do not blindly handle N-operand trees.
4595 Do not instantiate array-refs. Handle all constants and invariants.
4596 (instantiate_scev): Make instantiate_below parameter
4597 the edge into the region.
4598 (resolve_mixers): Use the loop preheader edge for the region
4599 parameter to instantiate_scev_r.
4600 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4602 2017-10-13 Richard Biener <rguenther@suse.de>
4604 PR tree-optimization/82525
4605 * graphite-isl-ast-to-gimple.c
4606 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4608 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4609 Fail code generation when we cannot represent the isl integer.
4610 (binary_op_to_tree): Elide modulo operations that are no-ops
4611 in the type we code generate. Remove now superfluous code
4614 2017-10-13 Richard Biener <rguenther@suse.de>
4616 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4617 (scop_detection::harmful_loop_in_region): Remove premature
4618 IV type restriction.
4619 (scop_detection::graphite_can_represent_scev): We can handle
4620 pointer IVs just fine.
4622 2017-10-13 Alan Modra <amodra@gmail.com>
4624 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4625 "Clobbers and Scratch Registers". Add paragraph on
4626 alternative to clobbers for scratch registers and OpenBLAS
4629 2017-10-13 Alan Modra <amodra@gmail.com>
4631 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4632 example of a memory input for a string of known length. Move
4633 commentary out of table. Add a number of new examples
4634 covering array memory inputs.
4636 2017-10-12 Martin Liska <mliska@suse.cz>
4638 PR tree-optimization/82493
4639 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4640 (test_range_functions): New function.
4641 (sbitmap_c_tests): Likewise.
4642 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4643 * selftest.h (sbitmap_c_tests): New function.
4645 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4647 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4649 * config/rs6000/amo.h: Fix spacing issue.
4651 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4654 * config/i386/i386.md (*ashl<mode>3_mask_1,
4655 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4656 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4659 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4661 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4662 (profile_probability): Set max_probability
4663 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4665 * profile-count.c (profile_probability::differs_from_p): Do not
4666 rely on max_probaiblity == 10000
4668 2017-10-12 Jeff Law <law@redhat.com>
4670 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4673 2017-10-12 Martin Sebor <msebor@redhat.com>
4677 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4678 (handle_alias_pairs): Call it.
4679 * common.opt (-Wattribute-alias): New option.
4680 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4681 * doc/invoke.texi (-Wattribute-alias): Document.
4683 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4686 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4688 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4690 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4691 (make_hard_regno_born, make_hard_regno_dead): Update
4692 bb_killed_pseudos and bb_gen_pseudos.
4694 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4696 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4698 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4700 * config/alpha/alpha.c (alpha_split_conditional_move):
4701 Use std::swap instead of manually swapping.
4702 (alpha_stdarg_optimize_hook): Ditto.
4703 (alpha_canonicalize_comparison): Ditto.
4705 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4707 * tree-loop-distribution.c (struct builtin_info): New struct.
4708 (struct partition): Refactor fields into struct builtin_info.
4709 (partition_free): Free struct builtin_info.
4710 (build_size_arg_loc, build_addr_arg_loc): Delete.
4711 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4712 information from struct builtin_info.
4713 (find_single_drs): New function refactored from classify_partition.
4714 Also moved builtin validity checks to this function.
4715 (compute_access_range, alloc_builtin): New functions.
4716 (classify_builtin_st, classify_builtin_ldst): New functions.
4717 (classify_partition): Refactor code into functions find_single_drs,
4718 classify_builtin_st and classify_builtin_ldst.
4719 (distribute_loop): Don't do runtime alias check when distributing
4721 (find_seed_stmts_for_distribution): New function.
4722 (pass_loop_distribution::execute): Refactor code finding seed
4723 stmts into above function. Support distribution for the innermost
4724 two-level loop nest. Adjust dump information.
4726 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4728 * tree-loop-distribution.c: Adjust the general comment.
4729 (NUM_PARTITION_THRESHOLD): New macro.
4730 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4731 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4732 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4733 in call to build_partition_graph.
4734 (finalize_partitions): New parameter. Make loop distribution more
4735 conservative by fusing more partitions.
4736 (distribute_loop): Don't do runtime alias check in case of loop nest
4738 (find_seed_stmts_for_distribution): New function.
4739 (prepare_perfect_loop_nest): New function.
4740 (pass_loop_distribution::execute): Refactor code finding seed stmts
4741 and loop nest into above functions. Support loop nest distribution.
4742 Adjust dump information accordingly.
4744 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4746 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4747 and set PTYPE_SEQUENTIAL for merged partition.
4749 2017-10-12 Richard Biener <rguenther@suse.de>
4751 PR tree-optimization/69728
4753 2017-09-19 Richard Biener <rguenther@suse.de>
4755 PR tree-optimization/69728
4756 * graphite-sese-to-poly.c (schedule_error): New global.
4757 (add_loop_schedule): Handle empty domain by failing the
4759 (build_original_schedule): Handle schedule_error.
4761 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4762 domain by returning an unchanged schedule.
4764 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4766 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4767 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4769 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4771 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4774 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4777 * expr.c (store_field): Don't optimize away bitsize == 0 store
4778 from CALL_EXPR with addressable return type.
4780 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4782 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4783 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4784 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4785 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4786 TARGET_ISEL instead of TARGET_ISEL<sel>.
4788 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4790 * config/rs6000/rs6000.c
4791 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4793 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4795 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4797 (reg_or_cint_operand): Fix comment.
4798 (reg_or_zero_operand): New predicate.
4799 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4800 * config/rs6000/rs6000.c (output_isel): Delete.
4801 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4802 instead of reg_or_cint_operand. Output instruction directly (not via
4804 (isel_unsigned_<mode>): Ditto.
4805 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4806 gpc_reg_operand. Add an instruction alternative for this. Output
4807 instruction directly.
4808 (*isel_reversed_unsigned_<mode>): Ditto.
4810 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4812 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4813 (TARGET_CANONICALIZE_COMPARISON): Define.
4815 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4818 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4819 Check whether the dest is REG before adding REG_EQUIV note.
4821 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4824 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4826 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4827 (make_hard_regno_born, make_hard_regno_dead): Update
4828 bb_killed_pseudos and bb_gen_pseudos.
4830 2017-10-11 Nathan Sidwell <nathan@acm.org>
4832 * incpath.h (enum incpath_kind): Name enum, prefix values.
4833 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4834 * incpath.c (heads, tails): Use INC_MAX.
4835 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4836 (merge_include_chains, split_quote_chain,
4837 register_include_chains): Update incpath_kind names.
4838 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4839 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4841 (add_framework_path, darwin_register_objc_includes): Likewise.
4842 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4844 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4846 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4847 Do not use float_operator operator predicate.
4848 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4849 * config/i386/predicates.md (float_operator): Remove predicate.
4851 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4853 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4854 (*jccxf_i387): Ditto.
4855 (*jcc<mode>_i387): Ditto.
4856 (*jccu<mode>_i387): Ditto.
4857 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4858 (*jcc_*_i387 splitters): Remove.
4859 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4860 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4861 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4864 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4866 * profile-count.h (slow_safe_scale_64bit): New function.
4867 (safe_scale_64bit): New inline.
4868 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4869 * profile-count.c: Include wide-int.h
4870 (slow_safe_scale_64bit): New.
4872 2017-10-11 Nathan Sidwell <nathan@acm.org>
4874 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4875 HAS_DECL_ASSEMBLER_NAME_P.
4876 * gimple-expr.c (gimple_decl_printable_name: Check
4877 HAS_DECL_ASSEMBLER_NAME_P too.
4878 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4879 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4880 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4881 * passes.c (rest_of_decl_compilation): Check
4882 HAS_DECL_ASSEMBLER_NAME_P too.
4883 * recog.c (verify_changes): Likewise.
4884 * tree-pretty-print.c (dump_decl_name): Likewise.
4885 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4887 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4888 (SET_DECL_ASSEMBLER_NAME): Use it.
4889 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4890 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4891 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4893 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4895 * config.gcc (i386, x86_64): Add extra objects.
4896 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4897 (ix86_min_insn_size): Declare.
4898 (ix86_issue_rate): Declare.
4899 (ix86_adjust_cost): Declare.
4900 (ia32_multipass_dfa_lookahead): Declare.
4901 (ix86_macro_fusion_p): Declare.
4902 (ix86_macro_fusion_pair_p): Declare.
4903 (ix86_bd_has_dispatch): Declare.
4904 (ix86_bd_do_dispatch): Declare.
4905 (ix86_core2i7_init_hooks): Declare.
4906 (ix86_atom_sched_reorder): Declare.
4907 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4908 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4909 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4910 (rip_relative_addr_p): Rename to ...
4911 (ix86_rip_relative_addr_p): ... this one; export.
4912 (memory_address_length): Update.
4913 (ix86_issue_rate): Move to x86-tune-sched.c.
4914 (ix86_flags_dependent): Move to x86-tune-sched.c.
4915 (ix86_agi_dependent): Move to x86-tune-sched.c.
4916 (exact_dependency_1): Move to x86-tune-sched.c.
4917 (exact_store_load_dependency): Move to x86-tune-sched.c.
4918 (ix86_adjust_cost): Move to x86-tune-sched.c.
4919 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4920 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4921 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4922 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4923 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4924 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4925 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4926 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4927 (min_insn_size): Rename to ...
4928 (ix86_min_insn_size): ... this one; export.
4929 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4930 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4931 (core2i7_first_cycle_multipass_backtrack): Move to
4932 x86-tune-sched-core.c.
4933 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4934 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4935 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4936 (ix86_avoid_jump_mispredicts): Update.
4937 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4938 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4939 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4940 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4941 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4942 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4943 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4944 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4945 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4946 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4947 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4948 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4949 (BIG): Move to ix86-tune-sched-bd.c.
4950 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4951 (enum insn_path): Move to ix86-tune-sched-bd.c.
4952 (get_mem_group): Move to ix86-tune-sched-bd.c.
4953 (is_cmp): Move to ix86-tune-sched-bd.c.
4954 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4955 (is_branch): Move to ix86-tune-sched-bd.c.
4956 (is_prefetch): Move to ix86-tune-sched-bd.c.
4957 (init_window): Move to ix86-tune-sched-bd.c.
4958 (allocate_window): Move to ix86-tune-sched-bd.c.
4959 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4960 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4961 (process_end_window): Move to ix86-tune-sched-bd.c.
4962 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4963 (find_constant): Move to ix86-tune-sched-bd.c.
4964 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4965 (has_immediate): Move to ix86-tune-sched-bd.c.
4966 (get_insn_path): Move to ix86-tune-sched-bd.c.
4967 (get_insn_group): Move to ix86-tune-sched-bd.c.
4968 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4969 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4970 (add_insn_window): Move to ix86-tune-sched-bd.c.
4971 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4972 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4973 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4974 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4975 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4976 (do_dispatch): Move to ix86-tune-sched-bd.c.
4977 (has_dispatch): Move to ix86-tune-sched-bd.c.
4978 * config/i386/t-i386: Add new object files.
4979 * config/i386/x86-tune-costs.h: New file.
4980 * config/i386/x86-tune-sched-atom.c: New file.
4981 * config/i386/x86-tune-sched-bd.c: New file.
4982 * config/i386/x86-tune-sched-core.c: New file.
4983 * config/i386/x86-tune-sched.c: New file.
4985 2017-10-11 Liu Hao <lh_mouse@126.com>
4987 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4988 the generic version below instead.
4989 (should_colorize): Recognize Windows consoles as terminals
4991 * pretty-print.c [__MINGW32__] (write_all): New function.
4992 [__MINGW32__] (find_esc_head): Likewise.
4993 [__MINGW32__] (find_esc_terminator): Likewise.
4994 [__MINGW32__] (eat_esc_sequence): Likewise.
4995 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4997 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5000 2017-10-11 Richard Biener <rguenther@suse.de>
5002 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5003 Properly call analyze_scalar_evolution with the loop of the stmt.
5005 2017-10-11 Richard Biener <rguenther@suse.de>
5007 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5008 * tree-core.h (tree_base): Add chrec_var union member.
5009 * tree.h (CHREC_VAR): Remove.
5010 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5011 * tree-chrec.h (build_polynomial_chrec): Adjust.
5012 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5013 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5015 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5017 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5018 * match.pd: ... here.
5019 ((T) X == (T) Y): Relax condition.
5021 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5023 PR tree-optimization/82472
5024 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5026 (break_alias_scc_partitions): Update postorder number.
5028 2017-10-11 Martin Liska <mliska@suse.cz>
5031 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5033 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5035 2017-10-11 Martin Liska <mliska@suse.cz>
5037 * print-rtl.c (print_insn): Move declaration of idbuf
5038 to same scope as name.
5040 2017-10-11 Martin Liska <mliska@suse.cz>
5045 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5048 2017-10-11 Martin Liska <mliska@suse.cz>
5051 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5054 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5056 * config.gcc (default_gnu_indirect_function): Default to yes for
5057 arm*-*-linux* with glibc.
5059 2017-10-11 Richard Biener <rguenther@suse.de>
5061 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5062 default-defs and types we do not want to analyze.
5063 (interpret_loop_phi): Replace unreachable code with an assert.
5064 (compute_scalar_evolution_in_loop): Remove and inline ...
5065 (analyze_scalar_evolution_1): ... here, replacing condition with
5066 what makes the intent clearer. Remove handling of cases
5067 get_scalar_evolution now handles.
5069 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5071 PR rtl-optimization/81434
5072 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5073 comment for main loop. In sched_group_found if, also add checks for
5074 pass and min_cost_group.
5076 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5078 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5079 (rs6000_insn_cost): New function.
5080 * config/rs6000/rs6000.md (cost): New attribute.
5082 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5083 H.J. Lu <hongjiu.lu@intel.com>
5087 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5088 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5089 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5090 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5091 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5092 OPTION_MASK_ISA_MMX and some other bit set require both
5093 mmx and the ISAs without the mmx bit.
5094 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5095 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5096 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5097 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5098 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5099 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5100 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5101 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5102 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5103 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5104 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5105 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5106 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5107 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5108 Add OPTION_MASK_ISA_MMX.
5110 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5112 * config.gcc (armv7*-*-freebsd*): New target.
5113 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5115 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5117 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5118 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5121 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5123 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5124 a template parameter.
5125 (WIDE_INT_REF_FOR): Update accordingly.
5126 * tree.h (wi::int_traits <const_tree>): Delete.
5127 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5128 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5129 (wi::tree_to_wide_ref): New typedef.
5130 (wi::to_wide): New function.
5131 * calls.c (get_size_range): Use wi::to_wide when operating on
5133 * cgraph.c (cgraph_node::create_thunk): Likewise.
5134 * config/i386/i386.c (ix86_data_alignment): Likewise.
5135 (ix86_local_alignment): Likewise.
5136 * dbxout.c (stabstr_O): Likewise.
5137 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5138 * expr.c (const_vector_from_tree): Likewise.
5139 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5140 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5141 (fold_negate_expr_1, int_const_binop_1, const_binop)
5142 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5143 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5144 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5145 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5146 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5147 (fold_not_const, round_up_loc): Likewise.
5148 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5149 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5150 (alloca_call_type): Likewise.
5151 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5152 * godump.c (go_output_typedef): Likewise.
5153 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5154 * internal-fn.c (get_min_precision): Likewise.
5155 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5156 * ipa-polymorphic-call.c
5157 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5158 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5159 (ipa_modify_call_arguments): Likewise.
5160 * match.pd: Likewise.
5161 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5162 * print-tree.c (print_node_brief, print_node): Likewise.
5163 * stmt.c (expand_case): Likewise.
5164 * stor-layout.c (layout_type): Likewise.
5165 * tree-affine.c (tree_to_aff_combination): Likewise.
5166 * tree-cfg.c (group_case_labels_stmt): Likewise.
5167 * tree-data-ref.c (dr_analyze_indices): Likewise.
5168 (prune_runtime_alias_test_list): Likewise.
5169 * tree-dump.c (dequeue_and_dump): Likewise.
5170 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5171 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5172 * tree-pretty-print.c (dump_generic_node): Likewise.
5173 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5174 (simple_iv_with_niters): Likewise.
5175 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5176 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5177 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5178 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5179 (refine_value_range_using_guard, number_of_iterations_ne_max)
5180 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5181 (get_cst_init_from_scev, record_nonwrapping_iv)
5182 (scev_var_range_cant_overflow): Likewise.
5183 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5184 * tree-ssa-pre.c (compute_avail): Likewise.
5185 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5186 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5187 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5188 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5189 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5190 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5191 (dump_case_nodes, try_switch_expansion): Likewise.
5192 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5193 (vect_do_peeling): Likewise.
5194 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5195 * tree-vect-stmts.c (vectorizable_load): Likewise.
5196 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5197 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5198 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5199 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5200 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5201 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5202 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5203 (evrp_dom_walker::before_dom_children): Likewise.
5204 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5205 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5206 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5207 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5208 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5209 (get_range_pos_neg): Likewise.
5210 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5211 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5212 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5213 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5214 * config/avr/avr.c (avr_fold_builtin): Likewise.
5215 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5216 * config/msp430/msp430.c (msp430_attr): Likewise.
5217 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5218 * config/powerpcspe/powerpcspe-c.c
5219 (altivec_resolve_overloaded_builtin): Likewise.
5220 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5221 (rs6000_expand_ternop_builtin): Likewise.
5222 * config/rs6000/rs6000-c.c
5223 (altivec_resolve_overloaded_builtin): Likewise.
5224 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5225 (rs6000_expand_ternop_builtin): Likewise.
5226 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5228 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5230 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5231 when copying loop nest with only one inner loop.
5233 2017-10-10 Richard Biener <rguenther@suse.de>
5235 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5236 blocks if SCEV is active.
5237 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5239 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5240 (scev_initialize): Assert we are not yet initialized.
5242 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5244 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5245 inner loop's exit stmt by making it always exit the loop, otherwise
5246 we would generate an infinite empty loop.
5248 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5250 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5251 renaming variables in new preheader if it's deleted.
5253 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5255 * tree-loop-distribution.c (struct partition): Remove unused field
5256 loops of the structure.
5257 (partition_alloc, partition_free): Ditto.
5258 (build_rdg_partition_for_vertex): Ditto.
5260 2017-10-09 Jeff Law <law@redhat.com>
5262 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5263 return type to match prototype and documentation.
5265 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5267 * config/rs6000/rs6000.c (processor_costs): Move to ...
5268 * config/rs6000/rs6000.h: ... here.
5269 (rs6000_cost): Declare.
5271 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5273 * except.c (setjmp_fn): New global variable.
5274 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5275 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5276 if DONT_USE_BUILTIN_SETJMP is defined.
5278 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5280 * target.def (insn_cost): New hook.
5281 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5282 * doc/tm.texi: Regenerate.
5283 * rtlanal.c (insn_cost): Use the new hook.
5285 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5287 * combine.c (combine_validate_cost): Compute the new insn_cost,
5288 not just pattern_cost.
5289 (try_combine): Adjust comment.
5291 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5293 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5295 * combine.c (uid_insn_cost): Adjust comment.
5296 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5298 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5299 * dse.c (find_shift_sequence): Ditto.
5300 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5301 (bb_valid_for_noce_process_p): Use pattern_cost.
5302 * rtl.h (insn_rtx_cost): Delete.
5303 (pattern_cost): New prototype.
5304 (insn_cost): New prototype.
5305 * rtlanal.c (insn_rtx_cost): Rename to...
5306 (pattern_cost): ... this.
5309 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5311 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5312 (*jcc<mode>_0_r_i387): Ditto.
5313 (*jccxf_r_i387): Ditto.
5314 (*jcc<mode>_r_i387): Ditto.
5315 (*jccu<mode>_r_i387): Ditto.
5316 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5317 (*jcc): Rename from *jcc_1.
5319 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5321 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5322 deferred rescans after the lvx/stvx recombination pre-pass.
5324 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5326 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5327 memory operation instruction support.
5328 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5329 (rs6000-ibm-aix[789]*): Likewise.
5330 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5331 Document new functions.
5333 2017-10-09 Richard Biener <rguenther@suse.de>
5335 PR tree-optimization/82397
5336 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5337 equality only for semantically equal trees.
5339 2017-10-09 Richard Biener <rguenther@suse.de>
5341 PR tree-optimization/82449
5342 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5343 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5344 allow constant addresses.
5345 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5348 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5350 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5353 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5356 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5359 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5362 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5364 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5367 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5368 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5370 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5372 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5373 (WI_BINARY_PREDICATE_RESULT): Likewise.
5374 (wi::binary_traits::operator_result): New type.
5375 (wi::binary_traits::predicate_result): Likewise.
5376 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5377 (generic_wide_int::operator==, generic_wide_int::operator!=)
5378 (generic_wide_int::operator&, generic_wide_int::and_not)
5379 (generic_wide_int::operator|, generic_wide_int::or_not)
5380 (generic_wide_int::operator^, generic_wide_int::operator+
5381 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5383 (operator~, unary operator-, operator==, operator!=, operator&)
5384 (operator|, operator^, operator+, binary operator-, operator*): New
5386 * expr.c (get_inner_reference): Use wi::bit_and_not.
5387 * fold-const.c (fold_binary_loc): Likewise.
5388 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5389 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5390 (bit_value_binop): Likewise.
5391 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5392 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5393 (extract_range_from_binary_expr_1): Likewise.
5394 (masked_increment): Likewise.
5395 (simplify_bit_ops_using_ranges): Likewise.
5397 2017-10-09 Martin Jambor <mjambor@suse.cz>
5400 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5401 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5402 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5404 (hsa_fixup_mov_insn_type): New function.
5405 (hsa_op_with_type::get_in_type): Use it.
5406 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5407 immediates in an assert.
5408 (hsa_op_with_type::extend_int_to_32bit): New method.
5409 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5410 types. Convert to dest type if necessary.
5411 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5412 (reg_for_gimple_ssa): Pass false as min32int to
5413 hsa_type_for_scalar_tree_type.
5414 (gen_hsa_addr): Fixup type when creating addresable temporary.
5415 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5416 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5417 necessary. Call hsa_fixup_mov_insn_type.
5418 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5419 extend operands and convert to dest type if necessary.
5420 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5421 to dest type if necessary.
5422 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5423 if conversion nt necessary and size matches.
5424 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5425 to dest type if necessary.
5426 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5427 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5429 (gen_hsa_clrsb): Likewise.
5430 (gen_hsa_ffs): Likewise.
5431 (gen_hsa_divmod): Extend operands and convert to dest type if
5433 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5435 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5437 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5438 Remove empty default arguments. Use a brace block as output
5440 (conditional return): Ditto.
5442 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5443 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5445 (group_ending_nop): Ditto.
5446 (doloop_end): Ditto.
5447 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5448 (splitters for those): Ditto.
5450 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5452 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5453 a conditional jump (and the compare for it) so that pc_rtx is the
5455 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5456 for the deleted and renamed ctr<mode>_internal[234] patterns.
5457 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5458 Delete second conditional return pattern.
5459 (ctr<mode>_internal2): Delete this second bdnz pattern.
5460 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5461 (ctr<mode>_internal4): Delete this second bdz pattern.
5463 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5465 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5466 (always_initialized_rtx_for_ssa_name_p): New predicate.
5467 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5468 (finish_out_of_ssa): Free new field of SA.
5469 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5470 * tree-ssa-coalesce.c: Include tree-ssa.h.
5471 (get_parm_default_def_partitions): Remove extern keyword.
5472 (get_undefined_value_partitions): New function.
5473 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5474 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5476 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5478 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5480 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5482 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5484 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5485 for targets that preffer 128bit.
5487 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5489 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5491 2017-10-08 Olivier Hainque <hainque@adacore.com>
5493 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5494 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5496 (thumb_set_return_address): Likewise.
5498 2017-10-08 Olivier Hainque <hainque@adacore.com>
5500 * common/config/arm/arm-common.c (arm_except_unwind_info):
5501 Handle DWARF2_UNWIND_INFO.
5503 2017-10-07 Michael Collison <michael.collison@arm.com>
5505 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5508 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5510 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5511 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5512 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5513 defined, force the creation of a new block for a dispatch label.
5515 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5517 * invoke.texi (Wsuggest-attribute=cold): Document.
5518 * common.opt (Wsuggest-attribute=cold): New
5519 * ipa-pure-const.c (warn_function_cold): New function.
5520 * predict.c (compute_function_frequency): Use it.
5521 * predict.h (warn_function_cold): Declare.
5523 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5525 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5528 2017-10-06 Martin Liska <mliska@suse.cz>
5530 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5531 keyword for member functions.
5532 (struct sanopt_tree_couple): New struct.
5533 (struct sanopt_tree_couple_hash): New function.
5534 (struct sanopt_ctx): Add new hash_map.
5535 (has_dominating_ubsan_ptr_check): New function.
5536 (record_ubsan_ptr_check_stmt): Likewise.
5537 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5538 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5539 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5541 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5544 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5545 aarch64_simd_valid_immediate on CONST_VECTORs.
5546 (aarch64_reg_or_bic_imm): Likewise.
5548 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5550 PR rtl-optimization/82396
5551 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5553 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5555 * graphite-dependences.c (scop_get_reads): Move code to...
5556 (scop_get_must_writes): Move code to...
5557 (scop_get_may_writes): Move code to...
5558 (scop_get_reads_and_writes): ... here.
5559 (scop_get_dependences): Call scop_get_reads_and_writes.
5561 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5563 PR tree-optimization/82434
5564 * fold-const.h (can_native_encode_type_p,
5565 can_native_encode_string_p): Remove.
5566 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5567 don't encode anything, just return what would be otherwise returned.
5568 (native_encode_fixed, native_encode_complex, native_encode_vector):
5570 (native_encode_string): Likewise. Inline by hand
5571 can_native_encode_string_p.
5572 (can_native_encode_type_p): Remove.
5573 (can_native_encode_string_p): Remove.
5574 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5575 STRING_CSTs using can_native_encode_string_p, test all
5576 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5577 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5578 argument from native_encode_expr.
5579 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5580 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5583 2017-10-06 Richard Biener <rguenther@suse.de>
5585 PR tree-optimization/82397
5586 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5587 operand_equal_p but rely on data_ref_compare_tree for detecting
5589 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5590 to match up with dr_group_sort_cmp.
5592 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5595 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5597 * config/s390/s390-builtin-types.def: Regenerate.
5599 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5602 * config/s390/s390-builtin-types.def: Regenerate.
5603 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5604 Change flag from B_VXE to B_VX.
5605 (s390_vec_min_dbl): Remove B_VXE flag.
5607 2017-10-06 Richard Biener <rguenther@suse.de>
5609 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5610 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5611 translate_isl_ast_to_gimple::is_valid_rename,
5612 translate_isl_ast_to_gimple::get_rename,
5613 translate_isl_ast_to_gimple::get_def_bb_for_const,
5614 translate_isl_ast_to_gimple::get_new_name,
5615 translate_isl_ast_to_gimple::collect_all_ssa_names,
5616 translate_isl_ast_to_gimple::copy_loop_phi_args,
5617 translate_isl_ast_to_gimple::collect_all_ssa_names,
5618 translate_isl_ast_to_gimple::copy_loop_phi_args,
5619 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5620 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5621 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5622 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5623 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5624 translate_isl_ast_to_gimple::copy_cond_phi_args,
5625 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5626 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5627 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5628 translate_isl_ast_to_gimple::rename_uses,
5629 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5630 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5631 (set_rename_for_each_def): Likewise.
5632 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5633 here. Handle rewriting SCEV analyzable uses here.
5634 (copy_bb_and_scalar_dependences): Generate code for PHI
5636 (graphite_regenerate_ast_isl): Adjust.
5637 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5638 (add_write, add_read): New functions.
5639 (build_cross_bb_scalars_def): Use it and simplify.
5640 (build_cross_bb_scalars_use): Likewise.
5641 (graphite_find_cross_bb_scalar_vars): Inline into...
5642 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5643 simulating out-of-SSA. Compute liveout and add dependencies.
5644 (build_scops): Force an empty entry block.
5645 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5647 (sese_build_liveouts): Declare.
5648 (sese_trivially_empty_bb_p): Likewise.
5649 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5650 compute liveout and debug_liveout.
5651 (sese_bad_liveouts_use): Remove.
5652 (sese_reset_debug_liveouts_bb): Likewise.
5653 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5654 (sese_build_liveouts): Build liveout and debug_liveout and store
5656 (new_sese_info): Adjust.
5657 (free_sese_info): Likewise.
5658 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5659 do not build liveout here.
5660 (move_sese_in_condition): Adjust region entry.
5661 (scev_analyzable_p): Match up with chrec_apply requirements.
5662 (sese_trivially_empty_bb_p): New.
5663 * tree-into-ssa.c (get_reaching_def): Properly support generating
5664 default-defs for incremental rewrite of anonymous names.
5666 2017-10-06 Richard Biener <rguenther@suse.de>
5668 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5669 precision do not perform modulo reduction.
5671 2017-10-06 Richard Biener <rguenther@suse.de>
5673 PR tree-optimization/82436
5674 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5675 conservatively choose the vectorization factor when checking
5676 whether we can perform the required load permutation.
5677 (vect_transform_slp_perm_load): Assert when we may not fail.
5679 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5681 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5682 message for incompatible -msdata=* and -mcall-* options.
5684 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5686 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5687 rate for post-reload scheduling.
5689 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5691 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5693 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5695 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5696 to improve monte carlo in scimark.
5698 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5700 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5701 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5702 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5703 pentium4_cost, nocona_cost): Set reassociation width to 1.
5704 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5705 width to 2 for fp operations and 1 otherwise.
5706 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5708 (atom_cost): Set reassociation width to 2.
5709 (slm_cost, generic_cost): Set fp reassociation width
5710 to 2 and 1 otherwise.
5711 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5712 (core_cost): Set fp reassociation width to 4 and vector to 2.
5713 (ix86_reassociation_width): Rewrite using cost table; special case
5714 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5715 and TARGET_AVX128_OPTIMAL.
5716 * config/i386/i386.h (processor_costs): Add
5717 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5718 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5719 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5720 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5721 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5722 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5724 2017-10-05 Nathan Sidwell <nathan@acm.org>
5726 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5728 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5730 * config/arm/arm.c (arm_test_fpu_data): New.
5731 (arm_run_selftests): Call arm_test_fpu_data.
5733 2017-10-04 Nathan Sidwell <nathan@acm.org>
5735 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5737 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5739 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5741 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5742 check type for aarch64_simd_valid_immediate.
5743 (aarch64_output_simd_mov_immediate): Update prototype.
5744 (aarch64_simd_valid_immediate): Update prototype.
5745 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5746 support for ORR-immediate.
5747 (and<mode>3): modified pattern to add support for BIC-immediate.
5748 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5749 now checks for valid immediate for BIC and ORR based on new enum
5751 (aarch64_output_simd_mov_immediate): Function now used to output
5752 BIC/ORR imm as well based on new enum argument.
5753 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5755 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5756 (aarch64_reg_or_bic_imm): Likewise.
5758 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5760 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5761 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5763 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5767 PR rtl-optimization/82396
5768 * haifa-sched.c (autopref_multipass_init): Simplify
5770 (autopref_rank_data): Simplify sort order.
5771 * sched-int.h (autopref_multipass_data_): Remove
5772 multi_mem_insn_p, min_offset and max_offset.
5774 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5776 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5778 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5780 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5783 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5785 * doc/sourcebuild.texi: Document vect_long_mult.
5787 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5789 PR tree-optimization/82413
5790 * fold-const.c (build_range_check): Use widest_int when comparing
5791 the maximum ETYPE value with HIGH.
5793 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5795 PR rtl-optimization/82396
5796 * haifa-sched.c (autopref_multipass_init): Simplify
5798 (autopref_rank_data): Simplify sort order.
5799 * sched-int.h (autopref_multipass_data_): Remove
5800 multi_mem_insn_p, min_offset and max_offset.
5802 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5804 PR tree-optimization/82381
5805 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5806 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5809 PR tree-optimization/82374
5810 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5811 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5812 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5813 current_function_decl to the new decl.
5815 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5817 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5818 helper macro for IEEE float128 hardware built-in functions.
5819 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5821 (TRUNCF128_ODD): Likewise.
5822 (ADDF128_ODD): Likewise.
5823 (SUBF128_ODD): Likewise.
5824 (MULF128_ODD): Likewise.
5825 (DIVF128_ODD): Likewise.
5826 (FMAF128_ODD): Likewise.
5827 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5828 UNSPEC_TRUNC_ROUND_TO_ODD.
5829 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5830 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5831 floating point round to odd instructions.
5832 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5833 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5834 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5835 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5836 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5837 (trunc<mode>sf2_hw): Change the truncate with round to odd
5838 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5839 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5840 to odd hardware instructions.
5841 (sub<mode>3_odd): Likewise.
5842 (mul<mode>3_odd): Likewise.
5843 (div<mode>3_odd): Likewise.
5844 (sqrt<mode>2_odd): Likewise.
5845 (fma<mode>4_odd): Likewise.
5846 (fms<mode>4_odd): Likewise.
5847 (nfma<mode>4_odd): Likewise.
5848 (nfms<mode>4_odd): Likewise.
5849 (trunc<mode>df2_odd): Change the truncate with round to odd
5850 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5852 * doc/extend.texi (PowerPC built-in functions): Update documentation
5853 for existing IEEE float128-bit built-in functions. Add built-in
5854 functions that generate the IEEE 128-bit floating point round to
5857 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5859 PR rtl-optimization/77729
5860 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5861 to (X&(C1&~C2))|C2 transformations.
5863 2017-10-03 Martin Jambor <mjambor@suse.cz>
5865 PR tree-optimization/82363
5866 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5867 mismatch, mark lacc written regardless of racc.
5869 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5871 PR tree-optimization/82381
5872 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5873 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5874 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5875 fallthrough into reassoc_stmt_dominates_stmt_p.
5878 * combine.c (combine_instructions): Don't combine in unreachable
5881 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5884 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5885 function to not use the have_cpu variable. Do not set cpu_index,
5886 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5888 (rs6000_valid_attribute_p): Remove duplicate initializations of
5889 old_optimize and func_optimize.
5890 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5891 (rs6000_activate_target_options): Make global.
5892 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5895 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5897 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5898 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5899 Set *poffset to 0 and *psize and *pmax_size to -1 if
5900 *poffset + *psize overflows in HOST_WIDE_INT.
5902 PR tree-optimization/82387
5903 PR tree-optimization/82388
5904 PR tree-optimization/82389
5905 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5906 instead of live_bytes non-NULL.
5908 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5911 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5914 2017-10-02 Richard Biener <rguenther@suse.de>
5916 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5917 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5918 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5920 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5922 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5923 requested precision matches the type's.
5924 * calls.c (alloc_max_size): Calculate the new candidate size as
5925 a widest_int and use wi::to_widest when comparing it with the
5926 current candidate size.
5927 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5928 zero rather than integer_zero_node.
5929 * match.pd: Check for a no-op conversion before using wi::add
5930 rather than after. Use tree_to_uhwi when summing small shift
5931 counts into an unsigned int.
5933 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5934 Alan Hayward <alan.hayward@arm.com>
5935 David Sherwood <david.sherwood@arm.com>
5938 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5939 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5940 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5941 POINTER_AND_FP_REGS.
5943 2017-10-02 Richard Biener <rguenther@suse.de>
5945 PR tree-optimization/82355
5946 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5947 a mapping for the enclosing loop but avoid generating one for
5949 (copy_bb_and_scalar_dependences): Remove premature codegen
5950 error on PHIs in blocks duplicated into multiple places.
5951 * graphite-scop-detection.c
5952 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5953 in the region use it as loop and nest to analyze the DR in.
5954 (try_generate_gimple_bb): Likewise.
5955 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5956 (add_loop_constraints): For blocks in a loop not in the region
5957 create a dimension with a single iteration.
5958 * sese.h (gbb_loop_at_index): Remove assert.
5960 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5962 * omp-expand.c (adjust_context_scope): New function.
5963 (expand_parallel_call): Call adjust_context_scope.
5965 2017-10-01 Jeff Law <law@redhat.com>
5967 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5968 dom_opt_dom_walker class with direct access to private members.
5969 Add comments. Call test_for_singularity.
5970 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5971 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5972 m_dummy_cond anymore.
5973 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5975 (pass_dominator:execute): Build the dummy_cond here and pass it
5976 to the dom_opt_dom_walker ctor.
5977 (test_for_singularity): New function.
5979 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5980 Maya Rashish <coypu@sdf.org>
5982 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5983 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5984 (alpha*-*-netbsd*) Use nbsd_tm_file.
5985 (arm*-*-netbsdelf*) Likewise.
5986 (i[34567]86-*-netbsdelf*) Likewise.
5987 (x86_64-*-netbsd*) Likewise.
5988 (mips*-*-netbsd*) Likewise.
5989 (powerpc-*-netbsd*) Likewise.
5990 (sh*-*-netbsd*) Likewise.
5991 (sparc-*-netbsdelf*) Likewise.
5992 (sparc64-*-netbsd*) Likewise.
5993 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5995 (vax-*-netbsdelf*) Likewise.
5996 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5997 (UINT_FAST8_TYPE) Likewise.
5998 (INT_FAST16_TYPE) Check CHAR_FAST16.
5999 (UINT_FAST16_TYPE) Likewise.
6001 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6004 * config/i386/i386.md
6005 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6006 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6007 *divmodsi4_zext_2): New define_insn_and_split.
6008 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6009 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6010 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6011 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6012 New define_insn_and_split.
6013 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6014 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6015 operands[1] having DImode when mode is SImode.
6017 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6018 always SImode for DIV and MOD in REG_EQUAL notes.
6020 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6023 * match.pd: Fix handling of NaNs in pattern.
6025 2017-09-29 Jeff Law <law@redhat.com>
6027 * sbitmap.c (bitmap_bit_in_range_p): New function.
6028 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6029 * tree-ssa-dse.c (live_bytes_read): New function.
6030 (dse_classify_store): Ignore reads of dead bytes.
6032 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6033 typos and whitespace errors.
6034 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6035 * config/s390/s390.c (s390_emit_prologue): Likewise.
6037 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6040 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6041 with a symbol for LRA.
6043 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6045 PR rtl-optimization/82338
6046 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6048 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6050 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6051 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6052 (qsort_chk): Declare.
6053 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6054 (qsort_chk): New function.
6056 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6058 PR tree-optimization/82337
6059 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6060 phi definition if the PHI result appears in an abnormal PHI.
6061 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6062 basis appears in an abnormal PHI.
6064 2017-09-29 Richard Biener <rguenther@suse.de>
6066 * graphite-isl-ast-to-gimple.c
6067 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6068 (binary_op_to_tree): Use it.
6069 (get_rename_from_scev): Likewise.
6070 (copy_loop_phi_nodes): Likewise.
6071 (copy_bb_and_scalar_dependences): Likewise.
6072 (translate_pending_phi_nodes): Likewise.
6074 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6077 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6078 for movabsq $(i32 << shift), r64.
6080 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6082 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6083 index when encoding %esp as %rsp to avoid 0x67 prefix.
6085 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6087 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6088 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6090 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6092 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6093 Extensions with more than 16 double VFP registers.
6094 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6095 to_clear_mask and all code related to it. Replace the remaining
6096 entry by a sbitmap and adapt code accordingly.
6098 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6100 * brig-builtins.def: Change pure attributes to const.
6102 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6104 * config.gcc (default_gnu_indirect_function): Default to yes for
6105 sparc*-*-linux* with glibc.
6107 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6109 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6110 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6111 when creating .init_array and .fini_array sections with priority
6114 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6117 * config/aarch64/aarch64.c
6118 (aarch64_builtin_support_vector_misalignment): Always return false
6119 when misalignment is unknown.
6121 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6123 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6124 this function to return false if the definition used by the swap
6125 instruction is artificial, or if the memory address from which the
6126 constant value is loaded is not represented by a base address held
6127 in a register or if the base address register is a frame or stack
6128 pointer. Additionally, return false if the base address of the
6129 loaded constant is a SYMBOL_REF but is not considered to be a
6131 (replace_swapped_load_constant): New function.
6132 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6133 loaded constant vector with a load of a swapped constant vector.
6135 2017-09-27 Carl Love <cel@us.ibm.com>
6137 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6138 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6139 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6142 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6144 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6145 first, always call autopref_rank_data otherwise.
6147 2017-09-27 Richard Biener <rguenther@suse.de>
6149 * graphite-scop-detection.c (find_scop_parameters): Move
6150 loop bound handling ...
6151 (gather_bbs::before_dom_children): ... here, avoiding the need
6152 to build scop_info->loop_nest.
6153 (record_loop_in_sese): Remove.
6154 * sese.h (sese_info_t::loop_nest): Remove.
6155 * sese.c (new_sese_info): Do not allocate loop_nest.
6156 (free_sese_info): Do not free loop_nest.
6158 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6161 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6162 lhs from calls if the lhs has addressable type.
6164 2017-09-27 Richard Biener <rguenther@suse.de>
6166 * graphite.h (scop::max_alias_set): New member.
6167 * graphite-scop-detection.c: Remove references to non-existing
6168 --param in comments.
6169 (build_alias_sets): Record the maximum alias set used for drs.
6170 (build_scops): Support zero as unlimited for
6171 --param graphite-max-arrays-per-scop.
6172 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6174 (build_poly_sr_1): ... here. Compute alias set based on the
6175 maximum alias set used for drs rather than
6176 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6178 2017-09-27 Richard Biener <rguenther@suse.de>
6180 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6181 --param loop-block-tile-size=0 to disable tiling.
6183 2017-09-27 Richard Biener <rguenther@suse.de>
6185 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6186 (graphite-max-nb-scop-params): Document special value zero.
6187 * domwalk.h (dom_walker::STOP): New symbolical constant.
6188 (dom_walker::dom_walker): Add optional parameter for bb to
6190 (dom_walker::~dom_walker): Declare.
6191 (dom_walker::before_dom_children): Document STOP return value.
6192 (dom_walker::m_user_bb_to_rpo): New member.
6193 (dom_walker::m_bb_to_rpo): Likewise.
6194 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6195 mapping here if not provided by the user.
6196 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6197 provided by the user.
6198 (dom_walker::STOP): Define.
6199 (dom_walker::walk): Do not compute bb to RPO mapping here.
6200 Support STOP return value from before_dom_children to stop
6202 * graphite-optimize-isl.c (optimize_isl): If the schedule
6203 is the same still generate code if -fgraphite-identity
6204 or -floop-parallelize-all are given.
6205 * graphite-scop-detection.c: Include cfganal.h.
6206 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6208 (gather_bbs::before_dom_children): Return STOP for BBs
6210 (build_scops): Compute bb to RPO mapping and pass it to
6211 the domwalk. Treat --param graphite-max-nb-scop-params=0
6212 as not limiting the number of params.
6213 * graphite.c (graphite_initialize): Remove limit on the
6214 number of basic-blocks in a function.
6215 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6216 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6217 default value of 10.
6219 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6221 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6222 Adjust code to eliminate needing to do the shift right 32-bits
6223 operation after XSCVDPSPN.
6225 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6227 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6228 ((X / Y) != 0 -> X >= Y): Likewise.
6230 2017-09-26 Carl Love <cel@us.ibm.com>
6232 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6233 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6234 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6235 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6236 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6237 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6238 definitions and overloading.
6239 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6240 statement for P9V_BUILTIN_XST_LEN_R.
6241 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6242 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6243 define_expand and define_insn for the instructions and builtins.
6244 * doc/extend.texi: Update the built-in documentation file for the new
6246 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6247 define_insn for the instructions
6249 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6252 * gcc/config/netbsd-protos.h: New file.
6253 * gcc/config/netbsd.c: New file.
6254 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6255 * gcc/config/t-netbsd: New file.
6256 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6257 (tmake_file) Add t-netbsd.
6258 (extra_objs) Add netbsd.o.
6260 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6264 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6266 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6268 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6269 sign extension from a vector register to a GPR by doing a 32-bit
6270 direct move and then an EXTSW.
6271 (extendsi<mode>2 splitter): Likewise.
6272 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6273 right or vector extract after doing XSCVDPSPN. Use
6274 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6276 (movdi_from_sf_zero_ext): Likewise.
6277 (reload_gpr_from_vsxsf): Likewise.
6278 (p8_mfvsrd_4_disf): Delete, no longer used.
6279 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6280 then needing to move the SFmode to a GPR to use the XSCVDPSP
6281 instruction instead of FRSP and XSCVDPSPN.
6282 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6283 it is adjacent to the other XSCVSPDP insns.
6284 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6285 SFmode to be in traditional Altivec registers.
6286 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6287 (vsx_xscvspdpn): Likewise.
6288 (vsx_xscvspdpn_scalar): Likewise.
6290 2017-09-26 Martin Jambor <mjambor@suse.cz>
6292 * tree-sra.c (compare_access_positions): Put integral types first,
6293 stabilize sorting of integral types, remove conditions putting
6294 non-full-precision integers last.
6295 (sort_and_splice_var_accesses): Disable scalarization if a
6296 non-integert would be represented by a non-full-precision integer.
6298 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6300 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6301 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6302 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6303 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6304 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6305 conditionals inside the function instead of around it. Call
6306 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6307 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6309 2017-09-26 Richard Biener <rguenther@suse.de>
6311 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6313 (scop_detection::build_scop_breadth): ... this. Removed.
6314 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6315 (scop_detection::harmful_stmt_in_bb): Likewise.
6316 (scop_detection::graphite_can_represent_stmt): Likewise.
6317 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6318 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6319 (scop_detection::can_represent_loop_1): ... this. Removed.
6320 (scop_detection::harmful_loop_in_region): Simplify after inlining
6321 the above and remove more quadraticness.
6322 (build_scops): Adjust.
6323 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6326 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6329 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6330 REGNO (base) == SP_REG if base is a REG.
6333 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6334 if it is different SSA_NAME.
6335 (optimize_range_tests_cmp_bitwise): New function.
6336 (optimize_range_tests): Call it.
6338 2017-09-26 Richard Biener <rguenther@suse.de>
6340 PR tree-optimization/82321
6341 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6342 for the def being inside the loop.
6344 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6346 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6348 * config/s390/s390-builtins.def: Fix constraint on op4.
6350 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6352 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6353 independent expanders.
6354 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6355 ("vec_ordered", "vec_unordered"): New expanders.
6357 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6359 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6362 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6364 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6365 vec_unpacks_lo_v16qi.
6366 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6368 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6370 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6371 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6372 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6374 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6376 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6378 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6380 ("*vec_slb<mode>"): New insn pattern.
6381 ("vec_shr_<mode>"): New expander.
6382 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6383 and force the shift count operand to V16QImode.
6384 ("vec_srb<mode>"): Set shift count mode to V16QI.
6386 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6388 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6389 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6390 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6392 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6395 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6397 2017-09-26 Richard Biener <rguenther@suse.de>
6399 PR tree-optimization/82320
6400 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6403 2017-09-25 Jeff Law <law@redhat.com>
6405 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6406 prototype for new argument.
6407 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6408 mostly extracted from rs6000_emit_allocate_stack.
6409 (rs6000_emit_probe_stack_range_stack_clash): New function.
6410 (rs6000_emit_allocate_stack): Call
6411 rs6000_emit_probe_stack_range_stack_clash as needed.
6412 (rs6000_emit_probe_stack_range): Add additional argument
6413 to call to gen_probe_stack_range{si,di}.
6414 (output_probe_stack_range): New.
6415 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6416 (output_probe_stack_range_stack_clash): New.
6417 (rs6000_emit_prologue): Emit notes into dump file as requested.
6418 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6419 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6420 Add additional operand and pass it to output_probe_stack_range.
6422 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6424 PR tree-optimization/82163
6425 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6426 (checking_verify_loop_closed_ssa): New parameter.
6427 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6428 (check_loop_closed_ssa_stmt): Delete.
6429 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6430 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6431 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6434 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6436 * brig-builtins.def: Treat HSAIL barrier builtins as
6437 setjmp/longjump style functions.
6439 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6441 * target.def (constant_alignment): New hook.
6442 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6443 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6444 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6445 * doc/tm.texi: Regenerate.
6446 * targhooks.h (default_constant_alignment): Declare.
6447 (constant_alignment_word_strings): Likewise.
6448 * targhooks.c (default_constant_alignment): New function.
6449 (constant_alignment_word_strings): Likewise.
6450 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6451 instead of CONSTANT_ALIGNMENT.
6452 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6453 (force_const_mem): Likewise.
6454 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6455 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6456 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6457 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6458 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6460 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6461 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6462 constant_alignment_word_strings.
6463 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6464 (CONSTANT_ALIGNMENT): Likewise.
6465 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6466 (arm_constant_alignment): New function.
6467 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6468 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6469 constant_alignment_word_strings.
6470 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6471 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6472 constant_alignment_word_strings.
6473 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6474 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6475 (cris_constant_alignment): New function.
6476 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6477 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6478 (epiphany_constant_alignment): New function.
6479 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6480 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6481 constant_alignment_word_strings.
6482 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6483 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6484 constant_alignment_word_strings.
6485 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6486 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6487 constant_alignment_word_strings.
6488 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6489 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6490 * config/i386/i386.c (ix86_constant_alignment): Make static.
6491 Use the same interface as the target hook.
6492 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6493 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6494 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6495 constant_alignment_word_strings.
6496 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6497 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6498 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6499 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6500 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6501 constant_alignment_word_strings.
6502 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6503 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6504 constant_alignment_word_strings.
6505 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6506 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6507 constant_alignment_word_strings.
6508 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6509 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6511 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6512 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6513 * config/mips/mips.c (mips_constant_alignment): New function.
6514 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6515 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6516 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6517 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6518 (mmix_constant_alignment): Make static. Use the same interface
6520 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6521 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6522 constant_alignment_word_strings.
6523 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6524 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6525 constant_alignment_word_strings.
6526 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6527 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6528 constant_alignment_word_strings.
6529 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6530 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6531 (rs6000_constant_alignment): New function.
6532 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6533 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6534 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6535 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6536 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6537 (rs6000_constant_alignment): New function.
6538 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6539 * config/s390/s390.c (s390_constant_alignment): New function.
6540 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6541 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6542 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6543 constant_alignment_word_strings.
6544 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6545 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6546 (sparc_constant_alignment): New function.
6547 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6548 * config/spu/spu.c (spu_constant_alignment): New function.
6549 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6550 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6551 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6552 constant_alignment_word_strings.
6553 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6554 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6555 constant_alignment_word_strings.
6556 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6557 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6558 constant_alignment_word_strings.
6559 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6560 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6561 (visium_constant_alignment): New function.
6562 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6563 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6564 (xtensa_constant_alignment): New function.
6565 * system.h (CONSTANT_ALIGNMENT): Poison.
6567 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6569 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6570 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6571 (rs6000_builtin_valid_without_lhs): New helper function.
6572 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6573 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6575 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6577 * target.h (vec_perm_indices): Use unsigned short rather than
6579 (auto_vec_perm_indices): Likewise.
6580 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6581 Use unsigned int rather than unsigned char.
6582 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6584 2017-09-25 Richard Biener <rguenther@suse.de>
6586 * cfgloop.h (sort_sibling_loops): Declare.
6587 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6588 (sort_sibling_loops): New function sorting the sibling loop list
6590 * graphite.c (graphite_transform_loops): Sort sibling loops.
6592 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6594 * target.def (vec_perm_const_ok): Change sel parameter to
6596 * optabs-query.c (can_vec_perm_p): Update accordingly.
6597 * doc/tm.texi: Regenerate.
6598 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6599 auto_vec_perm_indices and remove separate nelt field.
6600 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6601 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6602 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6603 (aarch64_expand_vec_perm_const): Update accordingly.
6604 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6605 to vec_perm_indices.
6606 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6607 auto_vec_perm_indices and remove separate nelt field.
6608 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6609 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6610 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6612 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6613 to vec_perm_indices.
6614 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6615 sel to vec_perm_indices.
6616 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6617 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6618 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6620 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6623 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6626 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6627 on the FUNCTION_DECL function context if it has a DIE that is a
6630 2017-09-25 Richard Biener <rguenther@suse.de>
6632 PR tree-optimization/82285
6633 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6636 2017-09-25 Tom de Vries <tom@codesourcery.com>
6640 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6643 2017-09-25 Richard Biener <rguenther@suse.de>
6645 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6646 ISL errors other than isl_error_quota happen. Dump if the
6647 schedule is the same.
6648 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6649 errors instead of aborting inside ISL.
6651 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6654 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6655 of libgcc_eh for m64.
6656 * config/i386/darwin64.h: Likewise.
6658 2017-09-25 Richard Biener <rguenther@suse.de>
6661 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6662 attribute for incomplete types nor twice for complete ones.
6664 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6667 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6668 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6671 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6674 * config/i386/i386.opt (mprefer-avx256): Use
6675 ix86_target_flags variable.
6676 * config/i386/i386.c (ix86_target_string): Move
6677 -mprefer-avx256 to flag2_opts.
6679 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6682 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6683 and x != -1 | y != -1 into (x & y) != -1.
6685 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6687 * config.gcc: Add new case statement to set
6688 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6689 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6690 s390x-*-linux* case statements. Added aarch64 to the list of
6691 supported architectures.
6693 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6695 PR tree-optimization/82289
6696 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6697 STMT_VINFO_RELEVANT_P.
6699 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6700 Alan Hayward <alan.hayward@arm.com>
6701 David Sherwood <david.sherwood@arm.com>
6703 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6704 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6705 (extract_range_from_binary_expr_1): Don't call
6706 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6708 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6709 Alan Hayward <alan.hayward@arm.com>
6710 David Sherwood <david.sherwood@arm.com>
6712 * target.def (preferred_vector_alignment): New hook.
6713 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6715 * doc/tm.texi: Regenerate.
6716 * targhooks.h (default_preferred_vector_alignment): Declare.
6717 * targhooks.c (default_preferred_vector_alignment): New function.
6718 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6720 (DR_TARGET_ALIGNMENT): New macro.
6721 (aligned_access_p): Update commentary.
6722 (vect_known_alignment_in_bytes): New function.
6723 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6725 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6726 Calculate the misalignment based on the target alignment rather than
6728 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6729 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6730 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6731 the target alignment, rather than masking the element misalignment
6732 with the number of elements in a vector. Also use the target
6733 alignment when calculating the maximum number of peels.
6734 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6735 instead of TYPE_ALIGN_UNIT.
6736 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6737 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6738 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6739 (vect_create_data_ref_ptr): Likewise.
6740 (vect_setup_realignment): Realign by ANDing with
6741 -DR_TARGET_MISALIGNMENT.
6742 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6743 the number of peels based on DR_TARGET_ALIGNMENT.
6744 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6745 with the guaranteed alignment boundary when deciding whether
6747 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6748 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6749 (ensure_base_align): Remove stmt_info parameter. Get the
6750 target base alignment from DR_TARGET_ALIGNMENT.
6751 (vectorizable_store): Update call accordingly. Interpret
6752 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6754 (vectorizable_load): Likewise.
6756 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6757 Alan Hayward <alan.hayward@arm.com>
6758 David Sherwood <david.sherwood@arm.com>
6760 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6761 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6762 (vect_enhance_data_refs_alignment): Likewise.
6764 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6766 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6767 error. Only quit immediately if parsing is complete.
6768 (BEGIN): Initialize fatal_err and parse_done.
6769 (begin fpu, end fpu): Check number of arguments.
6770 (begin arch, end arch): Likewise.
6771 (begin cpu, end cpu): Likewise.
6772 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6773 (optalias): Likewise.
6775 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6777 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6778 * config/arm/arm-isa.h: Delete. Move definitions to ...
6779 * arm-cpus.in: ... here. Use new feature and fgroup values.
6780 * config/arm/arm.c (arm_option_override): Use lower case for feature
6782 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6783 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6784 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6786 (print_isa_bits_for): New function.
6787 (gen_isa): New function.
6788 (gen_comm_data): Use print_isa_bits_for.
6789 (define feature): New keyword.
6790 (define fgroup): New keyword.
6791 * config/arm/t-arm (TM_H): Remove.
6792 (GTM_H): Add arm-isa.h.
6793 (arm-isa.h): Add rule to generate file.
6794 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6795 case for feature bit names.
6797 2017-09-22 Richard Biener <rguenther@suse.de>
6799 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6801 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6802 print of no dependency loops ...
6803 * graphite.c (graphite_transform_loops): ... here.
6804 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6806 (same_close_phi_node, remove_duplicate_close_phi,
6807 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6808 (canonicalize_loop_closed_ssa): ... here and simplify.
6809 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6810 (optimize_isl): Use dump_printf_loc to tell when we stopped
6811 optimizing because of an ISL timeout.
6813 2017-09-22 Richard Biener <rguenther@suse.de>
6815 PR tree-optimization/82291
6816 * tree-if-conv.c (predicate_mem_writes): Make sure to
6817 remove writes in blocks predicated with false.
6819 2017-09-22 Richard Biener <rguenther@suse.de>
6821 * sese.c: Include cfganal.h.
6822 (if_region_set_false_region): Remove.
6823 (create_if_region_on_edge): Likewise.
6824 (move_sese_in_condition): Re-implement without destroying
6827 2017-09-22 Richard Biener <rguenther@suse.de>
6829 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6830 Verify both BBs contain loop PHI nodes before dispatching to
6832 (graphite_regenerate_ast_isl): Do not recompute dominators,
6833 do not verify three times. Restructure for clarity.
6834 * graphite-scop-detection.c (same_close_phi_node,
6835 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6836 defined_in_loop_p, canonicalize_loop_closed_ssa,
6837 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6838 checking and SSA rewrite, move to ...
6839 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6840 (graphite_initialize): Do not pass in ctx, do not reset the
6841 SCEV cache, compute only dominators.
6842 (graphite_transform_loops): Allocate ISL ctx after
6843 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6844 Maintain post-dominators only around build_scops.
6845 * sese.c (if_region_set_false_region): Make static. Free
6846 and recompute dominators.
6847 (move_sese_in_condition): Assert we don't get called with
6848 post-dominators computed.
6849 * sese.h (if_region_set_false_region): Remove.
6851 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6853 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6854 mode attribute for TARGET_AVX512VL.
6856 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6858 * config/i386/i386.opt (mprefer-avx256): New option.
6859 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6861 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6862 for TARGET_PREFER_AVX256.
6863 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6865 2017-09-21 Jeff Law <law@redhat.com>
6867 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6868 Fix dump output if the only stack space is for pushed registers.
6870 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6872 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6875 2017-09-21 Martin Sebor <msebor@redhat.com>
6878 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6879 code (in C++) or code that triggers warnings.
6881 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6883 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6885 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6887 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6888 * sched-rgn.c: Ditto.
6889 * sel-sched-ir.c: Ditto.
6891 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6893 * toplev.h (set_random_seed): Adjust return type.
6894 * toplev.c (init_local_tick): Move eager initialization of random_seed
6895 to get_random_seed. Adjust comment.
6896 (init_random_seed): Inline to get_random_seed, delete.
6897 (get_random_seed): Initialize random_seed lazily.
6898 (set_random_seed): Do not return previous value.
6899 (print_switch_value): Do not call get_random_seed.
6901 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6903 * cgraph.c (delete_function_version): New, broken out from...
6904 (cgraph_node::delete_function_version): ...here. Rename to
6905 cgraph_node::delete_function_version_by_decl. Update all uses.
6906 (cgraph_node::remove): Call delete_function_version.
6908 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6911 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6912 VAR_DECLs to which addressable non-volatile parameters are mapped
6913 and for id->retvar after the return value assignment. Clear
6914 id->retval and id->retbnd after inlining.
6916 2017-09-21 Richard Biener <rguenther@suse.de>
6918 PR tree-optimization/82276
6919 PR tree-optimization/82244
6920 * tree-vrp.c (build_assert_expr_for): Set
6921 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6923 (remove_range_assertions): Revert earlier change.
6925 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6928 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6930 2017-09-21 Richard Biener <rguenther@suse.de>
6932 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6933 Restore valid IL after code generation errors.
6934 * graphite.c (graphite_transform_loops): Diagnose code
6935 generation issues as MSG_MISSED_OPTIMIZATION and continue
6936 with processing SCOPs.
6938 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6939 Alan Hayward <alan.hayward@arm.com>
6940 David Sherwood <david.sherwood@arm.com>
6942 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6943 rather than choosing between plus_constant and gen_rtx_<CODE>.
6944 * expr.c (emit_push_insn): Likewise.
6945 (expand_expr_real_2): Likewise.
6947 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6948 Alan Hayward <alan.hayward@arm.com>
6949 David Sherwood <david.sherwood@arm.com>
6951 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6953 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6954 Alan Hayward <alan.hayward@arm.com>
6955 David Sherwood <david.sherwood@arm.com>
6957 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6958 calling tree_to_uhwi.
6960 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6961 Alan Hayward <alan.hayward@arm.com>
6962 David Sherwood <david.sherwood@arm.com>
6964 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6965 INTEGER_CST rather than a negative test for ADDR_EXPR.
6967 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6968 Alan Hayward <alan.hayward@arm.com>
6969 David Sherwood <david.sherwood@arm.com>
6971 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6972 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6974 2017-09-21 Richard Biener <rguenther@suse.de>
6976 PR tree-optimization/71351
6977 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6978 graphite_create_new_loop_guard): Remove, fold remaining parts
6980 (translate_isl_ast_node_for): ... here and simplify.
6982 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6985 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6986 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6987 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6988 alternative always use QI mode, for -Os imov (=R,R) alternative
6989 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6992 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6993 Jeff Law <law@redhat.com>
6995 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6996 (allocate_stack_space): New function, partially extracted from
6998 (s390_emit_prologue): Track offset to most recent stack probe.
6999 Code to allocate space moved into allocate_stack_space.
7000 Dump actions when no stack is allocated.
7001 (s390_prologue_plus_offset): New function.
7002 (s390_emit_stack_probe): Likewise.
7004 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7006 * common.opt (Wa, Wl, Wp, g, gz=): Add
7008 (gno-column-info): Remove.
7009 (gcolumn-info): Drop RejectNegative.
7011 (gno-record-gcc-switches): Remove.
7012 (grecord-gcc-switches): Drop RejectNegative.
7013 (gno-split-dwarf): Remove.
7014 (gsplit-dwarf): Drop RejectNegative.
7015 (gno-strict-dwarf): Remove.
7016 (gstrict-dwarf): Drop RejectNegative.
7017 * config/darwin.opt (gfull, gused): Add RejectNegative.
7018 * dwarf2out.c (gen_producer_string): Drop
7019 gno-record-gcc-switches handler.
7020 * optc-gen.awk: Add g to prefixes with negative forms.
7021 * opts-common.c (remapping_prefix_p): New.
7022 (find_opt): Check it.
7023 (generate_canonical_option): Test g prefix.
7024 (option_map): Add -gno- mapping.
7025 (add_misspelling_candidates): Check remapping_prefix_p.
7027 2017-09-20 Jeff Law <law@redhat.com>
7029 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7030 thinko in stack clash protection support.
7032 * explow.c (compute_stack_clash_protection_loop_data): Use
7033 CONST_INT_P instead of explicit test. Verify object is a
7034 CONST_INT_P before looking at INTVAL.
7035 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7036 instead of explicit test.
7038 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7041 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7042 address instead of to r1 and r11.
7044 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7046 * config.gcc: Support "knm".
7047 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7048 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7050 * config/i386/i386.c (m_KNM): Define.
7051 (processor_target_table): Add "knm".
7053 (ix86_option_override_internal): Add "knm".
7054 (ix86_issue_rate): Add PROCESSOR_KNM.
7055 (ix86_adjust_cost): Ditto.
7056 (ia32_multipass_dfa_lookahead): Ditto.
7057 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7058 (fold_builtin_cpu): Add M_INTEL_KNM.
7059 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7060 (processor_type): Add PROCESSOR_KNM.
7061 * config/i386/x86-tune.def: Add m_KNM.
7062 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7064 2017-09-20 Richard Biener <rguenther@suse.de>
7066 PR tree-optimization/80213
7067 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7068 are allowed in empty BBs as well.
7069 (canonicalize_loop_closed_ssa): Also look for other complex
7071 (scop_detection::get_sese): Include the loop-closed PHI block
7073 (scop_detection::merge_sese): Remove code adding extra blocks.
7074 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7075 (build_scops): Assert the final returned scop is invalid.
7077 2017-09-20 Richard Biener <rguenther@suse.de>
7079 PR tree-optimization/82264
7080 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7082 (vn_phi_lookup): Likewise.
7083 (vn_phi_insert): Likewise.
7085 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7087 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7088 that fits into uhwi or shwi, add DW_AT_const_value regardless
7089 of early_dwarf without going through RTL, using add_AT_unsigned
7092 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7093 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7094 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7095 (DEBUG_MACRO_SECTION): Likewise.
7096 (DEBUG_LTO_MACRO_SECTION): Likewise.
7097 (DEBUG_STR_DWO_SECTION): Likewise.
7098 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7099 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7100 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7101 (DEBUG_LTO_LINE_STR_SECTION): Define.
7102 (init_sections_and_labels): Initialize debug_line_str_section
7103 variable. Initialize debug_loc_section for -gdwarf-5 to
7104 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7106 2017-09-20 Richard Biener <rguenther@suse.de>
7108 * graphite-sese-to-poly.c (extract_affine): Properly handle
7109 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7111 2017-09-20 Richard Biener <rguenther@suse.de>
7113 PR tree-optimization/81373
7114 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7115 Force SESE live-out defs to be handled even if they are
7118 2017-09-19 Jeff Law <law@redhat.com>
7120 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7121 nothing for stack adjustments with REG_STACK_CHECK.
7122 * sched-deps.c (parse_add_or_inc): Reject insns with
7123 REG_STACK_CHECK from dependency breaking.
7124 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7125 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7126 * reg-notes.def (STACK_CHECK): New note.
7128 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7129 (ix86_expand_prologue): Dump stack clash info as needed.
7130 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7132 * function.c (dump_stack_clash_frame_info): New function.
7133 * function.h (dump_stack_clash_frame_info): Prototype.
7134 (enum stack_clash_probes): New enum.
7136 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7137 flag_stack_clash_protection.
7138 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7139 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7140 (arm_frame_pointer_required): Likewise.
7141 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7142 (ia64_expand_prologue): Likewise.
7143 * config/mips/mips.c (mips_expand_prologue): Likewise.
7144 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7145 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7146 (sparc_flat_expand_prologue): Likewise.
7147 * config/spu/spu.c (spu_expand_prologue): Likewise.
7149 * explow.c: Include "params.h".
7150 (anti_adjust_stack_and_probe_stack_clash): New function.
7151 (get_stack_check_protect): Likewise.
7152 (compute_stack_clash_protection_loop_data): Likewise.
7153 (emit_stack_clash_protection_loop_start): Likewise.
7154 (emit_stack_clash_protection_loop_end): Likewise.
7155 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7156 Use anti_adjust_stack_and_probe_stack_clash.
7157 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7158 (emit_stack_clash_protection_loop_start): Likewise.
7159 (emit_stack_clash_protection_loop_end): Likewise.
7160 * rtl.h (get_stack_check_protect): Prototype.
7161 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7162 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7163 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7165 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7167 * doc/tm.texi: Rebuilt.
7168 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7169 get_stack_check_protect.
7170 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7171 * config/arm/arm.c (arm_expand_prologue): Likewise.
7172 (arm_frame_pointer_required): Likewise.
7173 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7174 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7175 * config/mips/mips.c (mips_expand_prologue): Likewise.
7176 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7177 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7178 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7179 (sparc_flat_expand_prologue): Likewise.
7181 * common.opt (-fstack-clash-protection): New option.
7182 * flag-types.h (enum stack_check_type): Note difference between
7183 -fstack-check= and -fstack-clash-protection.
7184 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7185 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7186 * toplev.c (process_options): Issue warnings/errors for cases
7187 not handled with -fstack-clash-protection.
7188 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7189 (-fstack-check): Note additional problem with -fstack-check=generic.
7190 Note that -fstack-check is primarily for Ada and refer users
7191 to -fstack-clash-protection for stack-clash-protection.
7192 Document new params for stack clash protection.
7194 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7196 * config/i386/i386.c (ix86_split_long_move): Do not handle
7197 address used for LEA in a special way.
7199 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7201 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7203 2017-09-19 Martin Sebor <msebor@redhat.com>
7206 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7207 of incompatible types.
7209 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7211 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7212 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7213 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7214 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7216 2017-09-19 Richard Biener <rguenther@suse.de>
7218 PR tree-optimization/82244
7219 * tree-vrp.c (remove_range_assertions): Do not propagate
7220 a constant to abnormals but replace the assert with a copy.
7222 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7224 PR rtl-optimization/57878
7225 PR rtl-optimization/68988
7226 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7227 avoidance test involving non_reload_pseudos. Move frequency test
7228 below the general fragmentation avoidance test.
7230 2017-09-19 Richard Biener <rguenther@suse.de>
7232 PR tree-optimization/69728
7233 * graphite-sese-to-poly.c (schedule_error): New global.
7234 (add_loop_schedule): Handle empty domain by failing the
7236 (build_original_schedule): Handle schedule_error.
7238 2017-09-19 Richard Biener <rguenther@suse.de>
7240 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7241 Do not iterate to sibling loops but only to siblings of inner
7244 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7247 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7248 that the registers are different.
7250 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7252 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7253 to processor_model and "amdfam17h" to arch_names_table.
7254 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7256 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7259 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7261 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7262 Alan Hayward <alan.hayward@arm.com>
7263 David Sherwood <david.sherwood@arm.com>
7265 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7267 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7269 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7270 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7271 vect_schedule_slp_instance.
7272 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7273 Update call to vect_slp_analyze_node_operations. Simplify return
7275 (vect_slp_analyze_bb_1): Update call accordingly.
7276 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7277 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7278 (vect_schedule_slp): Update call accordingly.
7280 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7281 Alan Hayward <alan.hayward@arm.com>
7282 David Sherwood <david.sherwood@arm.com>
7284 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7285 with types that aren't in fact scalar.
7287 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7289 * tree-vect-slp.c (vect_record_max_nunits): New function,
7291 (vect_build_slp_tree_1): ...here.
7292 (vect_build_slp_tree_2): Call it for phis too.
7294 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7296 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7297 to vect_get_vec_def_for_operand when getting the mask operand.
7299 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7300 Alan Hayward <alan.hayward@arm.com>
7301 David Sherwood <david.sherwood@arm.com>
7303 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7306 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7307 Alan Hayward <alan.hayward@arm.com>
7308 David Sherwood <david.sherwood@arm.com>
7310 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7311 calculation for vector booleans.
7313 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7314 Alan Hayward <alan.hayward@arm.com>
7315 David Sherwood <david.sherwood@arm.com>
7317 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7319 (vect_transform_stmt): ...here.
7320 (vect_analyze_stmt): Use it instead of calling
7321 vectorizable_live_operation directly.
7323 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7325 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7326 non-SIMT targets in acc vector loops.
7328 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7330 * configure.ac: Add arc and check if assembler supports gdwarf2.
7331 * configure: Regenerate.
7333 2017-09-18 Richard Biener <rguenther@suse.de>
7335 PR tree-optimization/82220
7336 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7337 epilogue niters from the min_profitable_iters compute.
7339 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7342 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7343 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7344 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7346 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7349 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7350 switching to a new text section.
7352 2017-09-18 Richard Biener <rguenther@suse.de>
7354 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7356 (build_alias_set): Reject aliases with no access function.
7358 2017-09-18 Richard Biener <rguenther@suse.de>
7360 PR tree-optimization/79622
7361 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7363 (build_cross_bb_scalars_use): Likewise.
7365 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7367 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7369 2017-09-18 Alan Modra <amodra@gmail.com>
7372 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7373 stack_pointer_rtx for count 0. Update comments. Break up
7374 large rtl expression.
7376 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7378 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7379 Increase to 20 bytes.
7380 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7381 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7382 or avx version of the stub.
7384 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7387 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7388 compute the minimum stack alignment. Also update preferred stack
7389 boundary for leaf functions.
7391 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7393 PR tree-optimization/82228
7394 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7397 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7399 * common/config/nds32/nds32-common.c
7400 (nds32_option_optimization_table): Refine formatting.
7401 (nds32_option_optimization_table): Use -fsched-pressure and
7402 -fomit-frame-pointer for specific optimization level.
7404 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7406 * config/nds32/nds32.c: Refine formatting and comments.
7407 * config/nds32/nds32.h: Likewise.
7408 * config/nds32/nds32.md: Likewise.
7409 * config/nds32/nds32-cost.c: Likewise.
7410 * config/nds32/nds32-isr.c: Likewise.
7411 * config/nds32/nds32-md-auxiliary.c: Likewise.
7412 * config/nds32/nds32-multiple.md: Likewise.
7413 * config/nds32/nds32-predicates.c: Likewise.
7415 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7416 Jakub Jelinek <jakub@redhat.com>
7418 Add support for -std=c++2a.
7419 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7421 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7423 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7426 * doc/extend.texi (Common Function Attributes): Add
7427 references to ARM, AArch64, and S/390 specific attributes.
7428 (Function Specific Option Pragmas): Add AArch64 and S/390
7429 to list of back ends that support the target pragma.
7431 2017-09-15 Nathan Sidwell <nathan@acm.org>
7433 * doc/standards.texi: Fix C++17 description. Update URLs for
7436 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7438 * common.opt (Wcast-align=strict): New warning option.
7439 * doc/invoke.texi: Document -Wcast-align=strict.
7441 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7443 * cgraph.h (cgraph_thunk_info): Add comments.
7444 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7445 assert for VIRTUAL_* arguments stricter.
7447 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7449 PR tree-optimization/71026
7450 * match.pd: Move RDIV patterns from fold-const.c
7451 * fold-const.c (distribute_real_division): Removed.
7452 (fold_binary_loc): Remove calls to distribute_real_divison.
7454 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7456 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7457 c++1z and gnu++1z as deprecated. Change other references to
7458 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7459 Change -Wc++1z-compat to -Wc++17-compat.
7460 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7461 * dwarf2out.c (highest_c_language): Handle C++17.
7462 (gen_compile_unit_die): Likewise.
7464 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7466 PR rtl-optimization/82192
7467 * combine.c (make_extraction): Don't look through non-paradoxical
7468 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7471 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7472 Alan Hayward <alan.hayward@arm.com>
7473 David Sherwood <david.sherwood@arm.com>
7475 * target.def (function_arg_offset): New hook.
7476 * targhooks.h (default_function_arg_offset): Declare.
7477 * targhooks.c (default_function_arg_offset): New function.
7478 * function.c (locate_and_pad_parm): Use
7479 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7480 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7481 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7482 * doc/tm.texi: Regenerate.
7483 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7484 * config/spu/spu.c (spu_function_arg_offset): New function.
7485 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7486 * system.h (FUNCTION_ARG_OFFSET): Poison.
7488 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7489 Alan Hayard <alan.hayward@arm.com>
7490 David Sherwood <david.sherwood@arm.com>
7492 * target.def (truly_noop_truncation): New hook.
7493 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7494 than TRULY_NOOP_TRUNCATION.
7495 * hooks.h (hook_bool_uint_uint_true): Declare.
7496 * hooks.c (hook_bool_uint_uint_true): New function.
7497 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7498 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7499 * doc/tm.texi: Regenerate.
7500 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7501 rather than TRULY_NOOP_TRUNCATION in comments.
7502 (simplify_comparison): Likewise.
7503 (record_truncated_value): Likewise.
7504 * expmed.c (extract_bit_field_1): Likewise.
7505 (extract_split_bit_field): Likewise.
7506 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7507 instead of TRULY_NOOP_TRUNCATION.
7508 * function.c (assign_parm_setup_block): Likewise.
7509 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7510 * rtlhooks.c: Include target.h.
7511 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7512 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7513 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7514 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7515 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7516 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7517 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7518 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7519 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7520 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7521 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7522 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7523 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7524 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7525 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7526 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7527 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7528 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7529 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7530 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7531 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7532 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7533 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7534 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7535 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7536 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7537 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7538 rather than TRULY_NOOP_TRUNCATION in comments.
7539 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7540 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7541 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7542 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7543 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7544 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7545 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7546 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7547 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7548 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7549 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7550 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7551 rather than TRULY_NOOP_TRUNCATION in comments.
7552 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7553 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7554 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7555 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7556 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7557 TRULY_NOOP_TRUNCATION condition.
7558 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7559 (TRULY_NOOP_TRUNCATION): Delete.
7560 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7561 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7562 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7563 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7564 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7565 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7566 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7567 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7568 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7569 rather than TRULY_NOOP_TRUNCATION in comments.
7570 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7571 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7572 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7573 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7574 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7575 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7577 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7580 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7581 (*cmp_ior): Likewise.
7582 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7583 (*ior_scc_scc_cmp): Likewise.
7584 (*and_scc_scc): Likewise.
7585 (*and_scc_scc_cmp): Likewise.
7587 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7588 Alan Hayard <alan.hayward@arm.com>
7589 David Sherwood <david.sherwood@arm.com>
7591 * target.def (can_change_mode_class): New hook.
7592 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7593 (hard_regno_nregs): Likewise.
7594 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7595 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7596 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7597 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7598 (LOAD_EXTEND_OP): Update accordingly.
7599 * doc/tm.texi: Regenerate.
7600 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7601 CANNOT_CHANGE_MODE_CLASS.
7602 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7603 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7604 * combine.c (simplify_set): Update accordingly.
7605 * emit-rtl.c (validate_subreg): Likewise.
7606 * recog.c (general_operand): Likewise.
7607 * regcprop.c (mode_change_ok): Likewise.
7608 * reload1.c (choose_reload_regs): Likewise.
7609 (inherit_piecemeal_p): Likewise.
7610 * rtlanal.c (simplify_subreg_regno): Likewise.
7611 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7612 instead of CANNOT_CHANGE_MODE_CLASS.
7613 (reload_cse_simplify_operands): Likewise.
7614 * reload.c (push_reload): Use targetm.can_change_mode_class
7615 instead of CANNOT_CHANGE_MODE_CLASS.
7616 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7617 REG_CANNOT_CHANGE_MODE_P.
7618 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7619 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7620 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7621 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7622 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7623 (arm_can_change_mode_class): New function.
7624 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7625 than CANNOT_CHANGE_MODE_CLASS in comments.
7626 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7627 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7628 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7629 (ix86_can_change_mode_class): ...this new function, inverting the
7630 sense of the return value.
7631 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7632 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7633 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7634 (ia64_can_change_mode_class): New function.
7635 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7636 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7637 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7638 (m32c_can_change_mode_class): ...this new function, inverting the
7639 sense of the return value.
7640 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7641 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7642 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7643 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7644 (mips_can_change_mode_class): ...this new function, inverting the
7645 sense of the return value.
7646 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7647 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7648 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7649 (msp430_can_change_mode_class): New function.
7650 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7651 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7652 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7653 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7654 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7655 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7656 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7657 (pa_cannot_change_mode_class): Replace with...
7658 (pa_can_change_mode_class): ...this new function, inverting the
7659 sense of the return value.
7660 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7661 than CANNOT_CHANGE_MODE_CLASS in comments.
7662 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7663 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7664 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7665 (pdp11_cannot_change_mode_class): Replace with...
7666 (pdp11_can_change_mode_class): ...this new function, inverting the
7667 sense of the return value.
7668 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7669 * config/powerpcspe/powerpcspe-protos.h
7670 (rs6000_cannot_change_mode_class_ptr): Delete.
7671 * config/powerpcspe/powerpcspe.c
7672 (rs6000_cannot_change_mode_class_ptr): Delete.
7673 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7674 (rs6000_option_override_internal): Assign to
7675 targetm.can_change_mode_class instead of
7676 rs6000_cannot_change_mode_class_ptr.
7677 (rs6000_cannot_change_mode_class): Replace with...
7678 (rs6000_can_change_mode_class): ...this new function, inverting the
7679 sense of the return value.
7680 (rs6000_debug_cannot_change_mode_class): Replace with...
7681 (rs6000_debug_can_change_mode_class): ...this new function.
7682 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7683 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7684 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7685 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7686 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7688 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7689 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7690 (rs6000_option_override_internal): Assign to
7691 targetm.can_change_mode_class instead of
7692 rs6000_cannot_change_mode_class_ptr.
7693 (rs6000_cannot_change_mode_class): Replace with...
7694 (rs6000_can_change_mode_class): ...this new function, inverting the
7695 sense of the return value.
7696 (rs6000_debug_cannot_change_mode_class): Replace with...
7697 (rs6000_debug_can_change_mode_class): ...this new function.
7698 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7699 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7700 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7701 (s390_can_change_mode_class): ...this new function, inverting the
7702 sense of the return value.
7703 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7704 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7705 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7706 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7707 (sh_cannot_change_mode_class): Replace with...
7708 (sh_can_change_mode_class): ...this new function, inverting the
7709 sense of the return value.
7710 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7711 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7712 (sparc_can_change_mode_class): New function.
7713 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7714 * config/spu/spu.c (spu_can_change_mode_class): New function.
7715 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7716 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7717 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7718 (visium_can_change_mode_class): New function.
7719 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7721 2017-09-15 Richard Biener <rguenther@suse.de>
7723 PR tree-optimization/82217
7724 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7725 but not undefined case.
7727 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7730 * postreload.c (reload_cse_simplify_operands): Skip
7731 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7733 2017-09-15 Richard Biener <rguenther@suse.de>
7735 PR tree-optimization/68823
7736 * graphite-scop-detection.c (build_alias_set): If we have a
7737 possible dependence check whether we can handle them by just
7738 looking at the DRs DR_ACCESS_FNs.
7739 (build_scops): If build_alias_set fails, fail the SCOP.
7741 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7743 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7744 to support float128 built-in functions that require the ISA 3.0
7746 (BU_FLOAT128_3_HW): Likewise.
7747 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7749 (FMAF128): Likewise.
7751 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7752 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7753 floating point instructions.
7754 (rs6000_invalid_builtin): Likewise.
7755 (rs6000_builtin_mask_names): Likewise.
7756 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7757 (RS6000_BTM_FLOAT128_HW): Likewise.
7758 (RS6000_BTM_COMMON): Likewise.
7759 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7761 * doc/extend.texi (RS/6000 built-in functions): Document the
7762 IEEE 128-bit floating point square root and fused multiply-add
7765 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7767 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7768 reg (r2) isn't in the set of registers defined in the prologue.
7770 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7771 Alan Hayward <alan.hayward@arm.com>
7772 David Sherwood <david.sherwood@arm.com>
7774 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7775 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7776 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7777 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7778 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7780 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7781 max_vectorization_factor.
7782 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7784 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7785 Alan Hayward <alan.hayward@arm.com>
7786 David Sherwood <david.sherwood@arm.com>
7788 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7789 (vect_worthwhile_without_simd_p): Declare.
7790 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7791 (vectorizable_reduction): Use it.
7792 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7793 (vectorizable_operation): Likewise.
7795 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7796 Alan Hayward <alan.hayward@arm.com>
7797 David Sherwood <david.sherwood@arm.com>
7799 * tree-vectorizer.h (vect_get_num_copies): New function.
7800 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7801 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7802 (vectorizable_induction): Likewise.
7803 (vectorizable_live_operation): Likewise.
7804 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7805 (vectorizable_bswap): Likewise.
7806 (vectorizable_call): Likewise.
7807 (vectorizable_conversion): Likewise.
7808 (vectorizable_assignment): Likewise.
7809 (vectorizable_shift): Likewise.
7810 (vectorizable_operation): Likewise.
7811 (vectorizable_store): Likewise.
7812 (vectorizable_load): Likewise.
7813 (vectorizable_condition): Likewise.
7814 (vectorizable_comparison): Likewise.
7815 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7817 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7818 Alan Hayward <alan.hayward@arm.com>
7819 David Sherwood <david.sherwood@arm.com>
7821 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7822 of vect_init_vector.
7824 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7825 Alan Hayward <alan.hayward@arm.com>
7826 David Sherwood <david.sherwood@arm.com>
7828 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7829 an inline wrapper that provides a location.
7830 (gimple_build_vector): Likewise.
7831 * gimple-fold.c (gimple_build_vector_from_val): New function.
7832 (gimple_build_vector): Likewise.
7833 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7834 functions to build the initial value. Always return a gimple value.
7835 (get_initial_defs_for_reduction): Likewise. Only compute
7837 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7838 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7839 (vectorizable_induction): Use gimple_build_vector rather than
7842 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7843 Alan Hayward <alan.hayward@arm.com>
7844 David Sherwood <david.sherwood@arm.com>
7846 * target.h (vec_perm_indices): New typedef.
7847 (auto_vec_perm_indices): Likewise.
7848 * optabs-query.h: Include target.h
7849 (can_vec_perm_p): Take a vec_perm_indices *.
7850 * optabs-query.c (can_vec_perm_p): Likewise.
7851 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7852 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7853 * tree-vect-generic.c (lower_vec_perm): Likewise.
7854 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7855 (vect_grouped_load_supported): Likewise.
7856 (vect_shift_permute_load_chain): Likewise.
7857 (vect_permute_store_chain): Use auto_vec_perm_indices.
7858 (vect_permute_load_chain): Likewise.
7859 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7860 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7861 Update uses of can_vec_perm_p.
7862 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7863 mode with a number of elements. Take a vec_perm_indices *.
7864 (vect_create_epilog_for_reduction): Update accordingly.
7865 Use auto_vec_perm_indices.
7866 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7867 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7868 (vect_transform_slp_perm_load): Likewise.
7869 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7870 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7871 (vect_gen_perm_mask_checked): Likewise.
7872 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7873 (vect_gen_perm_mask_checked): Likewise.
7874 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7875 (vectorizable_store): Likewise.
7876 (vectorizable_load): Likewise.
7877 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7878 (vectorizable_bswap): Likewise.
7880 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7881 Alan Hayward <alan.hayward@arm.com>
7882 David Sherwood <david.sherwood@arm.com>
7884 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7885 * tree.c (build_vector): Likewise.
7886 (build_vector_from_ctor): Update accordingly.
7887 (build_vector_from_val): Likewise.
7888 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7889 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7890 * tree-vect-generic.c (add_rshift): Likewise.
7891 (expand_vector_divmod): Likewise.
7892 (optimize_vector_constructor): Likewise.
7893 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7894 (vect_transform_slp_perm_load): Likewise.
7895 (vect_schedule_slp_instance): Likewise.
7896 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7897 (vectorizable_call): Likewise.
7898 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7899 * expmed.c (make_tree): Likewise.
7900 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7901 a vector passed to build_vector.
7902 (fold_convert_const): Likewise.
7903 (exact_inverse): Likewise.
7904 (fold_ternary_loc): Likewise.
7905 (fold_relational_const): Likewise.
7906 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7907 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7908 (const_unop): Likewise. Store the reduction accumulator in a
7909 variable rather than an array.
7910 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7911 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7912 the new vector, rather than constructing it after the input arrays.
7913 (native_interpret_vector): Use auto_vec<tree> when building
7914 a vector passed to build_vector. Add elements in order.
7915 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7916 auto_vec<tree> when building a vector passed to build_vector.
7917 (vect_create_epilog_for_reduction): Likewise.
7918 (vectorizable_induction): Likewise.
7919 (get_initial_def_for_reduction): Likewise. Fix indentation of
7921 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7923 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7924 passed to build_vector.
7926 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7927 Alan Hayward <alan.hayward@arm.com>
7928 David Sherwood <david.sherwood@arm.com>
7930 * tree-core.h (tree_base::u): Add an "nelts" field.
7931 (tree_vector): Use VECTOR_CST_NELTS as the length.
7932 * tree.c (tree_size): Likewise.
7933 (make_vector): Initialize VECTOR_CST_NELTS.
7934 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7935 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7936 TYPE_VECTOR_SUBPARTS.
7937 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7938 as the number of units, setting it from VECTOR_CST_NELTS.
7939 (const_vector_from_tree): Likewise.
7940 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7941 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7942 (fold_negate_expr_1): Likewise.
7943 (fold_convert_const): Likewise.
7944 (const_binop): Likewise. Differentiate the number of output and
7946 (const_unop): Likewise.
7947 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7948 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7949 in cases that did the opposite.
7951 2017-09-14 Richard Biener <rguenther@suse.de>
7953 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7956 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7958 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7960 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7963 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7964 if and where to split a bb, except for splitting before debug insn
7965 sequences followed by non-label real insn. Delete debug insns
7966 in between basic blocks.
7968 * combine.c (make_compound_operation_int): Formatting fixes.
7970 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7971 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7972 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7973 * config/sol2.h (LINK_EH_SPEC): Likewise.
7974 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7975 * config/s390/linux.h (LINK_SPEC): Likewise.
7976 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7977 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7978 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7979 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7980 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7981 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7982 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7983 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7985 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7987 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7989 (ENDFILE_LINUX_SPEC): Likewise.
7990 (LINK_EH_SPEC): Likewise.
7991 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7992 (LINK_OS_LINUX_SPEC32): Likewise.
7993 (LINK_OS_LINUX_SPEC64): Likewise.
7994 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7995 (LINK_OS_LINUX_SPEC): Likewise.
7997 2017-09-13 Martin Liska <mliska@suse.cz>
8000 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8001 CASE_HIGH is NULL_TREE.
8003 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8004 Alan Hayward <alan.hayward@arm.com>
8005 David Sherwood <david.sherwood@arm.com>
8007 * target.def (secondary_memory_needed): New hook.
8008 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8009 instead of SECONDARY_MEMORY_NEEDED.
8010 (secondary_memory_needed_mode): Likewise.
8011 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8012 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8013 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8014 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8015 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8016 * doc/tm.texi: Regenerate.
8017 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8018 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8019 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8020 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8021 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8022 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8023 mode argument first and change the reg_class arguments to reg_class_t.
8024 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8025 Make static. Update the call to inline_secondary_memory_needed.
8026 (ix86_register_move_cost): Update the call to
8027 inline_secondary_memory_needed.
8028 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8029 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8031 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8032 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8034 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8035 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8036 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8037 and match hook interface. Add comment from mips.h.
8038 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8039 * config/mmix/mmix.md (truncdfsf2): Refer to
8040 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8042 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8043 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8044 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8045 (pa_secondary_memory_needed): New function.
8046 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8047 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8048 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8049 (pdp11_secondary_memory_needed): Make static and match hook interface.
8050 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8051 * config/powerpcspe/powerpcspe-protos.h
8052 (rs6000_secondary_memory_needed_ptr): Delete.
8053 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8055 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8056 (rs6000_option_override_internal): Assign to
8057 targetm.secondary_memory_needed rather than
8058 rs6000_secondary_memory_needed_ptr.
8059 (rs6000_secondary_memory_needed): Match hook interface.
8060 (rs6000_debug_secondary_memory_needed): Likewise.
8061 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8062 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8063 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8064 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8065 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8066 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8068 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8069 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8070 (rs6000_option_override_internal): Assign to
8071 targetm.secondary_memory_needed rather than
8072 rs6000_secondary_memory_needed_ptr.
8073 (rs6000_secondary_memory_needed): Match hook interface.
8074 (rs6000_debug_secondary_memory_needed): Likewise.
8075 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8076 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8077 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8078 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8079 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8080 (sparc_secondary_memory_needed): New function.
8081 * lra-constraints.c (check_and_process_move): Refer to
8082 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8084 (curr_insn_transform): Likewise.
8085 (process_alt_operands): Use targetm.secondary_memory_needed
8086 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8087 (check_secondary_memory_needed_p): Likewise.
8088 (choose_split_class): Likewise.
8089 * reload.c: Unconditionally include code that was previously
8090 conditional on SECONDARY_MEMORY_NEEDED.
8091 (push_secondary_reload): Use targetm.secondary_memory_needed
8092 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8093 (push_reload): Likewise.
8094 * reload1.c: Unconditionally include code that was previously
8095 conditional on SECONDARY_MEMORY_NEEDED.
8096 (choose_reload_regs): Use targetm.secondary_memory_needed
8097 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8098 (gen_reload): Likewise.
8099 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8101 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8102 Alan Hayward <alan.hayward@arm.com>
8103 David Sherwood <david.sherwood@arm.com>
8105 * target.def (secondary_memory_needed_mode): New hook:
8106 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8107 * targhooks.h (default_secondary_memory_needed_mode): New function.
8108 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8109 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8110 * doc/tm.texi: Regenerate.
8111 * lra-constraints.c (check_and_process_move): Use
8112 targetm.secondary_memory_needed_mode instead of
8113 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8114 (curr_insn_transform): Likewise.
8115 * reload.c (get_secondary_mem): Likewise.
8116 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8117 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8119 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8120 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8121 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8122 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8123 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8125 * config/powerpcspe/powerpcspe-protos.h
8126 (rs6000_secondary_memory_needed_mode): Delete.
8127 * config/powerpcspe/powerpcspe.c
8128 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8129 (rs6000_secondary_memory_needed_mode): Make static.
8130 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8131 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8133 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8135 (rs6000_secondary_memory_needed_mode): Make static.
8136 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8137 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8138 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8139 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8140 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8142 (sparc_secondary_memory_needed_mode): New function.
8143 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8145 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8147 * config/aarch64/constraints.md (Umq): New constraint.
8148 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8150 (mov<mode>): Update condition.
8152 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8154 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8155 when bitposition is the same.
8157 2017-09-13 Richard Biener <rguenther@suse.de>
8159 * dwarf2out.c (output_die_symbol): Remove.
8160 (output_die): Do not output a DIEs symbol.
8162 2017-09-13 Richard Biener <rguenther@suse.de>
8165 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8166 default-def to avoid breaking iterator update with the weird
8167 interaction with cgraph_update_edges_for_call_stmt_node.
8169 2017-09-13 Richard Biener <rguenther@suse.de>
8171 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8172 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8173 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8174 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8176 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8178 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8179 Disable pc relative literal load irrespective of
8180 TARGET_FIX_ERR_A53_84341 for default.
8182 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8184 * config/sparc/sparc.c (output_return): Output the source location of
8185 the insn in the delay slot, if any.
8186 (output_sibcall): Likewise.
8188 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8191 * common.opt (-static-pie): New alias.
8192 (shared): Negate static-pie.
8193 (-no-pie): Update help text.
8195 (static-pie): New option.
8196 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8197 -static-pie support.
8198 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8199 (LINK_EH_SPEC): Likewise.
8200 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8201 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8202 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8203 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8204 (init_gcc_specs): Likewise.
8205 (init_spec): Likewise.
8206 (display_help): Update help message for -pie.
8207 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8210 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8212 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8213 (movdi_aarch64): Likewise.
8214 (movti_aarch64): Likewise.
8216 2017-09-12 Simon Wright <simon@pushface.org>
8219 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8220 calculation of the minor version, always output as 0.
8222 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8225 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8226 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8227 on it early, rather than manual conversion late. For
8228 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8229 instead of performing manual conversion.
8231 2017-09-12 Carl Love <cel@us.ibm.com>
8233 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8234 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8236 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8237 VMULOSW): Add definitions.
8238 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8239 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8240 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8241 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8242 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8244 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8246 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8248 (movti_aarch64): Likewise.
8249 (movdf_aarch64): Likewise.
8250 (movtf_aarch64): Likewise.
8251 (load_pairdi): Likewise.
8252 (store_pairdi): Likewise.
8253 (load_pairdf): Likewise.
8254 (store_pairdf): Likewise.
8255 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8256 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8257 (ldr_got_small_<mode>): Likewise.
8258 (ldr_got_small_28k_<mode>): Likewise.
8259 (ldr_got_tiny): Likewise.
8260 * config/aarch64/iterators.md (ldst_sz): New.
8261 (ldpstp_sz): Likewise.
8262 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8264 (thunderx_load): Split load_8 to load_16.
8265 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8267 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8268 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8269 (xgene1_store_pair): Split store_8 to store_16.
8270 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8271 (falkor_st_0_st_sd): Split store_8 to store_16.
8273 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8275 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8276 and store1/2/3/4 to store_4/8/12/16.
8277 * config/aarch64/aarch64.md: Update for rename.
8278 * config/arm/arm.md: Likewise.: Likewise.
8279 * config/arm/arm.c: Likewise.
8280 * config/arm/thumb1.md: Likewise.
8281 * config/arm/thumb2.md: Likewise.
8282 * config/arm/vfp.md: Likewise.
8283 * config/arm/arm-generic.md: Likewise.
8284 * config/arm/arm1020e.md: Likewise.
8285 * config/arm/arm1026ejs.md: Likewise.
8286 * config/arm/arm1136jfs.md: Likewise.
8287 * config/arm/arm926ejs.md: Likewise.
8288 * config/arm/cortex-a15.md: Likewise.
8289 * config/arm/cortex-a17.md: Likewise.
8290 * config/arm/cortex-a5.md: Likewise.
8291 * config/arm/cortex-a53.md: Likewise.
8292 * config/arm/cortex-a57.md: Likewise.
8293 * config/arm/cortex-a7.md: Likewise.
8294 * config/arm/cortex-a8.md: Likewise.
8295 * config/arm/cortex-a9.md: Likewise.
8296 * config/arm/cortex-m4.md: Likewise.
8297 * config/arm/cortex-m7.md: Likewise.
8298 * config/arm/cortex-r4.md: Likewise.
8299 * config/arm/exynos-m1.md: Likewise.
8300 * config/arm/fa526.md: Likewise.
8301 * config/arm/fa606te.md: Likewise.
8302 * config/arm/fa626te.md: Likewise.
8303 * config/arm/fa726te.md: Likewise.
8304 * config/arm/fmp626.md: Likewise.
8305 * config/arm/iwmmxt.md: Likewise.
8306 * config/arm/ldmstm.md: Likewise.
8307 * config/arm/marvell-pj4.md: Likewise.
8308 * config/arm/xgene1.md: Likewise.
8309 * config/aarch64/thunderx.md: Likewise.
8310 * config/aarch64/thunderx2t99.md: Likewise.
8311 * config/aarch64/falkor.md: Likewise.
8313 2017-09-12 Martin Liska <mliska@suse.cz>
8315 * attribs.c (private_lookup_attribute): New function.
8316 * attribs.h (private_lookup_attribute): Declared here.
8317 (lookup_attribute): Called from this place.
8319 2017-09-12 Richard Biener <rguenther@suse.de>
8321 PR tree-optimization/82157
8322 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8323 stmts with side-effects.
8325 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8326 Alan Hayward <alan.hayward@arm.com>
8327 David Sherwood <david.sherwood@arm.com>
8329 * target.def (hard_regno_nregs): New hook.
8330 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8331 * targhooks.h (default_hard_regno_nregs): Declare.
8332 * targhooks.c (default_hard_regno_nregs): New function.
8333 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8334 (TARGET_HARD_REGNO_NREGS): ...this hook.
8335 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8336 (CLASS_MAX_NREGS): Likewise.
8337 * doc/tm.texi: Regenerate.
8338 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8339 instead of HARD_REGNO_NREGS.
8340 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8341 HARD_REGNO_NREGS in the comment.
8342 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8343 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8344 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8345 Return an unsigned int.
8346 (TARGET_HARD_REGNO_NREGS): Redefine.
8347 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8348 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8349 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8350 (arc_hard_regno_nregs): New function.
8351 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8352 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8353 (arm_hard_regno_nregs): New function.
8354 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8355 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8356 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8357 (TARGET_HARD_REGNO_NREGS): Redefine.
8358 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8359 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8360 (HARD_REGNO_NREGS): Delete.
8361 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8362 (cr16_hard_regno_nregs): New function.
8363 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8364 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8365 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8366 (cris_hard_regno_nregs): New function.
8367 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8368 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8369 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8370 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8371 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8372 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8373 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8374 (frv_hard_regno_nregs): Make static. Take and return an
8376 (frv_class_max_nregs): Remove outdated copy of documentation.
8377 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8378 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8379 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8380 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8381 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8382 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8383 (TARGET_HARD_REGNO_NREGS): Redefine.
8384 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8385 (CLASS_MAX_NREGS): Update comment.
8386 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8387 (ia64_hard_regno_nregs): New function.
8388 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8389 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8390 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8391 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8392 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8394 (m32c_hard_regno_nregs): Likewise. Make static.
8395 (TARGET_HARD_REGNO_NREGS): Redefine.
8396 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8397 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8398 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8399 (m68k_hard_regno_nregs): New function.
8400 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8401 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8402 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8403 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8404 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8405 Take and return an unsigned int.
8406 (TARGET_HARD_REGNO_NREGS): Redefine.
8407 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8408 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8409 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8410 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8411 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8412 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8413 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8414 (msp430_hard_regno_nregs): Make static. Take and return an
8416 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8417 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8418 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8419 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8420 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8421 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8422 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8423 (TARGET_HARD_REGNO_NREGS): Redefine.
8424 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8425 (PA_HARD_REGNO_NREGS): ...this.
8426 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8427 (PA_HARD_REGNO_NREGS): ...this.
8428 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8429 (pa_hard_regno_nregs): New function.
8430 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8431 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8432 (pdp11_hard_regno_nregs): New function.
8433 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8434 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8435 (rs6000_hard_regno_nregs_hook): New function.
8436 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8437 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8438 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8439 Take and return an unsigned int. Move earlier in file.
8440 (TARGET_HARD_REGNO_NREGS): Redefine.
8441 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8442 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8443 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8444 (rl78_hard_regno_nregs): Make static. Take and return an
8446 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8447 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8448 (rs6000_hard_regno_nregs_hook): New function.
8449 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8450 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8451 (TARGET_HARD_REGNO_NREGS): Redefine.
8452 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8453 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8454 instead of HARD_REGNO_NREGS.
8455 (s390_hard_regno_nregs): New function.
8456 (s390_hard_regno_mode_ok): Add comment from s390.h.
8457 (TARGET_HARD_REGNO_NREGS): Redefine.
8458 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8459 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8460 (sh_hard_regno_nregs): New function.
8461 (sh_pass_in_reg_p): Use it.
8462 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8463 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8464 (sparc_hard_regno_nregs): New function.
8465 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8466 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8467 (spu_function_arg_advance): Use it, supplying a valid register number.
8468 (TARGET_HARD_REGNO_NREGS): Redefine.
8469 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8470 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8471 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8472 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8473 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8474 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8475 (CLASS_MAX_NREGS): Remove copy of old documentation.
8476 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8477 (visium_hard_regno_nregs): New function.
8478 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8479 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8480 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8481 xtensa_hard_regno_nregs): New function.
8482 * system.h (HARD_REGNO_NREGS): Poison.
8484 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8486 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8487 hard_regno_nregs instead of HARD_REGNO_NREGS.
8488 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8489 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8490 (c6x_expand_epilogue): Likewise.
8491 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8492 (frv_read_iacc_argument): Likewise.
8493 * config/sh/sh.c: Include regs.h.
8494 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8495 (regs_used): Likewise.
8496 (output_stack_adjust): Likewise.
8497 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8498 * expmed.c: Include regs.h.
8499 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8500 * ree.c: Include regs.h.
8501 (combine_reaching_defs): Use hard_regno_nregs instead of
8503 (add_removable_extension): Likewise.
8505 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8507 * regs.h (hard_regno_nregs): Turn into a function.
8508 (end_hard_regno): Update accordingly.
8509 * caller-save.c (setup_save_areas): Likewise.
8510 (save_call_clobbered_regs): Likewise.
8511 (replace_reg_with_saved_mem): Likewise.
8512 (insert_restore): Likewise.
8513 (insert_save): Likewise.
8514 * combine.c (can_change_dest_mode): Likewise.
8515 (move_deaths): Likewise.
8516 (distribute_notes): Likewise.
8517 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8518 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8519 (rs6000_split_multireg_move): Likewise.
8520 (rs6000_register_move_cost): Likewise.
8521 (rs6000_memory_move_cost): Likewise.
8522 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8523 (rs6000_split_multireg_move): Likewise.
8524 (rs6000_register_move_cost): Likewise.
8525 (rs6000_memory_move_cost): Likewise.
8526 * cselib.c (cselib_reset_table): Likewise.
8527 (cselib_lookup_1): Likewise.
8528 * emit-rtl.c (set_mode_and_regno): Likewise.
8529 * function.c (aggregate_value_p): Likewise.
8530 * ira-color.c (setup_profitable_hard_regs): Likewise.
8531 (check_hard_reg_p): Likewise.
8532 (calculate_saved_nregs): Likewise.
8533 (assign_hard_reg): Likewise.
8534 (improve_allocation): Likewise.
8535 (calculate_spill_cost): Likewise.
8536 * ira-emit.c (modify_move_list): Likewise.
8537 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8538 (ira_hard_reg_in_set_p): Likewise.
8539 * ira.c (setup_reg_mode_hard_regset): Likewise.
8540 (clarify_prohibited_class_mode_regs): Likewise.
8541 (check_allocation): Likewise.
8542 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8543 (lra_setup_reg_renumber): Likewise.
8544 (setup_try_hard_regno_pseudos): Likewise.
8545 (spill_for): Likewise.
8546 (assign_hard_regno): Likewise.
8547 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8548 * lra-constraints.c (in_class_p): Likewise.
8549 (lra_constraint_offset): Likewise.
8550 (simplify_operand_subreg): Likewise.
8551 (lra_constraints): Likewise.
8552 (split_reg): Likewise.
8553 (split_if_necessary): Likewise.
8554 (invariant_p): Likewise.
8555 (inherit_in_ebb): Likewise.
8556 * lra-lives.c (process_bb_lives): Likewise.
8557 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8558 (get_hard_regs): Likewise.
8559 (do_remat): Likewise.
8560 * lra-spills.c (assign_spill_hard_regs): Likewise.
8561 * mode-switching.c (create_pre_exit): Likewise.
8562 * postreload.c (reload_combine_recognize_pattern): Likewise.
8563 * recog.c (peep2_find_free_register): Likewise.
8564 * regcprop.c (kill_value_regno): Likewise.
8565 (set_value_regno): Likewise.
8566 (copy_value): Likewise.
8567 (maybe_mode_change): Likewise.
8568 (find_oldest_value_reg): Likewise.
8569 (copyprop_hardreg_forward_1): Likewise.
8570 * regrename.c (check_new_reg_p): Likewise.
8571 (regrename_do_replace): Likewise.
8572 * reload.c (push_reload): Likewise.
8573 (combine_reloads): Likewise.
8574 (find_dummy_reload): Likewise.
8575 (operands_match_p): Likewise.
8576 (find_reloads): Likewise.
8577 (find_equiv_reg): Likewise.
8578 (reload_adjust_reg_for_mode): Likewise.
8579 * reload1.c (count_pseudo): Likewise.
8580 (count_spilled_pseudo): Likewise.
8581 (find_reg): Likewise.
8582 (clear_reload_reg_in_use): Likewise.
8583 (free_for_value_p): Likewise.
8584 (allocate_reload_reg): Likewise.
8585 (choose_reload_regs): Likewise.
8586 (reload_adjust_reg_for_temp): Likewise.
8587 (emit_reload_insns): Likewise.
8588 (delete_output_reload): Likewise.
8589 * rtlanal.c (subreg_get_info): Likewise.
8590 * sched-deps.c (sched_analyze_reg): Likewise.
8591 * sel-sched.c (init_regs_for_mode): Likewise.
8592 (mark_unavailable_hard_regs): Likewise.
8593 (choose_best_reg_1): Likewise.
8594 (verify_target_availability): Likewise.
8595 * valtrack.c (dead_debug_insert_temp): Likewise.
8596 * var-tracking.c (track_loc_p): Likewise.
8597 (emit_note_insn_var_location): Likewise.
8598 * varasm.c (make_decl_rtl): Likewise.
8599 * reginfo.c (choose_hard_reg_mode): Likewise.
8600 (init_reg_modes_target): Refer directly to
8601 this_target_regs->x_hard_regno_nregs.
8603 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8605 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8606 instead of hard_regno_nregs.
8608 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8610 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8611 end_hard_regno instead of hard_regno_nregs.
8612 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8613 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8614 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8615 * ira-color.c (improve_allocation): Likewise.
8616 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8617 * lra-lives.c (mark_regno_live): Likewise.
8618 (mark_regno_dead): Likewise.
8619 * lra-remat.c (operand_to_remat): Likewise.
8620 * lra.c (collect_non_operand_hard_regs): Likewise.
8621 * postreload.c (reload_combine_note_store): Likewise.
8622 (move2add_valid_value_p): Likewise.
8623 * reload.c (regno_clobbered_p): Likewise.
8625 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8627 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8629 * config/v850/v850.c (v850_reorg): Likewise.
8630 * reload.c (refers_to_regno_for_reload_p): Likewise.
8631 (find_equiv_reg): Likewise.
8632 * reload1.c (reload_reg_reaches_end_p): Likewise.
8634 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8636 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8638 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8639 * config/arm/arm.c (output_move_neon): Likewise.
8640 (arm_attr_length_move_neon): Likewise.
8641 (neon_split_vcombine): Likewise.
8642 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8643 (c6x_mark_reg_written): Likewise.
8644 (c6x_dwarf_register_span): Likewise.
8645 * config/i386/i386.c (ix86_save_reg): Likewise.
8646 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8647 (rws_access_reg): Likewise.
8648 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8649 * mode-switching.c (create_pre_exit): Likewise.
8650 * ree.c (combine_reaching_defs): Likewise.
8651 (add_removable_extension): Likewise.
8652 * regcprop.c (find_oldest_value_reg): Likewise.
8653 (copyprop_hardreg_forward_1): Likewise.
8654 * reload.c (reload_inner_reg_of_subreg): Likewise.
8655 (push_reload): Likewise.
8656 (combine_reloads): Likewise.
8657 (find_dummy_reload): Likewise.
8658 (reload_adjust_reg_for_mode): Likewise.
8659 * reload1.c (find_reload_regs): Likewise.
8660 (forget_old_reloads_1): Likewise.
8661 (reload_reg_free_for_value_p): Likewise.
8662 (reload_adjust_reg_for_temp): Likewise.
8663 (emit_reload_insns): Likewise.
8664 (delete_output_reload): Likewise.
8665 * sel-sched.c (choose_best_reg_1): Likewise.
8666 (choose_best_pseudo_reg): Likewise.
8668 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8669 Alan Hayward <alan.hayward@arm.com>
8670 David Sherwood <david.sherwood@arm.com>
8672 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8673 * target.def (slow_unaligned_access): New hook.
8674 * targhooks.h (default_slow_unaligned_access): Declare.
8675 * targhooks.c (default_slow_unaligned_access): New function.
8676 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8677 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8678 * doc/tm.texi: Regenerate.
8679 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8680 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8681 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8683 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8684 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8686 (rs6000_slow_unaligned_access): New function.
8687 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8688 (expand_block_compare): Likewise.
8689 (expand_strn_compare): Likewise.
8690 (rs6000_rtx_costs): Likewise.
8691 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8692 (riscv_slow_unaligned_access): Likewise.
8693 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8694 (riscv_slow_unaligned_access_p): ...this and make static.
8695 (riscv_option_override): Update accordingly.
8696 (riscv_slow_unaligned_access): New function.
8697 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8698 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8699 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8700 (rs6000_slow_unaligned_access): New function.
8701 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8702 (rs6000_rtx_costs): Likewise.
8703 * config/rs6000/rs6000-string.c (expand_block_compare)
8704 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8705 of SLOW_UNALIGNED_ACCESS.
8706 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8707 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8708 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8709 of SLOW_UNALIGNED_ACCESS.
8710 * expmed.c (simple_mem_bitfield_p): Likewise.
8711 * expr.c (alignment_for_piecewise_move): Likewise.
8712 (emit_group_load_1): Likewise.
8713 (emit_group_store): Likewise.
8714 (copy_blkmode_from_reg): Likewise.
8715 (emit_push_insn): Likewise.
8716 (expand_assignment): Likewise.
8717 (store_field): Likewise.
8718 (expand_expr_real_1): Likewise.
8719 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8720 * lra-constraints.c (simplify_operand_subreg): Likewise.
8721 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8722 * gimple-ssa-store-merging.c: Likewise in block comment at start
8724 * tree-ssa-strlen.c: Include target.h.
8725 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8726 of SLOW_UNALIGNED_ACCESS.
8727 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8729 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8731 PR rtl-optimization/82185
8732 * expmed.c (emit_store_flag_int): Only test tem if it has been
8735 2017-09-12 Richard Biener <rguenther@suse.de>
8738 * match.pd ((FTYPE) N CMP CST): Fix typo.
8740 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8742 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8744 (mips_near_type_p): Add 'short_call' attribute as a synonym
8746 * doc/extend.texi (short_call): Document new function attribute.
8748 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8751 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8752 assertion check that in the condition.
8753 (get_atomic_generic_size): Likewise. Before testing if parameter
8754 has pointer type, if it has array type, call for C++
8755 default_conversion to perform array-to-pointer conversion.
8757 2017-09-12 Richard Biener <rguenther@suse.de>
8759 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8760 for operations we cannot scalarize.
8762 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8764 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8765 vectors heap vectors. Clean up comments.
8766 Make visited_bbs a reference.
8767 (profitable_jump_thread_path): Make GC
8768 vectors heap vectors. Clean up comments.
8770 (convert_and_register_jump_thread_path): Make GC vectors heap
8772 (check_subpath_and_update_thread_path): Same. Clean up comments.
8773 Make visited_bbs a reference.
8774 (handle_phi): Abstract common code to to
8775 register_jump_thread_path_if_profitable.
8776 Rename VAR_BB to DEF_BB.
8778 Make GC vectors heap vectors.
8779 Make visited_bbs a reference.
8780 (handle_assignment): Same.
8781 (register_jump_thread_path_if_profitable): New.
8782 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8784 Make GC vectors heap vectors. Clean up comments.
8785 Make visited_bbs a reference.
8786 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8787 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8790 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8793 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8794 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8796 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8798 Revert r251800 and r251799.
8800 2017-09-11 Martin Jambor <mjambor@suse.cz>
8803 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8804 arguments in advance.
8805 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8806 use it to find predecessor edges.
8807 (naive_outof_ssa): Collect vector of predecessors.
8809 2017-09-08 Jason Merrill <jason@redhat.com>
8811 PR c++/70029 - ICE with ref-qualifier and -flto
8812 * langhooks.h (struct lang_hooks_for_types): Add
8813 copy_lang_qualifiers.
8814 * attribs.c (build_type_attribute_qual_variant): Use it.
8815 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8817 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8818 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8820 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8823 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8824 (*mulsi3_sp64): New instruction.
8825 (mulsi3): New expander.
8827 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8829 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8831 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8833 * sancov.c: Include memmodel.h.
8835 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8838 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8841 2017-09-07 Carl Love <cel@us.ibm.com>
8843 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8844 the sldi instruction.
8846 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8848 * sancov.c: Include tm_p.h.
8850 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8853 * output.h (switch_to_other_text_partition): New declaration.
8854 * varasm.c (switch_to_other_text_partition): New function.
8855 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8856 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8857 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8858 to the other text partition before emitting LCL label and switch back
8859 after emitting the word after it.
8861 2017-09-07 Richard Biener <rguenther@suse.de>
8863 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8864 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8865 critical edges here, after loop init.
8866 (pass_data_pre): Remove PROP_no_crit_edges flags.
8867 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8868 for valueization of call args to avoid leaking VN_TOP.
8869 (visit_use): Assert we do not visit default defs.
8870 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8871 Use error_mark_node to more easily detect leaking VN_TOP.
8872 All default-defs are varying, not VN_TOP. Mark them visited.
8873 (run_scc_vn): Make code match comment.
8875 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8877 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8878 OPTION_MASK_FLOAT128_KEYWORD.
8879 (POWERPC_MASKS): Likewise.
8880 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8881 support for the -mfloat128-type option, and make -mfloat128
8882 default on PowerPC Linux systems. Define or undefine
8883 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8884 Define __float128 to be __ieee128 if IEEE 128-bit support is
8885 enabled, or undefine it.
8886 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8887 Delete defining __FLOAT128_TYPE__.
8888 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8889 -mfloat128-type option and make -mfloat128 default on PowerPC
8891 (TARGET_FLOAT128_TYPE): Likewise.
8892 (-mfloat128-type): Likewise.
8893 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8894 Delete the -mfloat128-type option and make -mfloat128 default on
8895 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8896 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8897 128-bit floating point is enabled. Change tests from using
8898 -mfloat128-type to -mfloat128.
8899 (rs6000_mangle_type): Use the correct mangling for the __float128
8900 type even if normal long double is restricted to 64-bits.
8901 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8903 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8904 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8905 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8907 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8908 documentation for -mfloat128.
8910 2017-09-06 Olivier Hainque <hainque@adacore.com>
8912 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8914 2017-09-06 Wish Wu <wishwu007@gmail.com>
8915 Jakub Jelinek <jakub@redhat.com>
8917 * asan.c (initialize_sanitizer_builtins): Add
8918 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8919 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8920 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8921 BT_FN_VOID_UINT64_PTR variables.
8922 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8923 (BT_FN_VOID_UINT16_UINT16): Likewise.
8924 (BT_FN_VOID_UINT32_UINT32): Likewise.
8925 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8926 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8927 (BT_FN_VOID_UINT64_PTR): Likewise.
8928 * common.opt (flag_sanitize_coverage): New variable.
8929 (fsanitize-coverage=trace-pc): Remove.
8930 (fsanitize-coverage=): Add.
8931 * flag-types.h (enum sanitize_coverage_code): New enum.
8932 * fold-const.c (fold_range_test): Disable non-short-circuit
8933 optimization if flag_sanitize_coverage.
8934 (fold_truth_andor): Likewise.
8935 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8936 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8937 (coverage_sanitizer_opts): New array.
8938 (get_closest_sanitizer_option): Add OPTS argument, handle also
8939 OPT_fsanitize_coverage_.
8940 (parse_sanitizer_options): Adjusted to also handle
8941 OPT_fsanitize_coverage_.
8942 (common_handle_option): Add OPT_fsanitize_coverage_.
8943 * sancov.c (instrument_comparison, instrument_switch): New function.
8944 (sancov_pass): Add trace-cmp support.
8945 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8946 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8947 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8948 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8949 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8950 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8951 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8952 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8953 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8954 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8956 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8958 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8959 error. Only quit immediately if parsing is complete.
8960 (BEGIN): Initialize fatal_err and parse_done.
8961 (begin fpu, end fpu): Check number of arguments.
8962 (begin arch, end arch): Likewise.
8963 (begin cpu, end cpu): Likewise.
8964 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8965 (optalias): Likewise.
8967 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8969 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8970 * config/arm/arm-isa.h: Delete. Move definitions to ...
8971 * arm-cpus.in: ... here. Use new feature and fgroup values.
8972 * config/arm/arm.c (arm_option_override): Use lower case for feature
8974 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8975 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8976 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8978 (print_isa_bits_for): New function.
8979 (gen_isa): New function.
8980 (gen_comm_data): Use print_isa_bits_for.
8981 (define feature): New keyword.
8982 (define fgroup): New keyword.
8983 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8984 (arm-isa.h): Add rule to generate file.
8985 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8986 case for feature bit names.
8988 2017-09-06 Richard Biener <rguenther@suse.de>
8990 * tree-ssa-pre.c (NECESSARY): Remove.
8991 (create_expression_by_pieces): Do not touch pass-local flags.
8992 (insert_into_preds_of_block): Likewise.
8993 (do_pre_regular_insertion): Likewise.
8994 (eliminate_insert): Likewise.
8995 (eliminate_dom_walker::before_dom_children): Likewise.
8996 (fini_eliminate): Do not look at inserted_exprs.
8997 (mark_operand_necessary): Remove.
8998 (remove_dead_inserted_code): Replace with simple work-list
8999 algorithm based on inserted_exprs and SSA uses.
9000 (pass_pre::execute): Re-order fini_eliminate and
9001 remove_dead_inserted_code.
9003 2017-09-06 Olivier Hainque <hainque@adacore.com>
9005 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9006 for VxWorks 7. Adjust surrounding comments.
9008 2017-09-06 Richard Biener <rguenther@suse.de>
9010 * gimple-ssa-strength-reduction.c
9011 (find_candidates_dom_walker::before_dom_children): Also allow
9014 2017-09-06 Richard Biener <rguenther@suse.de>
9016 PR tree-optimization/82108
9017 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9018 for gap in the non-permutation SLP case.
9020 2017-09-06 Martin Jambor <mjambor@suse.cz>
9022 PR tree-optimization/82078
9023 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9024 add_access_to_work_queue...
9025 (build_accesses_from_assign): ...here.
9026 (propagate_all_subaccesses): Make sure racc is the group
9027 representative, if there is one.
9029 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9032 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9035 2017-09-06 Richard Biener <rguenther@suse.de>
9037 * gimple-ssa-strength-reduction.c
9038 (find_candidates_dom_walker::before_doom_children): Use a
9039 type and not a mode check.
9041 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9044 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9046 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9048 2017-09-05 Jeff Law <law@redhat.com>
9050 PR tree-optimization/64910
9051 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9052 cases where we have 3 or more operands.
9054 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9057 * omp-low.c (lower_omp_for): Recompute tree invariant if
9058 gimple_omp_for_initial/final is ADDR_EXPR.
9061 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9062 into gimple val before gimplification fo the COND_EXPR.
9064 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9066 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9067 REGION_COPY argument.
9068 (thread_through_all_blocks): Remove unused argument to
9069 duplicate_thread_path.
9071 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9072 Alan Hayward <alan.hayward@arm.com>
9073 David Sherwood <david.sherwood@arm.com>
9075 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9076 Take a scalar_mode rather than a machine_mode.
9077 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9078 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9079 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9080 (aarch64_gen_adjusted_ldpstp): Likewise.
9081 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9083 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9084 Alan Hayward <alan.hayward@arm.com>
9085 David Sherwood <david.sherwood@arm.com>
9087 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9088 Take a scalar_int_mode instead of a machine_mode.
9089 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9090 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9091 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9092 (aarch64_simd_attr_length_rglist): Delete.
9093 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9094 a scalar_int_mode instead of a machine_mode.
9095 (aarch64_add_offset): Likewise.
9096 (aarch64_internal_mov_immediate): Likewise
9097 (aarch64_add_constant_internal): Likewise.
9098 (aarch64_add_constant): Likewise.
9099 (aarch64_movw_imm): Likewise.
9100 (aarch64_rtx_arith_op_extract_p): Likewise.
9101 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9102 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9103 Remove assert that the mode isn't a vector.
9104 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9105 (aarch64_expand_mov_immediate): Update calls after above changes.
9106 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9107 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9108 (aarch64_move_imm): Likewise.
9109 (aarch64_can_const_movi_rtx_p): Likewise.
9110 (aarch64_strip_extend): Likewise.
9111 (aarch64_extr_rtx_p): Likewise.
9112 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9113 a CONST_INT when the mode parameter is VOIDmode.
9114 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9116 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9118 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9119 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9120 (bitwise_type_for_mode): Update accordingly.
9122 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9124 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9125 * stor-layout.c (mode_for_size_tree): Likewise.
9126 (mode_for_array): Update accordingly.
9127 (layout_decl): Likewise.
9128 (compute_record_mode): Likewise. Only set the mode once.
9130 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9132 * target.def (get_mask_mode): Change return type to opt_mode.
9134 * doc/tm.texi: Regenerate.
9135 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9136 * targhooks.c (default_get_mask_mode): Likewise.
9137 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9138 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9139 targetm.get_mask_mode.
9140 * tree.c (build_truth_vector_type): Likewise.
9142 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9144 * machmode.h (mode_for_vector): Return an opt_mode.
9145 * stor-layout.c (mode_for_vector): Likewise.
9146 (mode_for_int_vector): Update accordingly.
9147 (layout_type): Likewise.
9148 * config/i386/i386.c (emit_memmov): Likewise.
9149 (ix86_expand_set_or_movmem): Likewise.
9150 (ix86_expand_vector_init): Likewise.
9151 (ix86_get_mask_mode): Likewise.
9152 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9154 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9155 * expmed.c (extract_bit_field_1): Likewise.
9156 * expr.c (expand_expr_real_2): Likewise.
9157 * optabs-query.c (can_vec_perm_p): Likewise.
9158 (can_vec_mask_load_store_p): Likewise.
9159 * optabs.c (expand_vec_perm): Likewise.
9160 * targhooks.c (default_get_mask_mode): Likewise.
9161 * tree-vect-stmts.c (vectorizable_store): Likewise.
9162 (vectorizable_load): Likewise.
9163 (get_vectype_for_scalar_type_and_size): Likewise.
9165 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9167 * machmode.h (mode_for_int_vector): New function.
9168 * stor-layout.c (mode_for_int_vector): Likewise.
9169 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9170 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9171 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9172 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9173 (s390_expand_vcond): Likewise.
9175 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9177 * machmode.h (opt_machine_mode): New type.
9178 (opt_mode<T>): Allow construction from anything that can be
9180 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9181 (mode_for_size): Return an opt_machine_mode.
9182 * stor-layout.c (mode_for_size): Likewise.
9183 (mode_for_size_tree): Update call accordingly.
9184 (bitwise_mode_for_mode): Likewise.
9185 (make_fract_type): Likewise.
9186 (make_accum_type): Likewise.
9187 * caller-save.c (replace_reg_with_saved_mem): Update call
9189 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9190 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9191 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9192 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9193 * expmed.c (extract_bit_field_1): Likewise.
9194 * reload.c (get_secondary_mem): Likewise.
9195 * varasm.c (assemble_integer): Likewise.
9196 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9199 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9201 * machmode.h (decimal_float_mode_for_size): New function.
9202 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9203 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9204 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9205 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9206 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9207 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9209 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9211 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9212 (get_builtin_sync_mode): Likewise.
9213 (expand_ifn_atomic_compare_exchange): Likewise.
9214 (expand_builtin_atomic_clear): Likewise.
9215 (expand_builtin_atomic_test_and_set): Likewise.
9216 (fold_builtin_atomic_always_lock_free): Likewise.
9217 * calls.c (compute_argument_addresses): Likewise.
9218 (emit_library_call_value_1): Likewise.
9219 (store_one_arg): Likewise.
9220 * combine.c (combine_instructions): Likewise.
9221 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9222 * config/arm/arm.c (arm_function_value): Likewise.
9223 (aapcs_allocate_return_reg): Likewise.
9224 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9225 * config/i386/i386.c (construct_container): Likewise.
9226 (ix86_gimplify_va_arg): Likewise.
9227 (ix86_expand_sse_cmp): Likewise.
9228 (emit_memmov): Likewise.
9229 (emit_memset): Likewise.
9230 (expand_small_movmem_or_setmem): Likewise.
9231 (ix86_expand_pextr): Likewise.
9232 (ix86_expand_pinsr): Likewise.
9233 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9234 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9236 * config/mips/mips.c (mips_function_value_1) Likewise.
9237 (mips_block_move_straight): Likewise.
9238 (mips_expand_ins_as_unaligned_store): Likewise.
9239 * config/powerpcspe/powerpcspe.c
9240 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9241 (rs6000_darwin64_record_arg_flush): Likewise.
9242 * config/rs6000/rs6000.c
9243 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9244 (rs6000_darwin64_record_arg_flush): Likewise.
9245 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9246 (sparc_function_value_1): Likewise.
9247 * config/spu/spu.c (adjust_operand): Likewise.
9248 (spu_emit_branch_or_set): Likewise.
9249 (arith_immediate_p): Likewise.
9250 * emit-rtl.c (gen_lowpart_common): Likewise.
9251 * expr.c (expand_expr_real_1): Likewise.
9252 * function.c (assign_parm_setup_block): Likewise.
9253 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9254 * reload1.c (alter_reg): Likewise.
9255 * stor-layout.c (mode_for_vector): Likewise.
9256 (layout_type): Likewise.
9258 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9260 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9261 (spu_convert_move): Likewise.
9262 * lower-subreg.c (resolve_simple_move): Likewise.
9264 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9267 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9268 define_insn to a define_expand.
9269 (altivec_vsum2sws_direct): New define_insn.
9270 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9272 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9274 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9277 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9282 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9283 realign stack if stack alignment needed is less than incoming
9286 2017-09-05 Marek Polacek <polacek@redhat.com>
9289 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9292 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9294 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9296 2017-09-05 Richard Biener <rguenther@suse.de>
9298 PR tree-optimization/82084
9299 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9301 2017-09-05 Richard Biener <rguenther@suse.de>
9303 PR tree-optimization/82102
9304 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9306 2017-09-05 Martin Liska <mliska@suse.cz>
9308 PR tree-optimization/82032
9309 * tree-cfg.c (generate_range_test): New function.
9310 * tree-cfg.h (generate_range_test): Declared here.
9311 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9312 (cleanup_control_expr_graph): Use it.
9313 * tree-switch-conversion.c (try_switch_expansion): Remove
9315 (emit_case_nodes): Use generate_range_test.
9317 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9320 * config/i386/i386.md (*<btsc><mode>_mask): Add
9321 TARGET_USE_BT to insn constraint.
9322 (*btr<mode>_mask): Ditto.
9324 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9326 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9327 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9329 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9332 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9333 TARGET_NEON and TARGET_IWMMXT.
9334 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9335 TARGET_NEON and TARGET_IWMMXT.
9336 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9338 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9340 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9341 (ix86_rewrite_tls_address): Ditto.
9342 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9343 (ix86_rewrite_tls_address_1): Ditto.
9344 (ix86_rewrite_tls_address): Ditto.
9345 * config/i386/predicates.md (tls_address_pattern): New predicate.
9346 * config/i386/i386.md (TLS address splitter): New splitter.
9348 2017-09-04 Richard Biener <rguenther@suse.de>
9350 PR tree-optimization/82084
9351 * fold-const.h (can_native_encode_string_p): Declare.
9352 * fold-const.c (can_native_encode_string_p): Factor out from ...
9353 (native_encode_string): ... here.
9354 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9355 vectorizing stores from constants we later cannot handle.
9357 2017-09-04 Marek Polacek <polacek@redhat.com>
9360 * doc/invoke.texi: Update -Wtautological-compare documentation.
9362 2017-09-04 Jeff Law <law@redhat.com>
9364 PR tree-optimization/64910
9365 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9366 swap the first and last operand if the last is a constant.
9368 2017-09-04 Marek Polacek <polacek@redhat.com>
9371 * convert.c (do_narrow): When sanitizing signed integer overflows,
9372 bail out for signed types.
9373 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9375 2017-09-04 Richard Biener <rguenther@suse.de>
9377 PR tree-optimization/82060
9378 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9379 Move devirtualization after stmt folding and before EH/AB/noreturn
9380 cleanup to get the stmt refs canonicalized. Use a bool instead
9381 of gimple_modified_p since that doesn't work for NOPs. Schedule
9382 NOPs generated by folding for removal.
9384 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9385 Alan Hayward <alan.hayward@arm.com>
9386 David Sherwood <david.sherwood@arm.com>
9388 * coretypes.h (pad_direction): New enum.
9389 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9390 (FUNCTION_ARG_PADDING): Likewise.
9391 * target.def (function_arg_padding): New hook.
9392 * targhooks.h (default_function_arg_padding): Declare.
9393 * targhooks.c (default_function_arg_padding): New function.
9394 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9395 (TARGET_FUNCTION_ARG_PADDING): ...this.
9396 * doc/tm.texi: Regenerate.
9397 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9398 instead of direction.
9399 (compute_argument_addresses): Likewise.
9400 (load_register_parameters): Likewise.
9401 (emit_library_call_value_1): Likewise.
9402 (store_one_arg): Use targetm.calls.function_arg_padding instead
9403 of FUNCTION_ARG_PADDING.
9404 (must_pass_in_stack_var_size_or_pad): Likewise.
9405 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9406 (emit_group_store): Likewise.
9407 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9408 instead of FUNCTION_ARG_PADDING.
9409 (emit_push_insn): Likewise, and propagate enum change throughout
9411 * function.h (direction): Delete.
9412 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9414 * function.c (assign_parm_find_stack_rtl): Likewise.
9415 (assign_parm_setup_block_p): Likewise.
9416 (assign_parm_setup_block): Likewise.
9417 (gimplify_parameters): Likewise.
9418 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9419 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9421 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9422 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9423 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9424 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9425 (aarch64_function_arg_padding): ...this new function.
9426 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9427 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9428 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9429 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9430 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9431 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9432 (arm_pad_arg_upward): Replace with...
9433 (arm_function_arg_padding): ...this new function.
9434 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9436 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9437 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9438 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9439 (ia64_hpux_function_arg_padding): Replace with...
9440 (ia64_function_arg_padding): ...this new function. Use pad_direction
9441 instead of direction. Check for TARGET_HPUX.
9442 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9443 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9444 (iq2000_function_arg_padding): New function.
9445 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9446 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9447 (mips_function_arg_padding): ...this new function.
9448 (mips_pad_reg_upward): Update accordingly.
9449 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9450 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9451 targetm.calls.function_arg_padding.
9452 (FUNCTION_ARG_PADDING): Delete.
9453 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9454 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9455 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9456 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9457 (nios2_block_reg_padding): Return pad_direction instead of direction.
9458 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9459 instead of direction.
9460 (nios2_function_arg_padding): Likewise. Make static.
9461 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9462 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9463 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9464 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9465 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9466 (pa_function_arg_padding): Make static. Return pad_direction instead
9468 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9469 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9470 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9471 instead of direction. Use targetm.calls.function_arg_padding.
9472 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9473 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9474 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9475 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9476 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9478 (function_arg_padding): Rename to...
9479 (rs6000_function_arg_padding): ...this. Make static. Return
9480 pad_direction instead of direction.
9481 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9482 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9483 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9484 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9485 instead of direction. Use targetm.calls.function_arg_padding.
9486 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9487 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9488 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9489 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9490 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9491 (function_arg_padding): Rename to...
9492 (rs6000_function_arg_padding): ...this. Make static. Return
9493 pad_direction instead of direction.
9494 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9495 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9496 * config/s390/s390.c (s390_function_arg_padding): New function.
9497 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9498 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9499 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9500 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9501 (function_arg_padding): Rename to...
9502 (sparc_function_arg_padding): ...this. Make static. Return
9503 pad_direction instead of direction.
9504 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9505 * config/spu/spu.c (spu_function_arg_padding): New function.
9506 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9507 * system.h (FUNCTION_ARG_PADDING): Poison.
9509 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9510 Alan Hayward <alan.hayward@arm.com>
9511 David Sherwood <david.sherwood@arm.com>
9513 * target.def (modes_tieable_p): New hook.
9514 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9515 (TARGET_MODES_TIEABLE_P): ...this.
9516 * doc/tm.texi.in: Regenerate.
9517 * hooks.h (hook_bool_mode_mode_true): Declare.
9518 * hooks.c (hook_bool_mode_mode_true): New function.
9519 * combine.c (subst): Use targetm.modes_tieable_p instead of
9521 * dse.c (find_shift_sequence): Likewise.
9522 * expmed.c (extract_low_bits): Likewise.
9523 * lower-subreg.c: Include target.h.
9524 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9526 * rtlanal.c (rtx_cost): Likewise.
9527 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9528 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9529 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9530 (TARGET_MODES_TIEABLE_P): Redefine.
9531 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9532 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9533 (TARGET_MODES_TIEABLE_P): Redefine.
9534 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9535 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9536 (arc_modes_tieable_p): New function.
9537 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9538 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9539 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9540 (arm_modes_tieable_p): Make static.
9541 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9542 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9543 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9544 (TARGET_MODES_TIEABLE_P): Redefine.
9545 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9546 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9547 (TARGET_MODES_TIEABLE_P): Redefine.
9548 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9549 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9550 (cr16_modes_tieable_p): New function.
9551 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9552 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9553 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9554 (TRULY_NOOP_TRUNCATION): Update comment.
9555 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9556 (TRULY_NOOP_TRUNCATION): Update comment.
9557 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9558 (frv_modes_tieable_p): New function.
9559 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9560 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9561 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9562 (TARGET_MODES_TIEABLE_P): Redefine.
9563 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9564 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9565 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9566 (TARGET_MODES_TIEABLE_P): Redefine.
9567 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9568 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9569 (ia64_modes_tieable_p): New function.
9570 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9571 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9572 (iq2000_modes_tieable_p): New function.
9573 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9574 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9575 (lm32_modes_tieable_p): New function.
9576 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9577 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9578 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9579 (TARGET_MODES_TIEABLE_P): Redefine.
9580 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9581 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9582 (m32r_modes_tieable_p): New function.
9583 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9584 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9585 (m68k_modes_tieable_p): New function.
9586 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9587 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9588 (mcore_modes_tieable_p): New function.
9589 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9590 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9592 (TARGET_MODES_TIEABLE_P): Redefine.
9593 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9594 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9595 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9596 (TARGET_MODES_TIEABLE_P): Redefine.
9597 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9598 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9599 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9600 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9601 (mn10300_modes_tieable_p): ...this and make static.
9602 (TARGET_MODES_TIEABLE_P): Redefine.
9603 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9604 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9605 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9606 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9607 (msp430_modes_tieable_p): Make static.
9608 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9609 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9610 (TARGET_MODES_TIEABLE_P): Redefine.
9611 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9612 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9613 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9614 (TARGET_MODES_TIEABLE_P): Redefine.
9615 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9616 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9617 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9618 (TARGET_MODES_TIEABLE_P): Redefine.
9619 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9620 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9621 (pdp11_modes_tieable_p): New function.
9622 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9623 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9624 (rs6000_modes_tieable_p): New function.
9625 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9626 * config/powerpcspe/powerpcspe.md: Update comment.
9627 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9628 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9629 (TARGET_MODES_TIEABLE_P): Redefine.
9630 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9631 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9632 (rl78_modes_tieable_p): New function.
9633 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9634 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9635 (rs6000_modes_tieable_p): New function.
9636 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9637 * config/rs6000/rs6000.md: Update comment.
9638 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9639 * config/rx/rx.c (rx_modes_tieable_p): New function.
9640 (TARGET_MODES_TIEABLE_P): Redefine.
9641 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9642 * config/s390/s390.c (s390_modes_tieable_p): New function.
9643 (TARGET_MODES_TIEABLE_P): Redefine.
9644 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9645 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9646 (sh_modes_tieable_p): New function.
9647 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9648 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9649 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9650 (sparc_modes_tieable_p): Make static.
9651 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9652 * config/spu/spu.c (spu_modes_tieable_p): New function.
9653 (TARGET_MODES_TIEABLE_P): Redefine.
9654 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9655 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9656 (TARGET_MODES_TIEABLE_P): Redefine.
9657 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9658 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9659 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9660 * config/v850/v850.c (v850_modes_tieable_p): New function.
9661 (TARGET_MODES_TIEABLE_P): Redefine.
9662 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9663 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9664 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9665 (visium_modes_tieable_p): New function.
9666 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9667 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9668 (xtensa_modes_tieable_p): New function.
9669 * system.h (MODES_TIEABLE_P): Poison.
9671 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9672 Alan Hayward <alan.hayward@arm.com>
9673 David Sherwood <david.sherwood@arm.com>
9675 * target.def (hard_regno_mode_ok): New hook.
9676 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9677 (TARGET_HARD_REGNO_MODE_OK): ...this.
9678 * doc/tm.texi.in: Regenerate.
9679 * hooks.h (hook_bool_uint_mode_true): Declare.
9680 * hooks.c (hook_bool_uint_mode_true): New function.
9681 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9683 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9684 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9685 instead of HARD_REGNO_MODE_OK.
9686 * caller-save.c: Include target.h.
9687 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9689 * combine.c (can_combine_p): Likewise.
9690 (combinable_i3pat): Likewise.
9691 (can_change_dest_mode): Likewise.
9692 * expr.c (init_expr_target): Likewise.
9693 (convert_move): Likewise.
9694 (convert_modes): Likewise.
9695 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9696 (setup_prohibited_mode_move_regs): Likewise.
9697 * ira.h (target_ira): Likewise.
9698 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9699 * lra-constraints.c (process_alt_operands): Likewise.
9700 (split_reg): Likewise.
9701 * recog.c (peep2_find_free_register): Likewise.
9702 * ree.c (combine_reaching_defs): Likewise.
9703 * regcprop.c (maybe_mode_change): Likewise.
9704 * reginfo.c (init_reg_sets_1): Likewise.
9705 (choose_hard_reg_mode): Likewise.
9706 (simplifiable_subregs): Likewise.
9707 * regrename.c (check_new_reg_p): Likewise.
9708 * reload.c (find_valid_class): Likewise.
9709 (find_valid_class_1): Likewise.
9710 (reload_inner_reg_of_subreg): Likewise.
9711 (push_reload): Likewise.
9712 (combine_reloads): Likewise.
9713 (find_dummy_reload): Likewise.
9714 (find_reloads): Likewise.
9715 * reload1.c (find_reg): Likewise.
9716 (set_reload_reg): Likewise.
9717 (allocate_reload_reg): Likewise.
9718 (choose_reload_regs): Likewise.
9719 (reload_adjust_reg_for_temp): Likewise.
9720 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9721 (simplify_subreg_regno): Likewise.
9722 * sel-sched.c (init_regs_for_mode): Likewise.
9723 * varasm.c (make_decl_rtl): Likewise.
9724 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9725 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9727 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9728 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9729 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9730 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9731 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9732 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9733 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9734 (arc_mode_class): Delete.
9735 (HARD_REGNO_MODE_OK): Delete.
9736 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9737 (arc_hard_regno_mode_ok): Rename old array to...
9738 (arc_hard_regno_mode_ok_modes): ...this.
9739 (arc_conditional_register_usage): Update accordingly.
9740 (arc_mode_class): Make static.
9741 (arc_hard_regno_mode_ok): New function.
9742 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9743 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9744 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9745 (arm_hard_regno_mode_ok): Make static.
9746 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9748 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9749 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9750 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9752 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9753 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9754 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9755 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9756 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9757 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9758 * config/bfin/predicates.md (valid_reg_operand): Use
9759 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9760 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9761 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9762 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9763 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9764 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9765 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9766 (cr16_hard_regno_mode_ok): Make static and return a bool.
9767 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9768 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9769 (cris_hard_regno_mode_ok): New function.
9770 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9771 (epiphany_mode_class): Delete.
9772 (HARD_REGNO_MODE_OK): Delete.
9773 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9774 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9775 (hard_regno_mode_ok): Rename to...
9776 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9777 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9778 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9780 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9781 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9782 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9783 (frv_hard_regno_mode_ok): Make static and return a bool.
9784 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9786 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9787 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9788 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9789 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9791 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9792 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9793 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9794 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9796 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9797 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9798 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9799 (ia64_hard_regno_mode_ok): New function.
9800 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9801 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9802 (iq2000_hard_regno_mode_ok): New function.
9803 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9804 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9805 (lm32_hard_regno_mode_ok): New function.
9806 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9807 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9808 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9809 instead of HARD_REGNO_MODE_OK.
9810 (m32c_hard_regno_ok): Rename to...
9811 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9812 (m32c_cannot_change_mode_class): Update accordingly.
9813 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9814 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9815 (m32r_mode_class): Delete.
9816 (HARD_REGNO_MODE_OK): Delete.
9817 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9818 (m32r_hard_regno_mode_ok): Rename to...
9819 (m32r_hard_regno_modes): ...this.
9820 (m32r_mode_class): Make static.
9821 (m32r_hard_regno_mode_ok): New function.
9822 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9823 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9824 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9825 (m68k_hard_regno_mode_ok): Make static.
9826 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9827 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9828 (mcore_hard_regno_mode_ok): New function.
9829 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9830 (HARD_REGNO_MODE_OK): Delete.
9831 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9833 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9834 (microblaze_hard_regno_mode_ok): New function.
9835 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9836 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9837 (mips_hard_regno_mode_ok): Delete.
9838 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9839 (mips_hard_regno_mode_ok_p): ...this and make static.
9840 (mips_hard_regno_mode_ok_p): Rename to...
9841 (mips_hard_regno_mode_ok_uncached): ...this.
9842 (mips_hard_regno_mode_ok): New function.
9843 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9844 of HARD_REGNO_MODE_OK.
9845 (mips_option_override): Update after above name changes.
9846 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9847 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9848 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9849 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9850 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9851 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9852 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9853 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9854 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9855 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9856 (msp430_hard_regno_mode_ok): Make static and return a bool.
9857 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9858 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9859 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9861 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9862 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9863 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9864 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9865 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9866 (PA_HARD_REGNO_MODE_OK): ...this
9867 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9868 (PA_HARD_REGNO_MODE_OK): ...this.
9869 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9870 (pa_hard_regno_mode_ok): New function.
9871 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9872 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9873 (pdp11_hard_regno_mode_ok): New function.
9874 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9875 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9877 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9879 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9880 (rs6000_hard_regno_mode_ok): Rename to...
9881 (rs6000_hard_regno_mode_ok_uncached): ...this.
9882 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9883 (rs6000_hard_regno_mode_ok): New function.
9884 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9885 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9886 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9887 (riscv_hard_regno_mode_ok): ...this and make static.
9888 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9889 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9890 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9891 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9892 (rl78_hard_regno_mode_ok): Make static and return bool.
9893 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9894 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9896 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9897 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9898 (rs6000_hard_regno_mode_ok): Rename to...
9899 (rs6000_hard_regno_mode_ok_uncached): ...this.
9900 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9901 (rs6000_hard_regno_mode_ok): New function.
9902 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9903 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9904 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9905 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9906 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9907 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9908 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9909 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9910 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9911 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9912 (sh_hard_regno_mode_ok): Make static.
9913 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9914 instead of HARD_REGNO_MODE_OK.
9915 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9916 (sparc_mode_class): Delete.
9917 (HARD_REGNO_MODE_OK): Delete.
9918 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9919 (hard_regno_mode_classes): Make static.
9920 (sparc_mode_class): Likewise.
9921 (sparc_hard_regno_mode_ok): New function.
9922 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9923 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9924 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9926 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9927 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9928 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9929 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9930 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9931 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9932 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9933 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9934 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9935 (visium_hard_regno_mode_ok): New function.
9936 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9937 instead of HARD_REGNO_MODE_OK.
9938 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9939 (HARD_REGNO_MODE_OK): Delete.
9940 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9941 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9942 (xtensa_option_override): Update accordingly.
9943 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9944 (xtensa_hard_regno_mode_ok): New function.
9945 * system.h (HARD_REGNO_MODE_OK): Poison.
9947 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9948 Alan Hayward <alan.hayward@arm.com>
9949 David Sherwood <david.sherwood@arm.com>
9951 * target.def (hard_regno_call_part_clobbered): New hook.
9952 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9953 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9954 * doc/tm.texi: Regenerate.
9955 * hooks.h (hook_bool_uint_mode_false): Declare.
9956 * hooks.c (hook_bool_uint_mode_false): New function.
9957 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9958 * cselib.c (cselib_process_insn): Use
9959 targetm.hard_regno_call_part_clobbered instead of
9960 HARD_REGNO_CALL_PART_CLOBBERED.
9961 * ira-conflicts.c (ira_build_conflicts): Likewise.
9962 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9963 * lra-constraints.c (need_for_call_save_p): Likewise.
9964 * lra-lives.c: Include target.h.
9965 (check_pseudos_live_through_calls): Use
9966 targetm.hard_regno_call_part_clobbered instead of
9967 HARD_REGNO_CALL_PART_CLOBBERED.
9968 * regcprop.c: Include target.h.
9969 (copyprop_hardreg_forward_1): Use
9970 targetm.hard_regno_call_part_clobbered instead of
9971 HARD_REGNO_CALL_PART_CLOBBERED.
9972 * reginfo.c (choose_hard_reg_mode): Likewise.
9973 * regrename.c (check_new_reg_p): Likewise.
9974 * reload.c (find_equiv_reg): Likewise.
9975 * reload1.c (emit_reload_insns): Likewise.
9976 * sched-deps.c (deps_analyze_insn): Likewise.
9977 * sel-sched.c (init_regs_for_mode): Likewise.
9978 (mark_unavailable_hard_regs): Likewise.
9979 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9980 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9981 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9983 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9984 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9985 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9987 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9989 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9990 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9991 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9993 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9994 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9995 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9997 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9998 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10000 * config/powerpcspe/powerpcspe.c
10001 (rs6000_hard_regno_call_part_clobbered): New function.
10002 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10003 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10004 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10006 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10007 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10008 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10010 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10011 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10012 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10014 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10015 Alan Hayward <alan.hayward@arm.com>
10016 David Sherwood <david.sherwood@arm.com>
10018 * rtl.h (subreg_memory_offset): Declare.
10019 * emit-rtl.c (subreg_memory_offset): New function.
10020 * expmed.c (store_bit_field_1): Use it.
10021 * expr.c (undefined_operand_subword_p): Likewise.
10022 * simplify-rtx.c (simplify_subreg): Likewise.
10024 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10026 PR rtl-optimization/57448
10029 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10030 using atomic_load pattern.
10031 (expand_atomic_store): Likewise.
10033 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10036 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10037 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10038 handling. Use replace_call_with_value with NULL instead of
10039 gsi_replace, unlink_stmt_vdef and release_defs.
10041 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10044 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10046 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10049 * rtl.h (emit_library_call_value_1): Declare.
10050 (emit_library_call): Replace declaration with a series of overloads.
10051 Remove the parameter count argument.
10052 (emit_library_call_value): Likewise.
10053 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10054 with an "rtx_mode_t *".
10055 (emit_library_call_value): Delete.
10056 (emit_library_call): Likewise.
10057 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10058 (asan_emit_allocas_unpoison): Likewise.
10059 * builtins.c (expand_builtin_powi): Likewise.
10060 (expand_asan_emit_allocas_unpoison): Likewise.
10061 * cfgexpand.c (expand_main_function): Likewise.
10062 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10063 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10064 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10065 * config/arm/arm.c (arm_trampoline_init): Likewise.
10066 (arm_call_tls_get_addr): Likewise.
10067 (arm_expand_divmod_libfunc): Likewise.
10068 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10069 (smulsi3_highpart): Likewise.
10070 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10071 (c6x_expand_compare): Likewise.
10072 (c6x_expand_movmem): Likewise.
10073 * config/frv/frv.c (frv_trampoline_init): Likewise.
10074 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10075 (ix86_expand_divmod_libfunc): Likewise.
10076 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10077 (ia64_expand_compare): Likewise.
10078 (ia64_profile_hook): Likewise.
10079 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10080 (nonlocal_goto): Likewise.
10081 (restore_stack_nonlocal): Likewise.
10082 * config/m32r/m32r.c (block_move_call): Likewise.
10083 (m32r_trampoline_init): Likewise.
10084 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10085 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10086 (m68k_call_m68k_read_tp): Likewise.
10087 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10088 (microblaze_expand_divide): Likewise.
10089 * config/mips/mips.h (mips_args): Likewise.
10090 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10091 (MIPS_ICACHE_SYNC): Likewise.
10092 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10093 (nios2_trampoline_init): Likewise.
10094 * config/pa/pa.c (hppa_tls_call): Likewise.
10095 (pa_trampoline_init): Likewise.
10096 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10097 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10098 (expand_strn_compare): Likewise.
10099 (rs6000_generate_compare): Likewise.
10100 (rs6000_expand_float128_convert): Likewise.
10101 (output_profile_hook): Likewise.
10102 (rs6000_trampoline_init): Likewise.
10103 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10104 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10105 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10106 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10107 (rs6000_generate_compare): Likewise.
10108 (rs6000_expand_float128_convert): Likewise.
10109 (output_profile_hook): Likewise.
10110 (rs6000_trampoline_init): Likewise.
10111 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10112 * config/sh/sh.c (sh_trampoline_init): Likewise.
10113 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10114 (sparc_emit_float_lib_cmp): Likewise.
10115 (sparc32_initialize_trampoline): Likewise.
10116 (sparc64_initialize_trampoline): Likewise.
10117 (sparc_profile_hook): Likewise.
10118 * config/spu/spu.c (ea_load_store): Likewise.
10119 * config/spu/spu.md (floatunssidf2): Likewise.
10120 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10121 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10122 * config/visium/visium.c (expand_block_move_4): Likewise.
10123 (expand_block_move_2): Likewise.
10124 (expand_block_move_1): Likewise.
10125 (expand_block_set_4): Likewise.
10126 (expand_block_set_2): Likewise.
10127 (expand_block_set_1): Likewise.
10128 (visium_trampoline_init): Likewise.
10129 (visium_profile_hook): Likewise.
10130 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10131 (xtensa_setup_frame_addresses): Likewise.
10132 (xtensa_trampoline_init): Likewise.
10133 * except.c (sjlj_emit_function_enter): Likewise.
10134 (sjlj_emit_function_exit): Likewise.
10135 * explow.c (allocate_dynamic_stack_space): Likewise.
10136 (probe_stack_range): Likewise.
10137 * expr.c (convert_mode_scalar): Likewise.
10138 * optabs.c (expand_binop): Likewise.
10139 (expand_twoval_binop_libfunc): Likewise.
10140 (expand_unop): Likewise.
10141 (prepare_cmp_insn): Likewise.
10142 (prepare_float_lib_cmp): Likewise.
10143 (expand_float): Likewise.
10144 (expand_fix): Likewise.
10145 (expand_fixed_convert): Likewise.
10146 (maybe_emit_sync_lock_test_and_set): Likewise.
10147 (expand_atomic_compare_and_swap): Likewise.
10148 (expand_mem_thread_fence): Likewise.
10149 (expand_atomic_fetch_op): Likewise.
10151 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10153 * doc/generic.texi (OpenACC): Adjust URL.
10154 * doc/invoke.texi (C Dialect Options): Ditto.
10156 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10158 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10159 predicate for operand 1. Add (m,<S>) constraint.
10160 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10161 Prevent memory operand 1 with register operand 2.
10163 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10165 PR rtl-optimization/82024
10166 * combine.c (try_combine): If the combination result is a PARALLEL,
10167 and we only need to retain the SET in there that would be placed
10168 at I2, check that we can place that at I3 instead, before doing so.
10170 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10173 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10175 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10176 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10179 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10180 Jeff Law <law@redhat.com>
10182 * varasm.c (bss_initializer_p): Do not put constants into .bss
10183 (categorize_decl_for_section): Handle bss_initializer_p returning
10184 false when DECL_INITIAL is NULL.
10186 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10189 * config/s390/s390.c (s390_can_inline_p): New function.
10191 2017-09-01 Jeff Law <law@redhat.com>
10193 PR tree-optimization/82052
10194 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10195 Always initialize the returned slot after a hash table miss
10196 when INSERT is true.
10198 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10200 * config/s390/s390.md (mem_signal_fence): Remove.
10201 * doc/md.texi (mem_signal_fence): Remove.
10202 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10204 * target-insns.def (mem_signal_fence): Remove.
10206 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10209 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10212 * asan.c (create_odr_indicator): Strip name encoding from assembler
10213 name before appending it after __odr_asan_.
10215 2017-09-01 Martin Liska <mliska@suse.cz>
10217 PR tree-optimization/82059
10218 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10219 frequency only when an edge is redirected.
10221 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10223 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10224 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10225 (arc_conditional_register_usage): Remove ARC600 lp_count
10227 (arc_file_start): Emit Tag_ARC_CPU_variation.
10228 (arc_can_use_doloop_p): New conditions to use ZOLs.
10229 (hwloop_fail): New function.
10230 (hwloop_optimize): Likewise.
10231 (hwloop_pattern_reg): Likewise.
10232 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10233 (arc_reorg_loops): New function, calls reorg_loops.
10234 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10235 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10237 (arc_loop_hazard): Remove function, functionality moved into
10239 (arc_hazard): Remove arc_loop_hazard call.
10240 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10241 into hwloop_optimize.
10242 (arc_label_align): Remove ZOL handling.
10243 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10244 * config/arc/arc.md (doloop_begin): Remove pattern.
10245 (doloop_begin_i): Likewise.
10246 (doloop_end_i): Likewise.
10247 (doloop_fallback): Likewise.
10248 (doloop_fallback_m): Likewise.
10249 (doloop_end): Reimplement expand.
10250 (arc_lp): New pattern for LP instruction.
10251 (loop_end): New pattern.
10252 (loop_fail): Likewise.
10253 (decrement_and_branch_until_zero): Likewise.
10254 * config/arc/arc.opt (mlpc-width): New option.
10255 * doc/invoke.texi (mlpc-width): Document option.
10257 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10259 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10260 (arc_ccfsm_advance): Fix checking for delay slots.
10261 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10263 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10265 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10267 (movhi_insn): Update store instruction constraint which are saving
10268 6-bit short immediates.
10269 (movsi_insn): Consider also short scaled load operations.
10270 (zero_extendhisi2_i): Use Usd constraint instead of T.
10271 (extendhisi2_i): Add q constraint.
10272 (arc_clzsi2): Add type and length attributes.
10273 (arc_ctzsi2): Likewise.
10274 * config/arc/constraints.md (Usc): Update constraint, the
10275 assembler can parse two relocations for a single instruction.
10277 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10279 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10280 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10282 2017-08-31 Olivier Hainque <hainque@adacore.com>
10284 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10285 match as powerpc-wrs-vxworks*.
10287 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10289 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10290 register constraint for by-element operand.
10291 (aarch64_mls_elt_merge<mode>): Likewise.
10293 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10295 * config/arc/arc.c (arc_can_follow_jump): Check for short
10298 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10300 * config.gcc: Use g.opt for arc.
10301 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10302 functionality moved to ...
10303 (legitimate_scaled_address_p): New function, ...here.
10304 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10305 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10306 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10308 (arc_override_options): Handle G option.
10309 (arc_output_pic_addr_const): Correct function definition.
10310 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10311 (arc_decl_anon_ns_mem_p): Delete.
10312 (arc_in_small_data_p): Overhaul this function to take into
10313 consideration the value given via G option.
10314 (arc_rewrite_small_data_1): Renamed and corrected old
10315 arc_rewrite_small_data function.
10316 (arc_rewrite_small_data): New function.
10317 (small_data_pattern): Don't use pic_offset_table_rtx.
10318 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10319 * config/arc/simdext.md (movmisalignv2hi): Use
10320 prepare_move_operands function.
10322 (movmisalign*): Likewise.
10323 * doc/invoke.texi (ARC options): Document -G option.
10325 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10327 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10329 * config/arc/arc.c (arc_print_operand): Output scalled address for
10330 sdata whenever is possible.
10331 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10332 load/stores are available.
10333 (compact_sda_memory_operand): Check for the alignment required by
10334 code density instructions.
10335 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10337 * config/arc/constraints.md (Usd): Update constraint.
10338 (Us0): New constraint.
10339 (Usc): Update constraint.
10341 2017-08-31 Richard Biener <rguenther@suse.de>
10343 PR middle-end/82054
10344 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10345 function only once.
10347 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10349 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10350 Resize type_signature.
10352 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10353 Alan Hayward <alan.hayward@arm.com>
10354 David Sherwood <david.sherwood@arm.com>
10356 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10357 subregs whose inner modes can be stored in GPRs.
10358 (aarch64_classify_index): Likewise.
10360 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10361 Alan Hayward <alan.hayward@arm.com>
10362 David Sherwood <david.sherwood@arm.com>
10364 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10365 (V_INT_EQUIV): ...this.
10366 (v_cmp_result): Rename to...
10367 (v_int_equiv): ...this.
10368 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10369 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10370 (copysign<mode>3): Likewise.
10371 (aarch64_simd_bsl<mode>_internal): Likewise.
10372 (aarch64_simd_bsl<mode>): Likewise.
10373 (vec_cmp<mode><mode>): Likewise.
10374 (vcond<mode><mode>): Likewise.
10375 (vcond<v_cmp_mixed><mode>): Likewise.
10376 (vcondu<mode><v_cmp_mixed>): Likewise.
10377 (aarch64_cm<optab><mode>): Likewise.
10378 (aarch64_cmtst<mode>): Likewise.
10379 (aarch64_fac<optab><mode>): Likewise.
10380 (vec_perm_const<mode>): Likewise.
10381 (vcond_mask_<mode><v_cmp_result>): Rename to...
10382 (vcond_mask_<mode><v_int_equiv>): ...this.
10383 (vec_cmp<mode><v_cmp_result>): Rename to...
10384 (vec_cmp<mode><v_int_equiv>): ...this.
10386 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10387 Alan Hayward <alan.hayward@arm.com>
10388 David Sherwood <david.sherwood@arm.com>
10390 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10392 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10393 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10394 (UNSPEC_LD4_DREG): New unspecs.
10395 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10396 (aarch64_ld2<mode>_dreg_be): Replace with...
10397 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10399 (aarch64_ld3<mode>_dreg_le)
10400 (aarch64_ld3<mode>_dreg_be): Replace with...
10401 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10403 (aarch64_ld4<mode>_dreg_le)
10404 (aarch64_ld4<mode>_dreg_be): Replace with...
10405 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10408 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10410 PR tree-optimization/81987
10411 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10412 insert an initializer in a location not dominated by the stride
10415 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10417 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10418 on the entire header of the finally block in the fallthru case.
10420 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10422 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10424 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10426 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10427 rs6000_emit_move_from_cr and call renamed function.
10428 (rs6000_emit_prologue): Call renamed functions.
10429 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10430 movesi_from_cr, remove volatile CRs.
10432 2017-08-30 Jon Beniston <jon@beniston.com>
10433 Richard Biener <rguenther@suse.de>
10435 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10436 of VECTOR_MODE_P check.
10437 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10438 element vector types.
10440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10442 * df.h (df_read_modify_subreg_p): Remove in favor of...
10443 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10444 const_rtx instead of an rtx.
10445 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10446 * df-problems.c (df_word_lr_mark_ref): Likewise.
10447 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10448 (mark_pseudo_reg_dead): Likewise.
10449 (mark_ref_dead): Likewise.
10450 * reginfo.c (init_subregs_of_mode): Likewise.
10451 * sched-deps.c (sched_analyze_1): Likewise.
10452 * df-scan.c (df_def_record_1): Likewise.
10453 (df_uses_record): Likewise.
10454 (df_read_modify_subreg_p): Remove in favor of...
10455 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10456 const_rtx instead of an rtx.
10458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10459 Alan Hayward <alan.hayward@arm.com>
10460 David Sherwood <david.sherwood@arm.com>
10462 * rtl.h (partial_subreg_p): New function.
10463 * caller-save.c (save_call_clobbered_regs): Use it.
10464 * calls.c (expand_call): Likewise.
10465 * combine.c (combinable_i3pat): Likewise.
10466 (simplify_set): Likewise.
10467 (make_extraction): Likewise.
10468 (make_compound_operation_int): Likewise.
10469 (gen_lowpart_or_truncate): Likewise.
10470 (force_to_mode): Likewise.
10471 (make_field_assignment): Likewise.
10472 (reg_truncated_to_mode): Likewise.
10473 (record_truncated_value): Likewise.
10474 (move_deaths): Likewise.
10475 * cse.c (record_jump_cond): Likewise.
10476 (cse_insn): Likewise.
10477 * cselib.c (cselib_lookup_1): Likewise.
10478 * expmed.c (extract_bit_field_using_extv): Likewise.
10479 * function.c (assign_parm_setup_reg): Likewise.
10480 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10481 * ira-build.c (create_insn_allocnos): Likewise.
10482 * lra-coalesce.c (merge_pseudos): Likewise.
10483 * lra-constraints.c (match_reload): Likewise.
10484 (simplify_operand_subreg): Likewise.
10485 (curr_insn_transform): Likewise.
10486 * lra-lives.c (process_bb_lives): Likewise.
10487 * lra.c (new_insn_reg): Likewise.
10488 (lra_substitute_pseudo): Likewise.
10489 * regcprop.c (mode_change_ok): Likewise.
10490 (maybe_mode_change): Likewise.
10491 (copyprop_hardreg_forward_1): Likewise.
10492 * reload.c (push_reload): Likewise.
10493 (find_reloads): Likewise.
10494 (find_reloads_subreg_address): Likewise.
10495 * reload1.c (alter_reg): Likewise.
10496 (eliminate_regs_1): Likewise.
10497 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10499 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10501 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10502 back to if statements, including unpack.
10504 2017-08-30 Martin Liska <mliska@suse.cz>
10506 PR inline-asm/82001
10507 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10509 (func_checker::compare_asm_inputs_outputs): ... this function.
10510 (func_checker::compare_gimple_asm): Use the function to compare
10511 also ASM constrains.
10512 * ipa-icf-gimple.h: Rename the function.
10514 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10515 Alan Hayward <alan.hayward@arm.com>
10516 David Sherwood <david.sherwood@arm.com>
10518 * coretypes.h (complex_mode): New type.
10519 * gdbhooks.py (build_pretty_printer): Handle it.
10520 * machmode.h (complex_mode): New class.
10521 (complex_mode::includes_p): New function.
10522 (is_complex_int_mode): Likewise.
10523 (is_complex_float_mode): Likewise.
10524 * genmodes.c (get_mode_class): Handle complex mode classes.
10525 * function.c (expand_function_end): Use is_complex_int_mode.
10527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10528 Alan Hayward <alan.hayward@arm.com>
10529 David Sherwood <david.sherwood@arm.com>
10531 * coretypes.h (scalar_mode_pod): New typedef.
10532 * gdbhooks.py (build_pretty_printer): Handle it.
10533 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10534 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10535 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10536 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10537 as_a <scalar_mode>.
10539 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10540 Alan Hayward <alan.hayward@arm.com>
10541 David Sherwood <david.sherwood@arm.com>
10543 * machmode.h (mode_for_vector): Take a scalar_mode instead
10545 * stor-layout.c (mode_for_vector): Likewise.
10546 * explow.c (promote_mode): Use as_a <scalar_mode>.
10547 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10550 Alan Hayward <alan.hayward@arm.com>
10551 David Sherwood <david.sherwood@arm.com>
10553 * target.def (preferred_simd_mode): Take a scalar_mode
10554 instead of a machine_mode.
10555 * targhooks.h (default_preferred_simd_mode): Likewise.
10556 * targhooks.c (default_preferred_simd_mode): Likewise.
10557 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10558 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10559 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10560 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10561 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10562 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10563 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10564 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10566 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10567 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10568 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10569 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10570 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10571 * doc/tm.texi: Regenerate.
10572 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10576 Alan Hayward <alan.hayward@arm.com>
10577 David Sherwood <david.sherwood@arm.com>
10579 * target.def (scalar_mode_supported_p): Take a scalar_mode
10580 instead of a machine_mode.
10581 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10582 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10583 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10584 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10585 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10586 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10587 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10588 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10589 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10590 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10591 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10592 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10593 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10594 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10596 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10597 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10598 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10599 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10600 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10602 * doc/tm.texi: Regenerate.
10604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10605 Alan Hayward <alan.hayward@arm.com>
10606 David Sherwood <david.sherwood@arm.com>
10608 * coretypes.h (opt_scalar_mode): New typedef.
10609 * gdbhooks.py (build_pretty_printers): Handle it.
10610 * machmode.h (mode_iterator::get_2xwider): Add overload for
10612 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10614 * expr.c (convert_mode_scalar): Likewise.
10615 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10616 * optabs.c (expand_float): Likewise.
10617 (expand_fix): Likewise.
10618 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10621 Alan Hayward <alan.hayward@arm.com>
10622 David Sherwood <david.sherwood@arm.com>
10624 * optabs.c (expand_float): Explicitly check for scalars before
10625 using a branching expansion.
10626 (expand_fix): Likewise.
10628 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10629 Alan Hayward <alan.hayward@arm.com>
10630 David Sherwood <david.sherwood@arm.com>
10632 * expr.c (convert_mode): Split scalar handling out into...
10633 (convert_mode_scalar): ...this new function. Treat the modes
10636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10637 Alan Hayward <alan.hayward@arm.com>
10638 David Sherwood <david.sherwood@arm.com>
10640 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10642 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10645 Alan Hayward <alan.hayward@arm.com>
10646 David Sherwood <david.sherwood@arm.com>
10648 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10649 rather than a machine_mode.
10650 (fixed_from_string): Likewise.
10651 (fixed_convert): Likewise.
10652 (fixed_convert_from_int): Likewise.
10653 (fixed_convert_from_real): Likewise.
10654 (real_convert_from_fixed): Likewise.
10655 * fixed-value.c (fixed_from_double_int): Likewise.
10656 (fixed_from_string): Likewise.
10657 (fixed_convert): Likewise.
10658 (fixed_convert_from_int): Likewise.
10659 (fixed_convert_from_real): Likewise.
10660 (real_convert_from_fixed): Likewise.
10661 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10664 Alan Hayward <alan.hayward@arm.com>
10665 David Sherwood <david.sherwood@arm.com>
10667 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10668 of separate mode class checks. Do not allow vector modes here.
10669 (immed_wide_int_const): Use as_a <scalar_mode>.
10670 * explow.c (trunc_int_for_mode): Likewise.
10671 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10672 (wi::shwi): Likewise.
10673 (wi::min_value): Likewise.
10674 (wi::max_value): Likewise.
10675 * dwarf2out.c (loc_descriptor): Likewise.
10676 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10677 for CONST_WIDE_INT.
10679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10680 Alan Hayward <alan.hayward@arm.com>
10681 David Sherwood <david.sherwood@arm.com>
10683 * tree.h (SCALAR_TYPE_MODE): New macro.
10684 * expr.c (expand_expr_addr_expr_1): Use it.
10685 (expand_expr_real_2): Likewise.
10686 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10687 (fold_convert_const_fixed_from_int): Likewise.
10688 (fold_convert_const_fixed_from_real): Likewise.
10689 (native_encode_fixed): Likewise
10690 (native_encode_complex): Likewise
10691 (native_encode_vector): Likewise.
10692 (native_interpret_fixed): Likewise.
10693 (native_interpret_real): Likewise.
10694 (native_interpret_complex): Likewise.
10695 (native_interpret_vector): Likewise.
10696 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10697 (simd_clone_adjust_argument_types): Likewise.
10698 (simd_clone_init_simd_arrays): Likewise.
10699 (simd_clone_adjust): Likewise.
10700 * stor-layout.c (layout_type): Likewise.
10701 * tree.c (build_minus_one_cst): Likewise.
10702 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10703 * tree-inline.c (estimate_move_cost): Likewise.
10704 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10705 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10706 (vectorizable_reduction): Likewise.
10707 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10708 (vect_recog_mixed_size_cond_pattern): Likewise.
10709 (check_bool_pattern): Likewise.
10710 (adjust_bool_pattern): Likewise.
10711 (search_type_for_mask_1): Likewise.
10712 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10713 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10714 (vectorizable_load): Likewise.
10715 (vectorizable_store): Likewise.
10716 * ubsan.c (ubsan_encode_value): Likewise.
10717 * varasm.c (output_constant): Likewise.
10719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10720 Alan Hayward <alan.hayward@arm.com>
10721 David Sherwood <david.sherwood@arm.com>
10723 * coretypes.h (scalar_mode): New class.
10724 * machmode.h (scalar_mode): Likewise.
10725 (scalar_mode::includes_p): New function.
10726 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10727 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10728 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10729 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10730 * expmed.c (store_bit_field_1): Likewise.
10731 (extract_bit_field_1): Likewise.
10732 * expr.c (write_complex_part): Likewise.
10733 (read_complex_part): Likewise.
10734 (emit_move_complex_push): Likewise.
10735 (expand_expr_real_2): Likewise.
10736 * function.c (assign_parm_setup_reg): Likewise.
10737 (assign_parms_unsplit_complex): Likewise.
10738 * optabs.c (expand_binop): Likewise.
10739 * rtlanal.c (subreg_get_info): Likewise.
10740 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10741 * varasm.c (output_constant_pool_2): Likewise.
10743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10744 Alan Hayward <alan.hayward@arm.com>
10745 David Sherwood <david.sherwood@arm.com>
10747 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10749 (expmed_mult_highpart_optab): Likewise.
10750 (expmed_mult_highpart): Likewise.
10752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10753 Alan Hayward <alan.hayward@arm.com>
10754 David Sherwood <david.sherwood@arm.com>
10756 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10757 instead of a machine_mode.
10758 (builtin_memset_read_str): Likewise.
10759 * builtins.c (c_readstr): Likewise.
10760 (builtin_memcpy_read_str): Likewise.
10761 (builtin_strncpy_read_str): Likewise.
10762 (builtin_memset_read_str): Likewise.
10763 (builtin_memset_gen_str): Likewise.
10764 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10765 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10766 instead of a machine_mode.
10767 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10769 (make_extraction): Likewise.
10770 (try_widen_shift_mode): Take and return scalar_int_modes instead
10772 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10773 a scalar_int_mode instead of a machine_mode.
10774 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10775 (avr_addr_space_pointer_mode): Likewise.
10776 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10777 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10778 (msp430_unwind_word_mode): Likewise.
10779 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10780 (spu_addr_space_pointer_mode): Likewise.
10781 (spu_addr_space_address_mode): Likewise.
10782 (spu_libgcc_cmp_return_mode): Likewise.
10783 (spu_libgcc_shift_count_mode): Likewise.
10784 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10785 (rl78_addr_space_pointer_mode): Likewise.
10786 (fl78_unwind_word_mode): Likewise.
10787 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10789 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10790 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10791 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10792 (mips_valid_pointer_mode): Likewise.
10793 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10794 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10795 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10797 (ft32_addr_space_address_mode): Likewise.
10798 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10799 scalar_int_mode instead of a machine_mode.
10800 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10802 (m32c_addr_space_address_mode): Likewise.
10803 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10804 (rs6000_eh_return_filter_mode): Likewise.
10805 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10806 (rs6000_eh_return_filter_mode): Likewise.
10807 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10808 (s390_libgcc_shift_count_mode): Likewise.
10809 (s390_unwind_word_mode): Likewise.
10810 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10812 * target.def (mode_rep_extended): Likewise.
10813 (valid_pointer_mode): Likewise.
10814 (addr_space.valid_pointer_mode): Likewise.
10815 (eh_return_filter_mode): Return a scalar_int_mode rather than
10817 (libgcc_cmp_return_mode): Likewise.
10818 (libgcc_shift_count_mode): Likewise.
10819 (unwind_word_mode): Likewise.
10820 (addr_space.pointer_mode): Likewise.
10821 (addr_space.address_mode): Likewise.
10822 * doc/tm.texi: Regenerate.
10823 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10825 (do_jump): Use scalar_int_mode for local variables.
10826 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10827 rather than a machine_mode.
10828 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10829 (scompare_loc_descriptor_wide): Likewise.
10830 (scompare_loc_descriptor_narrow): Likewise.
10831 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10833 * except.c (sjlj_emit_dispatch_table): Likewise.
10834 (expand_builtin_eh_copy_values): Likewise.
10835 * explow.c (convert_memory_address_addr_space_1): Likewise.
10836 Take a scalar_int_mode rather than a machine_mode.
10837 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10838 than a machine_mode.
10839 (memory_address_addr_space): Use scalar_int_mode for local variables.
10840 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10841 rather than a machine_mode.
10842 * expmed.c (mask_rtx): Likewise.
10843 (init_expmed_one_conv): Likewise.
10844 (expand_mult_highpart_adjust): Likewise.
10845 (extract_high_half): Likewise.
10846 (expmed_mult_highpart_optab): Likewise.
10847 (expmed_mult_highpart): Likewise.
10848 (expand_smod_pow2): Likewise.
10849 (expand_sdiv_pow2): Likewise.
10850 (emit_store_flag_int): Likewise.
10851 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10853 (extract_low_bits): Likewise.
10854 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10856 * expr.c (pieces_addr::adjust): Likewise.
10857 (can_store_by_pieces): Likewise.
10858 (store_by_pieces): Likewise.
10859 (clear_by_pieces_1): Likewise.
10860 (expand_expr_addr_expr_1): Likewise.
10861 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10862 (expand_expr_real_1): Likewise.
10863 (try_casesi): Likewise.
10864 * final.c (shorten_branches): Likewise.
10865 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10866 type of "mode" to machine_mode.
10867 * internal-fn.c (expand_arith_overflow_result_store): Take a
10868 scalar_int_mode rather than a machine_mode.
10869 (expand_mul_overflow): Use scalar_int_mode for local variables.
10870 * loop-doloop.c (doloop_modify): Likewise.
10871 (doloop_optimize): Likewise.
10872 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10873 than a machine_mode.
10874 (expand_doubleword_shift_condmove): Likewise.
10875 (expand_doubleword_shift): Likewise.
10876 (expand_doubleword_clz): Likewise.
10877 (expand_doubleword_popcount): Likewise.
10878 (expand_doubleword_parity): Likewise.
10879 (expand_absneg_bit): Use scalar_int_mode for local variables.
10880 (prepare_float_lib_cmp): Likewise.
10881 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10882 rather than a machine_mode.
10883 (convert_memory_address_addr_space): Likewise.
10884 (get_mode_bounds): Likewise.
10885 (get_address_mode): Return a scalar_int_mode rather than a
10887 * rtlanal.c (get_address_mode): Likewise.
10888 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10889 than a machine_mode.
10890 * targhooks.c (default_mode_rep_extended): Likewise.
10891 (default_valid_pointer_mode): Likewise.
10892 (default_addr_space_valid_pointer_mode): Likewise.
10893 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10894 than a machine_mode.
10895 (default_libgcc_cmp_return_mode): Likewise.
10896 (default_libgcc_shift_count_mode): Likewise.
10897 (default_unwind_word_mode): Likewise.
10898 (default_addr_space_pointer_mode): Likewise.
10899 (default_addr_space_address_mode): Likewise.
10900 * targhooks.h (default_eh_return_filter_mode): Likewise.
10901 (default_libgcc_cmp_return_mode): Likewise.
10902 (default_libgcc_shift_count_mode): Likewise.
10903 (default_unwind_word_mode): Likewise.
10904 (default_addr_space_pointer_mode): Likewise.
10905 (default_addr_space_address_mode): Likewise.
10906 (default_mode_rep_extended): Take a scalar_int_mode rather than
10908 (default_valid_pointer_mode): Likewise.
10909 (default_addr_space_valid_pointer_mode): Likewise.
10910 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10912 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10913 rather than a machine_mode.
10914 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10915 for local variables.
10916 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10917 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10918 than a machine_mode.
10920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10921 Alan Hayward <alan.hayward@arm.com>
10922 David Sherwood <david.sherwood@arm.com>
10924 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10925 the mode argument to scalar_int_mode.
10926 (do_jump_by_parts_zero_rtx): Likewise.
10927 (do_jump_by_parts_equality_rtx): Likewise.
10928 (do_jump_by_parts_greater): Take a mode argument.
10929 (do_jump_by_parts_equality): Likewise.
10930 (do_jump_1): Update calls accordingly.
10932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10933 Alan Hayward <alan.hayward@arm.com>
10934 David Sherwood <david.sherwood@arm.com>
10936 * is-a.h (safe_dyn_cast): New function.
10937 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10938 (jump_table_for_label): Likewise.
10939 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10940 instead of an rtx_insn *.
10941 (shorten_branches): Use dyn_cast instead of LABEL_P and
10942 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10943 rtx_jump_table_data::get_data_mode.
10944 (final_scan_insn): Likewise.
10946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10947 Alan Hayward <alan.hayward@arm.com>
10948 David Sherwood <david.sherwood@arm.com>
10950 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10951 trying to combine a full-register integer set with a subreg
10954 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10955 Alan Hayward <alan.hayward@arm.com>
10956 David Sherwood <david.sherwood@arm.com>
10958 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10959 that is always either address_mode or pointer_mode.
10961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10962 Alan Hayward <alan.hayward@arm.com>
10963 David Sherwood <david.sherwood@arm.com>
10965 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10966 when the two are known to be equal.
10968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10969 Alan Hayward <alan.hayward@arm.com>
10970 David Sherwood <david.sherwood@arm.com>
10972 * simplify-rtx.c (simplify_const_unary_operation): Use
10973 is_a <scalar_int_mode> instead of checking for a nonzero
10974 precision. Forcibly convert op_mode to a scalar_int_mode
10975 in that case. More clearly differentiate the operand and
10976 result modes and use the former when deciding what the value
10977 of a count-bits operation should be. Use is_int_mode instead
10978 of checking for a MODE_INT. Remove redundant check for whether
10979 this mode has a zero precision.
10981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10982 Alan Hayward <alan.hayward@arm.com>
10983 David Sherwood <david.sherwood@arm.com>
10985 * optabs.c (widen_leading): Change the type of the mode argument
10986 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10987 (widen_bswap): Likewise.
10988 (expand_parity): Likewise.
10989 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10990 (expand_ffs): Likewise.
10991 (epand_unop): Check for scalar integer modes before calling the
10994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10995 Alan Hayward <alan.hayward@arm.com>
10996 David Sherwood <david.sherwood@arm.com>
10998 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11000 (expand_expr_real_1): Update call accordingly.
11002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11003 Alan Hayward <alan.hayward@arm.com>
11004 David Sherwood <david.sherwood@arm.com>
11006 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11007 value_mode arguments. Use scalar_int_mode internally.
11008 (store_bit_field_1): Rename the new integer mode from imode
11009 to op0_mode and use it instead of GET_MODE (op0). Update calls
11010 to store_split_bit_field, store_bit_field_using_insv and
11011 store_fixed_bit_field.
11012 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11013 Use scalar_int_mode internally. Use a bit count rather than a mode
11014 when calculating the largest bit size for get_best_mode.
11015 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11016 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11017 Remove assertion that OP0 has a scalar integer mode.
11018 (store_split_bit_field): Add op0_mode and value_mode arguments.
11019 Update calls to extract_fixed_bit_field.
11020 (extract_bit_field_using_extv): Add an op0_mode argument.
11021 Use scalar_int_mode internally.
11022 (extract_bit_field_1): Rename the new integer mode from imode to
11023 op0_mode and use it instead of GET_MODE (op0). Update calls to
11024 extract_split_bit_field, extract_bit_field_using_extv and
11025 extract_fixed_bit_field.
11026 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11027 to extract_split_bit_field and extract_fixed_bit_field_1.
11028 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11029 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11030 on the target mode.
11031 (extract_split_bit_field): Add an op0_mode argument. Update call
11032 to extract_fixed_bit_field.
11034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11035 Alan Hayward <alan.hayward@arm.com>
11036 David Sherwood <david.sherwood@arm.com>
11038 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11039 * explow.c (hard_function_value): Likewise.
11040 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11041 convert_to_mode call outside the loop.
11042 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11043 for the mode iterator. Require the mode specified by max_pieces
11045 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11047 (copy_blkmode_to_reg): Likewise.
11048 (set_storage_via_setmem): Likewise.
11049 * optabs.c (prepare_cmp_insn): Likewise.
11050 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11051 * stor-layout.c (finish_bitfield_representative): Likewise.
11053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11054 Alan Hayward <alan.hayward@arm.com>
11055 David Sherwood <david.sherwood@arm.com>
11057 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11058 * expr.c (convert_move): Use them.
11059 (convert_modes): Likewise.
11060 (store_expr_with_bounds): Likewise.
11062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11063 Alan Hayward <alan.hayward@arm.com>
11064 David Sherwood <david.sherwood@arm.com>
11066 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11067 parameter for the mode of "x". Remove the "known_x", "known_mode"
11068 and "known_ret" arguments. Change the type of the mode argument
11069 to scalar_int_mode.
11070 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11071 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11072 (reg_num_sign_bit_copies_for_combine): Likewise.
11073 * rtlanal.c (nonzero_bits1): Likewise.
11074 (num_sign_bit_copies1): Likewise.
11075 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11076 (reg_num_sign_bit_copies_general): Likewise.
11077 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11078 (reg_nonzero_bits_general): Likewise.
11080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11081 Alan Hayward <alan.hayward@arm.com>
11082 David Sherwood <david.sherwood@arm.com>
11084 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11085 than in subroutines. Return 1 for non-integer modes.
11086 (cached_num_sign_bit_copies): Change the type of the mode parameter
11087 to scalar_int_mode.
11088 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11089 classes. Handle CONST_INT_P first and then check whether X also
11090 has a scalar integer mode. Check the same thing for inner registers
11091 of a SUBREG and for values that are being extended or truncated.
11093 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11094 Alan Hayward <alan.hayward@arm.com>
11095 David Sherwood <david.sherwood@arm.com>
11097 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11098 in subroutines. Return the mode mask for non-integer modes.
11099 (cached_nonzero_bits): Change the type of the mode parameter
11100 to scalar_int_mode.
11101 (nonzero_bits1): Likewise. Remove early exit for other mode
11102 classes. Handle CONST_INT_P first and then check whether X
11103 also has a scalar integer mode.
11105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11106 Alan Hayward <alan.hayward@arm.com>
11107 David Sherwood <david.sherwood@arm.com>
11109 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11110 Return a scalar_int_mode and assert that the size is greater than
11112 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11113 (op_by_pieces_d::op_by_pieces_d): Likewise.
11114 (op_by_pieces_d::run): Likewise.
11115 (can_store_by_pieces): Likewise.
11117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11118 Alan Hayward <alan.hayward@arm.com>
11119 David Sherwood <david.sherwood@arm.com>
11121 * combine.c (extract_left_shift): Add a mode argument and update
11123 (make_compound_operation_int): Change the type of the mode parameter
11124 to scalar_int_mode and update the call to extract_left_shift.
11126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11127 Alan Hayward <alan.hayward@arm.com>
11128 David Sherwood <david.sherwood@arm.com>
11130 * combine.c (simplify_and_const_int): Change the type of the mode
11131 parameter to scalar_int_mode.
11132 (simplify_and_const_int_1): Likewise. Update recursive call.
11134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11135 Alan Hayward <alan.hayward@arm.com>
11136 David Sherwood <david.sherwood@arm.com>
11138 * combine.c (simplify_compare_const): Check that the mode is a
11139 scalar_int_mode (rather than VOIDmode) before testing its
11141 (simplify_comparison): Move COMPARISON_P handling out of the
11142 loop and restrict the latter part of the loop to scalar_int_modes.
11143 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11144 and when considering SUBREG_REGs. Use is_int_mode instead of
11145 checking GET_MODE_CLASS against MODE_INT.
11147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11148 Alan Hayward <alan.hayward@arm.com>
11149 David Sherwood <david.sherwood@arm.com>
11151 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11152 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11153 shift_unit_mode and for modes involved in scalar shifts.
11155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11156 Alan Hayward <alan.hayward@arm.com>
11157 David Sherwood <david.sherwood@arm.com>
11159 * combine.c (force_int_to_mode): New function, split out from...
11160 (force_to_mode): ...here. Keep xmode up-to-date and use it
11161 instead of GET_MODE (x).
11163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11164 Alan Hayward <alan.hayward@arm.com>
11165 David Sherwood <david.sherwood@arm.com>
11167 * optabs-query.h (extraction_insn::struct_mode): Change type to
11168 opt_scalar_int_mode and update comment.
11169 (extraction_insn::field_mode): Change type to scalar_int_mode.
11170 (extraction_insn::pos_mode): Likewise.
11171 * combine.c (make_extraction): Update accordingly.
11172 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11173 (get_optab_extraction_insn): Likewise.
11174 * recog.c (simplify_while_replacing): Likewise.
11175 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11176 parameter to opt_scalar_int_mode.
11178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11179 Alan Hayward <alan.hayward@arm.com>
11180 David Sherwood <david.sherwood@arm.com>
11182 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11183 to a scalar_int_mode instead of a machine_mode.
11184 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11185 (get_best_mode): Return a boolean and use a pointer argument to store
11186 the selected mode. Replace the limit mode parameter with a bit limit.
11187 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11188 for the values returned by bit_field_mode_iterator::next_mode.
11189 (store_bit_field): Update call to get_best_mode.
11190 (store_fixed_bit_field): Likewise.
11191 (extract_fixed_bit_field): Likewise.
11192 * expr.c (optimize_bitfield_assignment_op): Likewise.
11193 * fold-const.c (optimize_bit_field_compare): Likewise.
11194 (fold_truth_andor_1): Likewise.
11195 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11196 Update for new type of m_mode.
11197 (get_best_mode): As above.
11199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11200 Alan Hayward <alan.hayward@arm.com>
11201 David Sherwood <david.sherwood@arm.com>
11203 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11204 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11205 (store_bit_field): Check is_a <scalar_int_mode> before calling
11206 strict_volatile_bitfield_p.
11207 (extract_bit_field): Likewise.
11209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11210 Alan Hayward <alan.hayward@arm.com>
11211 David Sherwood <david.sherwood@arm.com>
11213 * target.def (cstore_mode): Return a scalar_int_mode.
11214 * doc/tm.texi: Regenerate.
11215 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11216 * targhooks.h (default_cstore_mode): Likewise.
11217 * targhooks.c (default_cstore_mode): Likewise, using a forced
11219 * expmed.c (emit_cstore): Expect the target of the cstore to be
11222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11223 Alan Hayward <alan.hayward@arm.com>
11224 David Sherwood <david.sherwood@arm.com>
11226 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11228 (niter_desc): Likewise mode.
11229 (iv_analyze): Add a mode parameter.
11231 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11232 and change its type to scalar_int_mode.
11233 * loop-iv.c: Update commentary at head of file.
11234 (iv_constant): Pass the mode paraeter before the rtx it describes
11235 and change its type to scalar_int_mode. Remove VOIDmode handling.
11236 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11237 (iv_extend): Likewise.
11238 (shorten_into_mode): Likewise.
11239 (iv_add): Use scalar_int_mode.
11240 (iv_mult): Likewise.
11241 (iv_shift): Likewise.
11242 (canonicalize_iv_subregs): Likewise.
11243 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11244 it describes and change its mode to scalar_int_mode. Also change
11245 the type of the returned inner_mode to scalar_int_mode.
11246 (get_biv_step): Likewise, turning outer_mode from a pointer
11247 into a direct parameter. Update call to get_biv_step_1.
11248 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11249 iv_constant and get_biv_step.
11250 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11251 and change its type to scalar_int_mode. Don't initialise iv->mode
11252 to VOIDmode and remove later checks for its still being VOIDmode.
11253 Update calls to iv_analyze_op and iv_analyze_expr. Check
11254 is_a <scalar_int_mode> when changing the mode under consideration.
11255 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11256 Update call to iv_analyze_expr.
11257 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11258 inner register is not also a scalar_int_mode. Update call to
11260 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11261 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11262 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11263 separate mode class checks. Update calls to iv_analyze. Remove
11264 fix-up of VOIDmodes after iv_analyze_biv.
11265 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11266 don't have a scalar_int_mode. Update call to biv_p.
11268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11269 Alan Hayward <alan.hayward@arm.com>
11270 David Sherwood <david.sherwood@arm.com>
11272 * cfgexpand.c (convert_debug_memory_address): Use
11273 as_a <scalar_int_mode>.
11274 * combine.c (expand_compound_operation): Likewise.
11275 (make_extraction): Likewise.
11276 (change_zero_ext): Likewise.
11277 (simplify_comparison): Likewise.
11278 * cse.c (cse_insn): Likewise.
11279 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11280 (mem_loc_descriptor): Likewise.
11281 (loc_descriptor): Likewise.
11282 * expmed.c (init_expmed_one_mode): Likewise.
11283 (synth_mult): Likewise.
11284 (emit_store_flag_1): Likewise.
11285 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11286 of a comparison with size.
11287 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11288 (reduce_to_bit_field_precision): Likewise.
11289 * function.c (expand_function_end): Likewise.
11290 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11291 * loop-doloop.c (doloop_modify): Likewise.
11292 * optabs.c (expand_binop): Likewise.
11293 (expand_unop): Likewise.
11294 (expand_copysign_absneg): Likewise.
11295 (prepare_cmp_insn): Likewise.
11296 (maybe_legitimize_operand): Likewise.
11297 * recog.c (const_scalar_int_operand): Likewise.
11298 * rtlanal.c (get_address_mode): Likewise.
11299 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11300 (simplify_cond_clz_ctz): Likewise.
11301 * tree-nested.c (get_nl_goto_field): Likewise.
11302 * tree.c (build_vector_type_for_mode): Likewise.
11303 * var-tracking.c (use_narrower_mode): Likewise.
11305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11306 Alan Hayward <alan.hayward@arm.com>
11307 David Sherwood <david.sherwood@arm.com>
11309 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11310 * builtins.c (expand_builtin_signbit): Use it.
11311 * cfgexpand.c (expand_debug_expr): Likewise.
11312 * dojump.c (do_jump): Likewise.
11313 (do_compare_and_jump): Likewise.
11314 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11315 * expmed.c (make_tree): Likewise.
11316 * expr.c (expand_expr_real_2): Likewise.
11317 (expand_expr_real_1): Likewise.
11318 (try_casesi): Likewise.
11319 * fold-const-call.c (fold_const_call_ss): Likewise.
11320 * fold-const.c (unextend): Likewise.
11321 (extract_muldiv_1): Likewise.
11322 (fold_single_bit_test): Likewise.
11323 (native_encode_int): Likewise.
11324 (native_encode_string): Likewise.
11325 (native_interpret_int): Likewise.
11326 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11327 * internal-fn.c (expand_addsub_overflow): Likewise.
11328 (expand_neg_overflow): Likewise.
11329 (expand_mul_overflow): Likewise.
11330 (expand_arith_overflow): Likewise.
11331 * match.pd: Likewise.
11332 * stor-layout.c (layout_type): Likewise.
11333 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11334 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11335 * tree-ssanames.c (get_range_info): Likewise.
11336 * tree-switch-conversion.c (array_value_type) Likewise.
11337 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11338 (vect_recog_divmod_pattern): Likewise.
11339 (vect_recog_mixed_size_cond_pattern): Likewise.
11340 * tree-vrp.c (extract_range_basic): Likewise.
11341 (simplify_float_conversion_using_ranges): Likewise.
11342 * tree.c (int_fits_type_p): Likewise.
11343 * ubsan.c (instrument_bool_enum_load): Likewise.
11344 * varasm.c (mergeable_string_section): Likewise.
11345 (narrowing_initializer_constant_valid_p): Likewise.
11346 (output_constant): Likewise.
11348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11349 Alan Hayward <alan.hayward@arm.com>
11350 David Sherwood <david.sherwood@arm.com>
11352 * machmode.h (NARROWEST_INT_MODE): New macro.
11353 * expr.c (alignment_for_piecewise_move): Use it instead of
11354 GET_CLASS_NARROWEST_MODE (MODE_INT).
11355 (push_block): Likewise.
11356 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11358 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11361 Alan Hayward <alan.hayward@arm.com>
11362 David Sherwood <david.sherwood@arm.com>
11364 * postreload.c (move2add_valid_value_p): Change the type of the
11365 mode parameter to scalar_int_mode.
11366 (move2add_use_add2_insn): Add a mode parameter and use it instead
11368 (move2add_use_add3_insn): Likewise.
11369 (reload_cse_move2add): Update accordingly.
11371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11372 Alan Hayward <alan.hayward@arm.com>
11373 David Sherwood <david.sherwood@arm.com>
11375 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11377 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11378 * optabs.c (expand_unop): Likewise.
11380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11381 Alan Hayward <alan.hayward@arm.com>
11382 David Sherwood <david.sherwood@arm.com>
11384 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11385 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11386 (popcount_loc_descriptor): Likewise.
11387 (bswap_loc_descriptor): Likewise.
11388 (rotate_loc_descriptor): Likewise.
11389 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11390 calling the functions above.
11392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11393 Alan Hayward <alan.hayward@arm.com>
11394 David Sherwood <david.sherwood@arm.com>
11396 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11398 (try_combine): Likewise.
11399 (simplify_if_then_else): Likewise.
11400 * cse.c (cse_insn): Likewise.
11401 * dwarf2out.c (mem_loc_descriptor): Likewise.
11402 * emit-rtl.c (gen_lowpart_common): Likewise.
11403 * simplify-rtx.c (simplify_truncation): Likewise.
11404 (simplify_binary_operation_1): Likewise.
11405 (simplify_const_relational_operation): Likewise.
11406 (simplify_ternary_operation): Likewise.
11407 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11410 Alan Hayward <alan.hayward@arm.com>
11411 David Sherwood <david.sherwood@arm.com>
11413 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11414 * reload.c (push_reload): Likewise.
11415 (find_reloads): Likewise.
11417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11418 Alan Hayward <alan.hayward@arm.com>
11419 David Sherwood <david.sherwood@arm.com>
11421 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11422 (make_compound_operation_int): Likewise.
11423 (change_zero_ext): Likewise.
11424 * expr.c (convert_move): Likewise.
11425 (convert_modes): Likewise.
11426 * fwprop.c (forward_propagate_subreg): Likewise.
11427 * loop-iv.c (get_biv_step_1): Likewise.
11428 * optabs.c (widen_operand): Likewise.
11429 * postreload.c (move2add_valid_value_p): Likewise.
11430 * recog.c (simplify_while_replacing): Likewise.
11431 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11432 (simplify_binary_operation_1): Likewise. Remove redundant
11433 mode equality check.
11435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11436 Alan Hayward <alan.hayward@arm.com>
11437 David Sherwood <david.sherwood@arm.com>
11439 * combine.c (combine_simplify_rtx): Add checks for
11440 is_a <scalar_int_mode>.
11441 (simplify_if_then_else): Likewise.
11442 (make_field_assignment): Likewise.
11443 (simplify_comparison): Likewise.
11444 * ifcvt.c (noce_try_bitop): Likewise.
11445 * loop-invariant.c (canonicalize_address_mult): Likewise.
11446 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11449 Alan Hayward <alan.hayward@arm.com>
11450 David Sherwood <david.sherwood@arm.com>
11452 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11453 is_a <scalar_int_mode> instead of != BLKmode.
11455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11456 Alan Hayward <alan.hayward@arm.com>
11457 David Sherwood <david.sherwood@arm.com>
11459 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11460 instead of != VOIDmode.
11461 * combine.c (if_then_else_cond): Likewise.
11462 (change_zero_ext): Likewise.
11463 * dwarf2out.c (mem_loc_descriptor): Likewise.
11464 (loc_descriptor): Likewise.
11465 * rtlanal.c (canonicalize_condition): Likewise.
11466 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11469 Alan Hayward <alan.hayward@arm.com>
11470 David Sherwood <david.sherwood@arm.com>
11472 * simplify-rtx.c (simplify_binary_operation_1): Use
11473 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11475 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11476 Alan Hayward <alan.hayward@arm.com>
11477 David Sherwood <david.sherwood@arm.com>
11479 * wide-int.h (int_traits<unsigned char>) New class.
11480 (int_traits<unsigned short>) Likewise.
11481 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11482 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11484 * combine.c (set_nonzero_bits_and_sign_copies): Use
11485 is_a <scalar_int_mode>.
11486 (find_split_point): Likewise.
11487 (combine_simplify_rtx): Likewise.
11488 (simplify_logical): Likewise.
11489 (expand_compound_operation): Likewise.
11490 (expand_field_assignment): Likewise.
11491 (make_compound_operation): Likewise.
11492 (extended_count): Likewise.
11493 (change_zero_ext): Likewise.
11494 (simplify_comparison): Likewise.
11495 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11496 (ucompare_loc_descriptor): Likewise.
11497 (minmax_loc_descriptor): Likewise.
11498 (mem_loc_descriptor): Likewise.
11499 (loc_descriptor): Likewise.
11500 * expmed.c (init_expmed_one_mode): Likewise.
11501 * lra-constraints.c (lra_constraint_offset): Likewise.
11502 * optabs.c (prepare_libcall_arg): Likewise.
11503 * postreload.c (move2add_note_store): Likewise.
11504 * reload.c (operands_match_p): Likewise.
11505 * rtl.h (load_extend_op): Likewise.
11506 * rtlhooks.c (gen_lowpart_general): Likewise.
11507 * simplify-rtx.c (simplify_truncation): Likewise.
11508 (simplify_unary_operation_1): Likewise.
11509 (simplify_binary_operation_1): Likewise.
11510 (simplify_const_binary_operation): Likewise.
11511 (simplify_const_relational_operation): Likewise.
11512 (simplify_subreg): Likewise.
11513 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11514 * var-tracking.c (adjust_mems): Likewise.
11515 (prepare_call_arguments): Likewise.
11517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11518 Alan Hayward <alan.hayward@arm.com>
11519 David Sherwood <david.sherwood@arm.com>
11521 * machmode.h (is_int_mode): New fuction.
11522 * combine.c (find_split_point): Use it.
11523 (combine_simplify_rtx): Likewise.
11524 (simplify_if_then_else): Likewise.
11525 (simplify_set): Likewise.
11526 (simplify_shift_const_1): Likewise.
11527 (simplify_comparison): Likewise.
11528 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11529 * cse.c (notreg_cost): Likewise.
11530 (cse_insn): Likewise.
11531 * cselib.c (cselib_lookup_1): Likewise.
11532 * dojump.c (do_jump_1): Likewise.
11533 (do_compare_rtx_and_jump): Likewise.
11534 * dse.c (get_call_args): Likewise.
11535 * dwarf2out.c (rtl_for_decl_init): Likewise.
11536 (native_encode_initializer): Likewise.
11537 * expmed.c (emit_store_flag_1): Likewise.
11538 (emit_store_flag): Likewise.
11539 * expr.c (convert_modes): Likewise.
11540 (store_field): Likewise.
11541 (expand_expr_real_1): Likewise.
11542 * fold-const.c (fold_read_from_constant_string): Likewise.
11543 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11544 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11545 * optabs.c (expand_binop): Likewise.
11546 (expand_unop): Likewise.
11547 (expand_abs_nojump): Likewise.
11548 (expand_one_cmpl_abs_nojump): Likewise.
11549 * simplify-rtx.c (mode_signbit_p): Likewise.
11550 (val_signbit_p): Likewise.
11551 (val_signbit_known_set_p): Likewise.
11552 (val_signbit_known_clear_p): Likewise.
11553 (simplify_relational_operation_1): Likewise.
11554 * tree.c (vector_type_mode): Likewise.
11556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11557 Alan Hayward <alan.hayward@arm.com>
11558 David Sherwood <david.sherwood@arm.com>
11560 * machmode.h (smallest_mode_for_size): Fix formatting.
11561 (smallest_int_mode_for_size): New function.
11562 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11563 instead of smallest_mode_for_size.
11564 * combine.c (make_extraction): Likewise.
11565 * config/arc/arc.c (arc_expand_movmem): Likewise.
11566 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11567 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11568 * config/s390/s390.c (s390_expand_insv): Likewise.
11569 * config/sparc/sparc.c (assign_int_registers): Likewise.
11570 * config/spu/spu.c (spu_function_value): Likewise.
11571 (spu_function_arg): Likewise.
11572 * coverage.c (get_gcov_type): Likewise.
11573 (get_gcov_unsigned_t): Likewise.
11574 * dse.c (find_shift_sequence): Likewise.
11575 * expmed.c (store_bit_field_1): Likewise.
11576 * expr.c (convert_move): Likewise.
11577 (store_field): Likewise.
11578 * internal-fn.c (expand_arith_overflow): Likewise.
11579 * optabs-query.c (get_best_extraction_insn): Likewise.
11580 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11581 * stor-layout.c (layout_type): Likewise.
11582 (initialize_sizetypes): Likewise.
11583 * targhooks.c (default_get_mask_mode): Likewise.
11584 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11587 Alan Hayward <alan.hayward@arm.com>
11588 David Sherwood <david.sherwood@arm.com>
11590 * machmode.h (opt_mode::else_blk): New function.
11591 (int_mode_for_mode): Declare.
11592 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11593 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11595 * cfgexpand.c (expand_debug_expr): Likewise.
11596 * combine.c (gen_lowpart_or_truncate): Likewise.
11597 (gen_lowpart_for_combine): Likewise.
11598 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11599 * config/avr/avr.c (avr_to_int_mode): Likewise.
11600 (avr_out_plus_1): Likewise.
11601 (avr_out_plus): Likewise.
11602 (avr_out_round): Likewise.
11603 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11604 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11605 (s390_expand_vcond): Likewise.
11606 * config/spu/spu.c (spu_split_immediate): Likewise.
11607 (spu_expand_mov): Likewise.
11608 * dse.c (get_stored_val): Likewise.
11609 * expmed.c (store_bit_field_1): Likewise.
11610 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11612 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11613 (extract_low_bits): Likewise.
11614 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11615 handling rather than repeating the check.
11616 (emit_group_store): Likewise.
11617 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11618 * optabs.c (expand_absneg_bit): Likewise.
11619 (expand_copysign_absneg): Likewise.
11620 (expand_copysign_bit): Likewise.
11621 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11622 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11623 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11624 * var-tracking.c (prepare_call_arguments): Likewise.
11625 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11626 int_mode_for_mode instead of mode_for_size.
11627 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11630 Alan Hayward <alan.hayward@arm.com>
11631 David Sherwood <david.sherwood@arm.com>
11633 * machmode.h (int_mode_for_size): New function.
11634 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11635 instead of mode_for_size.
11636 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11638 * combine.c (expand_field_assignment): Use int_mode_for_size
11639 instead of mode_for_size.
11640 (make_extraction): Likewise.
11641 (simplify_shift_const_1): Likewise.
11642 (simplify_comparison): Likewise.
11643 * dojump.c (do_jump): Likewise.
11644 * dwarf2out.c (mem_loc_descriptor): Likewise.
11645 * emit-rtl.c (init_derived_machine_modes): Likewise.
11646 * expmed.c (flip_storage_order): Likewise.
11647 (convert_extracted_bit_field): Likewise.
11648 * expr.c (copy_blkmode_from_reg): Likewise.
11649 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11650 * internal-fn.c (expand_mul_overflow): Likewise.
11651 * lower-subreg.c (simple_move): Likewise.
11652 * optabs-libfuncs.c (init_optabs): Likewise.
11653 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11654 * tree.c (vector_type_mode): Likewise.
11655 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11656 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11657 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11658 * tree-vect-stmts.c (vectorizable_load): Likewise.
11659 (vectorizable_store): Likewise.
11661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11662 Alan Hayward <alan.hayward@arm.com>
11663 David Sherwood <david.sherwood@arm.com>
11665 * coretypes.h (pod_mode): New type.
11666 (scalar_int_mode_pod): New typedef.
11667 * machmode.h (pod_mode): New class.
11668 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11669 * genmodes.c (emit_mode_int_n): Update accordingly.
11670 * lower-subreg.h (target_lower_subreg): Change type to
11671 scalar_int_mode_pod.
11672 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11673 scalar_int_mode_pod.
11675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11676 Alan Hayward <alan.hayward@arm.com>
11677 David Sherwood <david.sherwood@arm.com>
11679 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11680 machine_mode to scalar_int_mode.
11681 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11682 (rs6000_option_override_internal): Remove cast to int.
11683 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11684 machine_mode to scalar_int_mode.
11685 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11686 (rs6000_option_override_internal): Remove cast to int.
11687 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11688 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11690 * config/s390/s390.c (s390_expand_builtin): Likewise.
11691 * coretypes.h (scalar_int_mode): New type.
11692 (opt_scalar_int_mode): New typedef.
11693 * machmode.h (scalar_int_mode): New class.
11694 (scalar_int_mode::includes_p): New function.
11695 (byte_mode): Change type to scalar_int_mode.
11696 (word_mode): Likewise.
11697 (ptr_mode): Likewise.
11698 * emit-rtl.c (byte_mode): Likewise.
11699 (word_mode): Likewise.
11700 (ptr_mode): Likewise.
11701 (init_derived_machine_modes): Update accordingly.
11702 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11703 and MODE_PARTIAL_INT.
11704 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11705 opt_scalar_int_mode.
11707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11708 Alan Hayward <alan.hayward@arm.com>
11709 David Sherwood <david.sherwood@arm.com>
11711 * target.def (libgcc_floating_mode_supported_p): Take a
11713 * doc/tm.texi: Regenerate.
11714 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11716 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11717 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11721 Alan Hayward <alan.hayward@arm.com>
11722 David Sherwood <david.sherwood@arm.com>
11724 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11725 * doc/tm.texi: Regenerate.
11726 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11727 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11728 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11729 * targhooks.h (default_floatn_mode): Likewise.
11730 * targhooks.c (default_floatn_mode): Likewise.
11731 * tree.c (build_common_tree_nodes): Update accordingly.
11733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11734 Alan Hayward <alan.hayward@arm.com>
11735 David Sherwood <david.sherwood@arm.com>
11737 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11738 (mode_iterator::iterate_p): Likewise.
11739 (mode_iterator::get_wider): Likewise.
11740 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11743 Alan Hayward <alan.hayward@arm.com>
11744 David Sherwood <david.sherwood@arm.com>
11746 * coretypes.h (opt_scalar_float_mode): New typedef.
11747 * machmode.h (float_mode_for_size): New function.
11748 * emit-rtl.c (double_mode): Delete.
11749 (init_emit_once): Use float_mode_for_size.
11750 * stor-layout.c (layout_type): Likewise.
11751 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11754 Alan Hayward <alan.hayward@arm.com>
11755 David Sherwood <david.sherwood@arm.com>
11757 * output.h (assemble_real): Take a scalar_float_mode.
11758 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11759 * config/arm/arm.md (consttable_4): Likewise.
11760 (consttable_8): Likewise.
11761 (consttable_16): Likewise.
11762 * config/mips/mips.md (consttable_float): Likewise.
11763 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11764 * varasm.c (assemble_real): Take a scalar_float_mode.
11765 (output_constant_pool_2): Update accordingly.
11766 (output_constant): Likewise.
11768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11769 Alan Hayward <alan.hayward@arm.com>
11770 David Sherwood <david.sherwood@arm.com>
11772 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11773 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11774 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11775 (native_encode_real): Likewise.
11776 (native_interpret_real): Likewise.
11777 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11778 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11781 Alan Hayward <alan.hayward@arm.com>
11782 David Sherwood <david.sherwood@arm.com>
11784 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11785 <scalar_float_mode>. Simplify.
11786 (gen_extend_conv_libfunc): Likewise.
11788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11789 Alan Hayward <alan.hayward@arm.com>
11790 David Sherwood <david.sherwood@arm.com>
11792 * coretypes.h (scalar_float_mode): New type.
11793 * machmode.h (mode_traits::from_int): Use machine_mode if
11794 USE_ENUM_MODES is defined.
11795 (is_a): New function.
11797 (dyn_cast): Likewise.
11798 (scalar_float_mode): New class.
11799 (scalar_float_mode::includes_p): New function.
11800 (is_float_mode): Likewise.
11801 * gdbhooks.py (MachineModePrinter): New class.
11802 (build_pretty_printer): Use it for scalar_float_mode.
11803 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11804 (format_helper::format_helper): Turn into a template.
11805 * genmodes.c (get_mode_class): New function.
11806 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11807 or machine_mode if none.
11808 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11809 as_a <scalar_float_mode>.
11810 * dwarf2out.c (mem_loc_descriptor): Likewise.
11811 (insert_float): Likewise.
11812 (add_const_value_attribute): Likewise.
11813 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11814 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11815 (expand_unop): Update accordingly.
11816 (expand_abs_nojump): Likewise.
11817 (expand_copysign_absneg): Take a scalar_float_mode.
11818 (expand_copysign_bit): Likewise.
11819 (expand_copysign): Update accordingly.
11821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11822 Alan Hayward <alan.hayward@arm.com>
11823 David Sherwood <david.sherwood@arm.com>
11825 * coretypes.h (opt_mode): New class.
11826 * machmode.h (opt_mode): Likewise.
11827 (opt_mode::else_void): New function.
11828 (opt_mode::require): Likewise.
11829 (opt_mode::exists): Likewise.
11830 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11831 (GET_MODE_2XWIDER_MODE): Likewise.
11832 (mode_iterator::get_wider): Update accordingly.
11833 (mode_iterator::get_2xwider): Likewise.
11834 (mode_iterator::get_known_wider): Likewise, turning into a template.
11835 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11836 forcing a wider mode to exist.
11837 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11838 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11839 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11840 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11841 * lower-subreg.c (init_lower_subreg): Likewise.
11842 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11843 on the final iteration.
11844 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11845 a wider mode exists before asking for a move pattern.
11846 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11847 forcing a wider mode to exist.
11848 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11849 returning false if no such mode exists.
11850 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11851 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11852 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11853 Avoid checking for a MODE_INT if we already know the mode is not a
11855 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11856 forcing a wider mode to exist.
11857 (expmed_mult_highpart_optab): Likewise.
11858 (expmed_mult_highpart): Likewise.
11859 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11861 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11862 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11863 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11864 * internal-fn.c (expand_mul_overflow): Update use of
11865 GET_MODE_2XWIDER_MODE.
11866 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11867 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11868 GET_MODE_WIDER_MODE.
11869 (convert_plusminus_to_widen): Likewise.
11870 * tree-switch-conversion.c (array_value_type): Likewise.
11871 * var-tracking.c (emit_note_insn_var_location): Likewise.
11872 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11873 Return false inside rather than outside the loop if no wider mode
11875 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11876 and GET_MODE_2XWIDER_MODE
11877 (can_compare_p): Use else_void.
11878 * gdbhooks.py (OptMachineModePrinter): New class.
11879 (build_pretty_printer): Use it for opt_mode.
11881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11882 Alan Hayward <alan.hayward@arm.com>
11883 David Sherwood <david.sherwood@arm.com>
11885 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11886 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11889 Alan Hayward <alan.hayward@arm.com>
11890 David Sherwood <david.sherwood@arm.com>
11892 * machmode.h (mode_traits): New structure.
11893 (get_narrowest_mode): New function.
11894 (mode_iterator::start): Likewise.
11895 (mode_iterator::iterate_p): Likewise.
11896 (mode_iterator::get_wider): Likewise.
11897 (mode_iterator::get_known_wider): Likewise.
11898 (mode_iterator::get_2xwider): Likewise.
11899 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11900 (FOR_EACH_MODE): Likewise.
11901 (FOR_EACH_MODE_FROM): Likewise.
11902 (FOR_EACH_MODE_UNTIL): Likewise.
11903 (FOR_EACH_WIDER_MODE): Likewise.
11904 (FOR_EACH_2XWIDER_MODE): Likewise.
11905 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11906 * combine.c (simplify_comparison): Likewise
11907 * config/i386/i386.c (type_natural_mode): Likewise.
11908 * cse.c (cse_insn): Likewise.
11909 * dse.c (find_shift_sequence): Likewise.
11910 * emit-rtl.c (init_derived_machine_modes): Likewise.
11911 (init_emit_once): Likewise.
11912 * explow.c (hard_function_value): Likewise.
11913 * expmed.c (extract_fixed_bit_field_1): Likewise.
11914 (extract_bit_field_1): Likewise.
11915 (expand_divmod): Likewise.
11916 (emit_store_flag_1): Likewise.
11917 * expr.c (init_expr_target): Likewise.
11918 (convert_move): Likewise.
11919 (alignment_for_piecewise_move): Likewise.
11920 (widest_int_mode_for_size): Likewise.
11921 (emit_block_move_via_movmem): Likewise.
11922 (copy_blkmode_to_reg): Likewise.
11923 (set_storage_via_setmem): Likewise.
11924 (compress_float_constant): Likewise.
11925 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11926 * optabs-query.c (get_best_extraction_insn): Likewise.
11927 * optabs.c (expand_binop): Likewise.
11928 (expand_twoval_unop): Likewise.
11929 (expand_twoval_binop): Likewise.
11930 (widen_leading): Likewise.
11931 (widen_bswap): Likewise.
11932 (expand_parity): Likewise.
11933 (expand_unop): Likewise.
11934 (prepare_cmp_insn): Likewise.
11935 (prepare_float_lib_cmp): Likewise.
11936 (expand_float): Likewise.
11937 (expand_fix): Likewise.
11938 (expand_sfix_optab): Likewise.
11939 * postreload.c (move2add_use_add2_insn): Likewise.
11940 * reg-stack.c (reg_to_stack): Likewise.
11941 * reginfo.c (choose_hard_reg_mode): Likewise.
11942 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11943 * stor-layout.c (mode_for_size): Likewise.
11944 (smallest_mode_for_size): Likewise.
11945 (mode_for_vector): Likewise.
11946 (finish_bitfield_representative): Likewise.
11947 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11948 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11949 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11950 * var-tracking.c (prepare_call_arguments): Likewise.
11952 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11953 Alan Hayward <alan.hayward@arm.com>
11954 David Sherwood <david.sherwood@arm.com>
11956 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11957 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11958 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11959 * machmode.h (mode_size): Move earlier in file.
11960 (mode_precision): Likewise.
11961 (mode_inner): Likewise.
11962 (mode_nunits): Likewise.
11963 (mode_unit_size): Likewise.
11964 (unit_unit_precision): Likewise.
11965 (mode_wider): Likewise.
11966 (mode_2xwider): Likewise.
11967 (machine_mode): New class.
11968 (mode_to_bytes): New function.
11969 (mode_to_bits): Likewise.
11970 (mode_to_precision): Likewise.
11971 (mode_to_inner): Likewise.
11972 (mode_to_unit_size): Likewise.
11973 (mode_to_unit_precision): Likewise.
11974 (mode_to_nunits): Likewise.
11975 (GET_MODE_SIZE): Use mode_to_bytes.
11976 (GET_MODE_BITSIZE): Use mode_to_bits.
11977 (GET_MODE_PRECISION): Use mode_to_precision.
11978 (GET_MODE_INNER): Use mode_to_inner.
11979 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11980 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11981 (GET_MODE_NUNITS): Use mode_to_nunits.
11982 * system.h (ALWAYS_INLINE): New macro.
11983 * config/powerpcspe/powerpcspe-c.c
11984 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11985 int for arg1_mode and arg2_mode.
11987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11988 Alan Hayward <alan.hayward@arm.com>
11989 David Sherwood <david.sherwood@arm.com>
11991 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11992 Prefix mode names with E_ in case statements.
11993 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11994 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11995 (aarch64_split_simd_move): Likewise.
11996 (aarch64_gen_storewb_pair): Likewise.
11997 (aarch64_gen_loadwb_pair): Likewise.
11998 (aarch64_gen_store_pair): Likewise.
11999 (aarch64_gen_load_pair): Likewise.
12000 (aarch64_get_condition_code_1): Likewise.
12001 (aarch64_constant_pool_reload_icode): Likewise.
12002 (get_rsqrte_type): Likewise.
12003 (get_rsqrts_type): Likewise.
12004 (get_recpe_type): Likewise.
12005 (get_recps_type): Likewise.
12006 (aarch64_gimplify_va_arg_expr): Likewise.
12007 (aarch64_simd_container_mode): Likewise.
12008 (aarch64_emit_load_exclusive): Likewise.
12009 (aarch64_emit_store_exclusive): Likewise.
12010 (aarch64_expand_compare_and_swap): Likewise.
12011 (aarch64_gen_atomic_cas): Likewise.
12012 (aarch64_emit_bic): Likewise.
12013 (aarch64_emit_atomic_swap): Likewise.
12014 (aarch64_emit_atomic_load_op): Likewise.
12015 (aarch64_evpc_trn): Likewise.
12016 (aarch64_evpc_uzp): Likewise.
12017 (aarch64_evpc_zip): Likewise.
12018 (aarch64_evpc_ext): Likewise.
12019 (aarch64_evpc_rev): Likewise.
12020 (aarch64_evpc_dup): Likewise.
12021 (aarch64_gen_ccmp_first): Likewise.
12022 (aarch64_gen_ccmp_next): Likewise.
12023 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12024 (alpha_emit_xfloating_libcall): Likewise.
12025 (emit_insxl): Likewise.
12026 (alpha_arg_type): Likewise.
12027 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12028 (arc_preferred_simd_mode): Likewise.
12029 (arc_secondary_reload): Likewise.
12030 (get_arc_condition_code): Likewise.
12031 (arc_print_operand): Likewise.
12032 (arc_legitimate_constant_p): Likewise.
12033 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12034 * config/arc/arc.md (casesi_load): Likewise.
12035 (casesi_compact_jump): Likewise.
12036 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12037 (cc_use_register): Likewise.
12038 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12039 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12040 (arm_init_iwmmxt_builtins): Likewise.
12041 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12042 (neon_expand_vector_init): Likewise.
12043 (arm_attr_length_move_neon): Likewise.
12044 (maybe_get_arm_condition_code): Likewise.
12045 (arm_emit_vector_const): Likewise.
12046 (arm_preferred_simd_mode): Likewise.
12047 (arm_output_iwmmxt_tinsr): Likewise.
12048 (thumb1_output_casesi): Likewise.
12049 (thumb2_output_casesi): Likewise.
12050 (arm_emit_load_exclusive): Likewise.
12051 (arm_emit_store_exclusive): Likewise.
12052 (arm_expand_compare_and_swap): Likewise.
12053 (arm_evpc_neon_vuzp): Likewise.
12054 (arm_evpc_neon_vzip): Likewise.
12055 (arm_evpc_neon_vrev): Likewise.
12056 (arm_evpc_neon_vtrn): Likewise.
12057 (arm_evpc_neon_vext): Likewise.
12058 (arm_validize_comparison): Likewise.
12059 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12060 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12061 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12062 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12063 (c6x_preferred_simd_mode): Likewise.
12064 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12065 (epiphany_rtx_costs): Likewise.
12066 * config/epiphany/predicates.md (proper_comparison_operator):
12068 * config/frv/frv.c (condexec_memory_operand): Likewise.
12069 (frv_emit_move): Likewise.
12070 (output_move_single): Likewise.
12071 (output_condmove_single): Likewise.
12072 (frv_hard_regno_mode_ok): Likewise.
12073 (frv_matching_accg_mode): Likewise.
12074 * config/h8300/h8300.c (split_adds_subs): Likewise.
12075 (h8300_rtx_costs): Likewise.
12076 (h8300_print_operand): Likewise.
12077 (compute_mov_length): Likewise.
12078 (output_logical_op): Likewise.
12079 (compute_logical_op_length): Likewise.
12080 (compute_logical_op_cc): Likewise.
12081 (h8300_shift_needs_scratch_p): Likewise.
12082 (output_a_shift): Likewise.
12083 (compute_a_shift_length): Likewise.
12084 (compute_a_shift_cc): Likewise.
12085 (expand_a_rotate): Likewise.
12086 (output_a_rotate): Likewise.
12087 * config/i386/i386.c (classify_argument): Likewise.
12088 (function_arg_advance_32): Likewise.
12089 (function_arg_32): Likewise.
12090 (function_arg_64): Likewise.
12091 (function_value_64): Likewise.
12092 (ix86_gimplify_va_arg): Likewise.
12093 (ix86_legitimate_constant_p): Likewise.
12094 (put_condition_code): Likewise.
12095 (split_double_mode): Likewise.
12096 (ix86_avx256_split_vector_move_misalign): Likewise.
12097 (ix86_expand_vector_logical_operator): Likewise.
12098 (ix86_split_idivmod): Likewise.
12099 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12100 (ix86_build_const_vector): Likewise.
12101 (ix86_build_signbit_mask): Likewise.
12102 (ix86_match_ccmode): Likewise.
12103 (ix86_cc_modes_compatible): Likewise.
12104 (ix86_expand_branch): Likewise.
12105 (ix86_expand_sse_cmp): Likewise.
12106 (ix86_expand_sse_movcc): Likewise.
12107 (ix86_expand_int_sse_cmp): Likewise.
12108 (ix86_expand_vec_perm_vpermi2): Likewise.
12109 (ix86_expand_vec_perm): Likewise.
12110 (ix86_expand_sse_unpack): Likewise.
12111 (ix86_expand_int_addcc): Likewise.
12112 (ix86_split_to_parts): Likewise.
12113 (ix86_vectorize_builtin_gather): Likewise.
12114 (ix86_vectorize_builtin_scatter): Likewise.
12115 (avx_vpermilp_parallel): Likewise.
12116 (inline_memory_move_cost): Likewise.
12117 (ix86_tieable_integer_mode_p): Likewise.
12118 (x86_maybe_negate_const_int): Likewise.
12119 (ix86_expand_vector_init_duplicate): Likewise.
12120 (ix86_expand_vector_init_one_nonzero): Likewise.
12121 (ix86_expand_vector_init_one_var): Likewise.
12122 (ix86_expand_vector_init_concat): Likewise.
12123 (ix86_expand_vector_init_interleave): Likewise.
12124 (ix86_expand_vector_init_general): Likewise.
12125 (ix86_expand_vector_set): Likewise.
12126 (ix86_expand_vector_extract): Likewise.
12127 (emit_reduc_half): Likewise.
12128 (ix86_emit_i387_round): Likewise.
12129 (ix86_mangle_type): Likewise.
12130 (ix86_expand_round_sse4): Likewise.
12131 (expand_vec_perm_blend): Likewise.
12132 (canonicalize_vector_int_perm): Likewise.
12133 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12134 (expand_vec_perm_1): Likewise.
12135 (expand_vec_perm_interleave3): Likewise.
12136 (expand_vec_perm_even_odd_pack): Likewise.
12137 (expand_vec_perm_even_odd_1): Likewise.
12138 (expand_vec_perm_broadcast_1): Likewise.
12139 (ix86_vectorize_vec_perm_const_ok): Likewise.
12140 (ix86_expand_vecop_qihi): Likewise.
12141 (ix86_expand_mul_widen_hilo): Likewise.
12142 (ix86_expand_sse2_abs): Likewise.
12143 (ix86_expand_pextr): Likewise.
12144 (ix86_expand_pinsr): Likewise.
12145 (ix86_preferred_simd_mode): Likewise.
12146 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12147 * config/i386/sse.md (*andnot<mode>3): Likewise.
12148 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12149 (*<code><mode>3): Likewise.
12150 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12151 (ia64_expand_atomic_op): Likewise.
12152 (ia64_arg_type): Likewise.
12153 (ia64_mode_to_int): Likewise.
12154 (ia64_scalar_mode_supported_p): Likewise.
12155 (ia64_vector_mode_supported_p): Likewise.
12156 (expand_vec_perm_broadcast): Likewise.
12157 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12158 (iq2000_function_arg_advance): Likewise.
12159 (iq2000_function_arg): Likewise.
12160 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12161 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12162 (m68k_libcall_value): Likewise.
12163 (m68k_function_value): Likewise.
12164 (sched_attr_op_type): Likewise.
12165 * config/mcore/mcore.c (mcore_output_move): Likewise.
12166 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12168 (microblaze_function_arg): Likewise.
12169 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12170 (mips_print_operand): Likewise.
12171 (mips_mode_ok_for_mov_fmt_p): Likewise.
12172 (mips_vector_mode_supported_p): Likewise.
12173 (mips_preferred_simd_mode): Likewise.
12174 (mips_expand_vpc_loongson_even_odd): Likewise.
12175 (mips_expand_vec_unpack): Likewise.
12176 (mips_expand_vi_broadcast): Likewise.
12177 (mips_expand_vector_init): Likewise.
12178 (mips_expand_vec_reduc): Likewise.
12179 (mips_expand_msa_cmp): Likewise.
12180 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12181 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12182 (cc_flags_for_mode): Likewise.
12183 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12184 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12185 (nds32_output_casesi_pc_relative): Likewise.
12186 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12187 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12188 (nvptx_gen_unpack): Likewise.
12189 (nvptx_gen_pack): Likewise.
12190 (nvptx_gen_shuffle): Likewise.
12191 (nvptx_gen_wcast): Likewise.
12192 (nvptx_preferred_simd_mode): Likewise.
12193 * config/pa/pa.c (pa_secondary_reload): Likewise.
12194 * config/pa/predicates.md (base14_operand): Likewise.
12195 * config/powerpcspe/powerpcspe-c.c
12196 (altivec_resolve_overloaded_builtin): Likewise.
12197 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12199 (rs6000_preferred_simd_mode): Likewise.
12200 (output_vec_const_move): Likewise.
12201 (rs6000_expand_vector_extract): Likewise.
12202 (rs6000_split_vec_extract_var): Likewise.
12203 (reg_offset_addressing_ok_p): Likewise.
12204 (rs6000_legitimate_offset_address_p): Likewise.
12205 (rs6000_legitimize_address): Likewise.
12206 (rs6000_emit_set_const): Likewise.
12207 (rs6000_const_vec): Likewise.
12208 (rs6000_emit_move): Likewise.
12209 (spe_build_register_parallel): Likewise.
12210 (rs6000_darwin64_record_arg_recurse): Likewise.
12211 (swap_selector_for_mode): Likewise.
12212 (spe_init_builtins): Likewise.
12213 (paired_init_builtins): Likewise.
12214 (altivec_init_builtins): Likewise.
12215 (do_load_for_compare): Likewise.
12216 (rs6000_generate_compare): Likewise.
12217 (rs6000_expand_float128_convert): Likewise.
12218 (emit_load_locked): Likewise.
12219 (emit_store_conditional): Likewise.
12220 (rs6000_output_function_epilogue): Likewise.
12221 (rs6000_handle_altivec_attribute): Likewise.
12222 (rs6000_function_value): Likewise.
12223 (emit_fusion_gpr_load): Likewise.
12224 (emit_fusion_p9_load): Likewise.
12225 (emit_fusion_p9_store): Likewise.
12226 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12227 (fusion_gpr_mem_load): Likewise.
12228 (fusion_addis_mem_combo_load): Likewise.
12229 (fusion_addis_mem_combo_store): Likewise.
12230 * config/rs6000/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/rs6000-c.c (altivec_resolve_overloaded_builtin):
12236 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12237 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12238 (rs6000_preferred_simd_mode): Likewise.
12239 (output_vec_const_move): Likewise.
12240 (rs6000_expand_vector_extract): Likewise.
12241 (rs6000_split_vec_extract_var): Likewise.
12242 (reg_offset_addressing_ok_p): Likewise.
12243 (rs6000_legitimate_offset_address_p): Likewise.
12244 (rs6000_legitimize_address): Likewise.
12245 (rs6000_emit_set_const): Likewise.
12246 (rs6000_const_vec): Likewise.
12247 (rs6000_emit_move): Likewise.
12248 (rs6000_darwin64_record_arg_recurse): Likewise.
12249 (swap_selector_for_mode): Likewise.
12250 (paired_init_builtins): Likewise.
12251 (altivec_init_builtins): Likewise.
12252 (rs6000_expand_float128_convert): Likewise.
12253 (emit_load_locked): Likewise.
12254 (emit_store_conditional): Likewise.
12255 (rs6000_output_function_epilogue): Likewise.
12256 (rs6000_handle_altivec_attribute): Likewise.
12257 (rs6000_function_value): Likewise.
12258 (emit_fusion_gpr_load): Likewise.
12259 (emit_fusion_p9_load): Likewise.
12260 (emit_fusion_p9_store): Likewise.
12261 * config/rx/rx.c (rx_gen_move_template): Likewise.
12262 (flags_from_mode): Likewise.
12263 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12264 (s390_slb_comparison): Likewise.
12265 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12266 (s390_vector_mode_supported_p): Likewise.
12267 (s390_cc_modes_compatible): Likewise.
12268 (s390_match_ccmode_set): Likewise.
12269 (s390_canonicalize_comparison): Likewise.
12270 (s390_emit_compare_and_swap): Likewise.
12271 (s390_branch_condition_mask): Likewise.
12272 (s390_rtx_costs): Likewise.
12273 (s390_secondary_reload): Likewise.
12274 (__SECONDARY_RELOAD_CASE): Likewise.
12275 (s390_expand_cs): Likewise.
12276 (s390_preferred_simd_mode): Likewise.
12277 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12278 * config/sh/sh.c (sh_print_operand): Likewise.
12279 (dump_table): Likewise.
12280 (sh_secondary_reload): Likewise.
12281 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12282 * config/sh/sh.md (casesi_worker_1): Likewise.
12283 (casesi_worker_2): Likewise.
12284 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12285 (fcc_comparison_operator): Likewise.
12286 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12287 (emit_soft_tfmode_cvt): Likewise.
12288 (sparc_preferred_simd_mode): Likewise.
12289 (output_cbranch): Likewise.
12290 (sparc_print_operand): Likewise.
12291 (sparc_expand_vec_perm_bmask): Likewise.
12292 (vector_init_bshuffle): Likewise.
12293 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12294 (spu_vector_mode_supported_p): Likewise.
12295 (spu_expand_insv): Likewise.
12296 (spu_emit_branch_or_set): Likewise.
12297 (spu_handle_vector_attribute): Likewise.
12298 (spu_builtin_splats): Likewise.
12299 (spu_builtin_extract): Likewise.
12300 (spu_builtin_promote): Likewise.
12301 (spu_expand_sign_extend): Likewise.
12302 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12303 (tilegx_simd_int): Likewise.
12304 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12305 (tilepro_simd_int): Likewise.
12306 * config/v850/v850.c (const_double_split): Likewise.
12307 (v850_print_operand): Likewise.
12308 (ep_memory_offset): Likewise.
12309 * config/vax/vax.c (vax_rtx_costs): Likewise.
12310 (vax_output_int_move): Likewise.
12311 (vax_output_int_add): Likewise.
12312 (vax_output_int_subtract): Likewise.
12313 * config/visium/predicates.md (visium_branch_operator): Likewise.
12314 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12315 (visium_print_operand_address): Likewise.
12316 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12317 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12318 (xtensa_expand_conditional_branch): Likewise.
12319 (xtensa_copy_incoming_a7): Likewise.
12320 (xtensa_output_literal): Likewise.
12321 * dfp.c (decimal_real_maxval): Likewise.
12322 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12325 Alan Hayward <alan.hayward@arm.com>
12326 David Sherwood <david.sherwood@arm.com>
12328 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12329 (mode_nunits_inline): Likewise.
12330 (mode_inner_inline): Likewise.
12331 (mode_unit_size_inline): Likewise.
12332 (mode_unit_precision_inline): Likewise.
12333 (emit_insn_modes_h): Likewise. Also emit a #define of the
12335 (emit_mode_wider): Add an E_ prefix to mode names.
12336 (emit_mode_complex): Likewise.
12337 (emit_mode_inner): Likewise.
12338 (emit_mode_adjustments): Likewise.
12339 (emit_mode_int_n): Likewise.
12340 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12341 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12342 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12343 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12344 (CRC32_BUILTIN, ENTRY): Likewise.
12345 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12346 (aarch64_pop_regs): Likewise.
12347 (aarch64_process_components): Likewise.
12348 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12349 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12350 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12351 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12352 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12353 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12355 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12356 (pcmpistr): Likewise.
12357 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12358 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12359 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12361 * config/rl78/rl78.c (mduc_regs): Likewise.
12362 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12363 (htm_expand_builtin): Likewise.
12364 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12365 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12366 * config/xtensa/xtensa.c (print_operand): Likewise.
12367 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12368 (NUM_MODE_VECTOR_INT): Likewise.
12369 * genoutput.c (null_operand): Likewise.
12370 (output_operand_data): Likewise.
12371 * genrecog.c (print_parameter_value): Likewise.
12372 * lra.c (debug_operand_data): Likewise.
12374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12376 * dwarf2out.c (scompare_loc_descriptor_wide)
12377 (scompare_loc_descriptor_narrow): New functions, split out from...
12378 (scompare_loc_descriptor): ...here.
12379 * expmed.c (emit_store_flag_int): New function, split out from...
12380 (emit_store_flag): ...here.
12382 2017-08-30 Richard Biener <rguenther@suse.de>
12384 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12385 (dwarf2out_early_finish): Move setting of AT_pubnames from
12386 early debug output to early finish.
12388 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12390 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12391 and -mdata-region to the assembler.
12393 2017-08-30 Richard Biener <rguenther@suse.de>
12395 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12397 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12398 (dwarf2out_early_global_decl): Only generate a DIE for the
12399 abstract origin if it doesn't already exist or is a declaration DIE.
12400 (resolve_addr): Do not add the linkage name twice when
12401 generating a stub DIE for the DW_TAG_GNU_call_site target.
12403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12405 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12406 Use machine_mode rather than int for arg1_mode.
12408 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12411 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12412 that the second argument of the built-in functions to unpack
12413 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12414 switch statement instead a lot of if statements.
12415 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12416 Allow 64-bit values to be in Altivec registers as well as
12417 traditional floating point registers.
12418 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12420 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12422 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12423 MAX_REGS_PER_ADDRESS == 1.
12425 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12427 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12428 * config/i386/i386.c (ix86_option_override_internal): Simplify
12429 setting of opts->x_flag_entry.
12431 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12432 Jakub Jelinek <jakub@redhat.com>
12433 Richard Biener <rguenther@suse.de>
12435 PR tree-optimization/81503
12436 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12437 folded constant fits in the target type; reorder tests for clarity.
12439 2017-08-29 Martin Liska <mliska@suse.cz>
12441 * passes.def: Include pass_lower_switch.
12442 * stmt.c (dump_case_nodes): Remove and move to
12443 tree-switch-conversion.
12444 (case_values_threshold): Likewise.
12445 (expand_switch_as_decision_tree_p): Likewise.
12446 (emit_case_decision_tree): Likewise.
12447 (expand_case): Likewise.
12448 (balance_case_nodes): Likewise.
12449 (node_has_low_bound): Likewise.
12450 (node_has_high_bound): Likewise.
12451 (node_is_bounded): Likewise.
12452 (emit_case_nodes): Likewise.
12453 (struct simple_case_node): New struct.
12454 (add_case_node): Remove.
12455 (emit_case_dispatch_table): Use vector instead of case_list.
12456 (reset_out_edges_aux): Remove.
12457 (compute_cases_per_edge): Likewise.
12458 (expand_case): Build list of simple_case_node.
12459 (expand_sjlj_dispatch_table): Use it.
12460 * tree-switch-conversion.c (struct case_node): Moved from
12461 stmt.c and adjusted.
12462 (emit_case_nodes): Likewise.
12463 (node_has_low_bound): Likewise.
12464 (node_has_high_bound): Likewise.
12465 (node_is_bounded): Likewise.
12466 (case_values_threshold): Likewise.
12467 (reset_out_edges_aux): Likewise.
12468 (compute_cases_per_edge): Likewise.
12469 (add_case_node): Likewise.
12470 (dump_case_nodes): Likewise.
12471 (balance_case_nodes): Likewise.
12472 (expand_switch_as_decision_tree_p): Likewise.
12473 (emit_jump): Likewise.
12474 (emit_case_decision_tree): Likewise.
12475 (try_switch_expansion): Likewise.
12476 (do_jump_if_equal): Likewise.
12477 (emit_cmp_and_jump_insns): Likewise.
12478 (fix_phi_operands_for_edge): New function.
12479 (record_phi_operand_mapping): Likewise.
12480 (class pass_lower_switch): New pass.
12481 (pass_lower_switch::execute): New function.
12482 (make_pass_lower_switch): Likewise.
12483 (conditional_probability):
12484 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12485 * tree-pass.h: Add make_pass_lower_switch.
12487 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12490 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12493 2017-08-29 Richard Biener <rguenther@suse.de>
12495 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12496 we do not add a DW_AT_inline attribute twice.
12497 (gen_subprogram_die): Remove code setting DW_AT_inline on
12498 DECL_ABSTRACT_P nodes.
12500 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12502 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12503 calls to internal functions.
12504 (gimplify_modify_expr): Likewise.
12505 * tree-call-cdce.c (use_internal_fn): Likewise.
12506 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12507 (convert_to_divmod): Set the nothrow flag.
12508 * tree-if-conv.c (predicate_mem_writes): Likewise.
12509 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12510 (vectorizable_call): Likewise.
12511 (vectorizable_store): Likewise.
12512 (vectorizable_load): Likewise.
12513 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12514 (vect_recog_mask_conversion_pattern): Likewise.
12516 2017-08-29 Martin Liska <mliska@suse.cz>
12519 * gcc.c (driver_handle_option): Add new argument.
12520 * opts-common.c (handle_option): Pass
12521 target_option_override_hook.
12522 * opts-global.c (lang_handle_option): Add new option.
12523 (set_default_handlers): Add new argument.
12524 (decode_options): Likewise.
12525 * opts.c (target_handle_option): Likewise.
12526 (common_handle_option): Call target_option_override_hook.
12527 * opts.h (struct cl_option_handler_func): Add hook for
12528 target option override.
12529 (struct cl_option_handlers): Likewise.
12530 (set_default_handlers): Add new argument.
12531 (decode_options): Likewise.
12532 (common_handle_option): Likewise.
12533 (target_handle_option): Likewise.
12534 * toplev.c (toplev::main): Pass targetm.target_option.override
12537 2017-08-29 Richard Biener <rguenther@suse.de>
12538 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12540 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12541 life to the active subtree.
12543 2017-08-28 Jeff Law <law@redhat.com>
12545 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12546 derive_equivalences.
12547 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12548 Code moved into....
12549 (edge_info::derive_equivalences): New private member function
12551 * tree-ssa-dom.c (class edge_info): Changed from a struct
12552 to a class. Add ctor/dtor, methods and data members.
12553 (edge_info::edge_info): Renamed from allocate_edge_info.
12554 Initialize additional members.
12555 (edge_info::~edge_info): New.
12556 (free_dom_edge_info): Delete the edge info.
12557 (record_edge_info): Use new class & associated member functions.
12558 Tighten forms for testing for edge equivalences.
12559 (record_temporary_equivalences): Iterate over the simple
12560 equivalences rather than assuming there's only one per edge.
12561 (cprop_into_successor_phis): Iterate over the simple
12562 equivalences rather than assuming there's only one per edge.
12563 (optimize_stmt): Use operand_equal_p rather than pointer
12564 equality for mini-DSE code.
12566 2017-08-28 Nathan Sidwell <nathan@acm.org>
12568 * gcc.c (execute): Fold SIGPIPE handling into switch
12569 statement. Adjust internal error message.
12571 2017-08-28 Richard Biener <rguenther@suse.de>
12574 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12575 Do nothing for removed DIEs.
12577 2017-08-28 Richard Biener <rguenther@suse.de>
12579 PR tree-optimization/81977
12580 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12583 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12586 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12587 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12588 using targetm.gen_mem_thread_fence.
12590 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12593 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12594 predicate to register_operand. Reorder operands.
12595 (*btr<mode>): Ditto.
12596 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12597 (*btr<mode>_mask): Ditto.
12599 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12601 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12602 * config/rs6000/xmmintrin.h: New file.
12603 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12605 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12608 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12609 parameter and_insn and return it.
12610 (recombine_lvx_pattern): Insert a copy to ensure availability of
12611 the base register of the copied masking operation at the point of
12612 the instruction replacement.
12613 (recombine_stvx_pattern): Likewise.
12615 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12617 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12618 undocumented switches.
12619 (-mpower9-dform-vector): Likewise.
12620 (-mpower9-dform): Likewise.
12621 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12622 comments to delete references to -mpower9-dform* switches.
12623 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12624 Delete reference to -mpower9-dform* switches, test for
12625 -mpower9-vector instead.
12626 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12627 (OTHER_P9_VECTOR_MASKS): Likewise.
12628 (POWERPC_MASKS): Likewise.
12629 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12630 tests against -mpower9-dform* to -mpower9-vector. Delete code
12631 that checked for -mpower9-dform* consistancy with other options.
12632 Add test for -mpower9-misc to enable other power9 switches.
12633 (rs6000_init_hard_regno_mode_ok): Likewise.
12634 (rs6000_option_override_internal): Likewise.
12635 (rs6000_emit_prologue): Likewise.
12636 (rs6000_emit_epilogue): Likewise.
12637 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12638 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12639 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12641 (emit_fusion_p9_store): Likewise.
12642 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12643 resetting these macros if the assembler does not support ISA 3.0
12645 (TARGET_P9_DFORM_VECTOR): Likewise.
12646 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12647 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12649 2017-08-25 Alan Modra <amodra@gmail.com>
12651 PR rtl-optimization/81747
12652 * cse.c (cse_extended_basic_block): Don't attempt to record
12653 equivalences for degenerate conditional branches that branch
12654 to their fall-through.
12656 2017-08-24 Martin Sebor <msebor@redhat.com>
12658 PR middle-end/81908
12659 * gimple-fold.c (size_must_be_zero_p): New function.
12660 (gimple_fold_builtin_memory_op): Call it.
12662 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12664 * config/rs6000/mm_malloc.h: New file.
12666 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12668 PR tree-optimization/81913
12669 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12670 analysis when either IVs in condition can wrap.
12672 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12674 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12675 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12677 2017-08-24 Richard Biener <rguenther@suse.de>
12680 * targhooks.c (default_target_can_inline_p): Properly
12681 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12682 is present and always compare.
12683 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12684 infer -mfpmath=sse from TARGET_SSE_P.
12685 (ix86_can_inline_p): Properly use target_option_default_node when
12686 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12688 2017-08-24 Richard Biener <rguenther@suse.de>
12691 * dwarf2out.c (output_die): Handle flag_generate_offload like
12693 (output_comp_unit): Likewise.
12694 (gen_array_type_die): Likewise.
12695 (dwarf2out_early_finish): Likewise.
12696 (note_variable_value_in_expr): Likewise.
12697 (dwarf2out_finish): Likewise. Adjust assert.
12698 * cgraphunit.c (symbol_table::compile): Move setting of
12699 flag_generate_offload earlier ...
12700 (symbol_table::finalize_compilation_unit): ... here, before
12701 early debug finalization.
12703 2017-08-24 Richard Biener <rguenther@suse.de>
12705 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12706 and ipa-fnsummary.h.
12707 (ix86_can_inline_p): When ix86_fpmath flags do not match
12708 check whether the callee uses FP math at all.
12710 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12712 PR middle-end/81931
12713 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12714 instead of TYPE_PRECISION.
12716 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12717 Alan Hayward <alan.hayward@arm.com>
12718 David Sherwood <david.sherwood@arm.com>
12720 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12722 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12724 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12726 (ix86_frame::stack_realign_allocate): New field.
12727 (struct machine_frame_state): Modify comments.
12728 (machine_frame_state::sp_realigned_fp_end): New field.
12729 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12730 layout calculation.
12731 (sp_valid_at): Add assertion to assure no attempt to access invalid
12732 offset of a realigned stack.
12733 (fp_valid_at): Likewise.
12734 (choose_baseaddr): Modify comments.
12735 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12736 ix86_expand_prologue.
12737 (ix86_expand_prologue): Modify stack realignment and allocation.
12738 (ix86_expand_epilogue): Modify comments.
12739 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12740 avx2_runtime, avx512f, and avx512f_runtime.
12742 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12744 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12745 (mstackrealign): Do not init to -1.
12746 * config/i386/i386.c (ix86_option_override_internal):
12747 Check opts_set, not opts when setting default value of
12748 opts->x_ix86_force_align_arg_pointer.
12750 2017-08-23 Richard Biener <rguenther@suse.de>
12752 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12753 lang_hooks.decl_printable_name.
12754 * print-rtl-function.c (print_rtx_function): Likewise.
12755 * tree-pretty-print.c (dump_function_header): Likewise.
12757 2017-08-23 Richard Biener <rguenther@suse.de>
12760 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12761 -g0 at compile-time.
12763 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12765 PR middle-end/19706
12766 * doc/sourcebuild.texi (Other hardware attributes):
12769 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12771 PR middle-end/19706
12772 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12773 Fix single-use check.
12775 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12777 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12779 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12781 * doc/install.texi: Modify to add more details on running selected
12784 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12786 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12787 is combined with -mabi=ms.
12788 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12791 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12793 PR tree-optimization/81488
12794 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12795 and cached_basis fields.
12796 (MAX_SPREAD): New constant.
12797 (alloc_cand_and_find_basis): Initialize new fields.
12798 (clear_visited): New function.
12799 (create_phi_basis_1): Rename from create_phi_basis, set visited
12800 and cached_basis fields.
12801 (create_phi_basis): New wrapper function.
12802 (phi_add_costs_1): Rename from phi_add_costs, add spread
12803 parameter, set visited field, short-circuit when limits reached.
12804 (phi_add_costs): New wrapper function.
12805 (record_phi_increments_1): Rename from record_phi_increments, set
12807 (record_phi_increments): New wrapper function.
12808 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12809 (phi_incr_cost): New wrapper function.
12810 (all_phi_incrs_profitable_1): Rename from
12811 all_phi_incrs_profitable, set visited field.
12812 (all_phi_incrs_profitable): New wrapper function.
12814 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12815 Alan Hayward <alan.hayward@arm.com>
12816 David Sherwood <david.sherwood@arm.com>
12818 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12819 that takes the outer and inner modes.
12820 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12821 comparison as the canonical test for a paradoxical subreg.
12822 * combine.c (simplify_set): Use paradoxical_subreg_p.
12823 (make_extraction): Likewise.
12824 (force_to_mode): Likewise.
12825 (rtx_equal_for_field_assignment_p): Likewise.
12826 (gen_lowpart_for_combine): Likewise.
12827 (simplify_comparison): Likewise.
12828 * cse.c (equiv_constant): Likewise.
12829 * expmed.c (store_bit_field_1): Likewise.
12830 * final.c (alter_subreg): Likewise.
12831 * fwprop.c (propagate_rtx): Likewise.
12832 (forward_propagate_subreg): Likewise.
12833 * ira-conflicts.c (ira_build_conflicts): Likewise.
12834 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12835 * lra-constraints.c (curr_insn_transform): Likewise.
12836 (split_reg): Likewise.
12837 * lra-eliminations.c (move_plus_up): Likewise.
12838 (lra_eliminate_regs_1): Likewise.
12839 * recog.c (general_operand): Likewise.
12840 * ree.c (combine_reaching_defs): Likewise.
12841 * reload.c (push_reload): Likewise.
12842 (find_reloads): Likewise.
12843 * reload1.c (elimination_effects): Likewise.
12844 (compute_reload_subreg_offset): Likewise.
12845 (choose_reload_regs): Likewise.
12846 * rtlanal.c (subreg_lsb_1): Likewise.
12847 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12848 (simplify_subreg): Likewise.
12849 * var-tracking.c (track_loc_p): Likewise.
12850 * emit-rtl.c (byte_lowpart_offset): Likewise.
12851 (paradoxical_subreg_p): Delete out-of-line definition.
12853 2017-08-22 Jeff Law <law@redhat.com>
12855 PR tree-optimization/81741
12856 PR tree-optimization/71947
12857 * tree-ssa-dom.c: Include tree-inline.h.
12858 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12859 equivalences if one is more expensive to compute than the other.
12860 * tree-ssa-scopedtables.h (class const_or_copies): Make
12861 record_const_or_copy_raw method private.
12862 (class avail_exprs_stack): New method simplify_binary_operation.
12863 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12864 avail_exprs_stack::simplify_binary_operation as needed.
12865 (avail_exprs_stack::simplify_binary_operation): New function.
12867 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12869 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12870 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12871 (DOT_SYMBOLS): Likewise.
12872 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12873 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12874 (RS6000_ABI_NAME): Likewise.
12875 (TARGET_CMODEL): Likewise.
12876 (TOC_SECTION_ASM_OP): Likewise.
12877 (SET_CMODEL): New macro.
12878 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12880 2017-08-22 Richard Biener <rguenther@suse.de>
12882 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12883 to allow for free-lang-data replacements similar to verify_type_variant.
12885 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12886 Alan Hayward <alan.hayward@arm.com>
12887 David Sherwood <david.sherwood@arm.com>
12889 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12890 VOIDmode for the LABEL_REF.
12892 2017-08-22 Richard Biener <rguenther@suse.de>
12894 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12896 2017-08-22 Richard Biener <rguenther@suse.de>
12898 * common.opt (feliminate-dwarf2-dups): Ignore.
12899 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12900 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12901 same_die_p_wrap, compute_section_prefix,
12902 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12903 (comdat_symbol_id, comdat_symbol_number): Likewise.
12904 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12906 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12907 (output_die): Mark unreachable path unreachable.
12908 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12909 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12910 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12911 (dwarf2out_early_finish): Likewise.
12913 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12915 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12917 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12920 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12921 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12922 (avr_attribute_table) <io, io_low, address>: Initialize
12923 .decl_required with true.
12925 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12927 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12928 undocumented debugging options.
12929 (-mvsx-scalar-double): Likewise.
12930 (-mallow-df-permute): Likewise.
12931 (-mvectorize-builtins): Likewise.
12932 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12933 (rs6000_builtin_vectorized_function): Likewise.
12934 (rs6000_builtin_md_vectorized_function): Likewise.
12935 (rs6000_opt_vars): Likewise.
12937 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12940 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12941 (*btrq_imm): Rename from *btrq.
12942 (*btcq_imm): Rename from *btcq.
12943 (btsc): New code attribute.
12944 (*<btsc><mode>): New insn pattern.
12945 (*btr<mode>): Ditto.
12946 (*<btsc><mode>_mask): New insn_and_split pattern.
12947 (*btr<mode>_mask): Ditto.
12949 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12950 Alan Hayward <alan.hayward@arm.com>
12951 David Sherwood <david.sherwood@arm.com>
12953 * function.c (pad_below): Simplify padding calculation.
12955 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12956 Alan Hayward <alan.hayward@arm.com>
12957 David Sherwood <david.sherwood@arm.com>
12959 * target.def (function_prologue): Remove frame size argument.
12960 (function_epilogue): Likewise.
12961 * doc/tm.texi: Regenerate.
12962 * final.c (final_start_function): Update call to function_prologue.
12963 (final_end_function): Update call to function_epilogue.
12964 (default_function_pro_epilogue): Remove frame size argument.
12965 * output.h (default_function_pro_epilogue): Likewise.
12966 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12967 (arm_output_function_prologue): Likewise.
12968 * config/frv/frv.c (frv_function_prologue): Likewise.
12969 (frv_function_epilogue): Likewise.
12970 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12971 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12972 (ia64_output_function_epilogue): Likewise.
12973 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12974 (m32r_output_function_epilogue): Likewise.
12975 * config/microblaze/microblaze.c (microblaze_function_prologue)
12976 (microblaze_function_epilogue): Likewise.
12977 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12978 (mips_output_function_epilogue): Likewise.
12979 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12980 (mmix_target_asm_function_epilogue): Likewise.
12981 * config/msp430/msp430.c (msp430_start_function): Likewise.
12982 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12983 (nds32_asm_function_epilogue): Likewise.
12984 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12985 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12986 (pa_output_function_epilogue): Likewise.
12987 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12988 (rs6000_output_function_epilogue): Likewise.
12989 * config/rl78/rl78.c (rl78_start_function): Likewise.
12990 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12991 (rs6000_output_function_epilogue): Likewise.
12992 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12993 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12994 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12995 (sparc_asm_function_epilogue): Likewise.
12997 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12999 * tree.h (type_has_mode_precision_p): New function.
13000 * convert.c (convert_to_integer_1): Use it.
13001 * expr.c (expand_expr_real_2): Likewise.
13002 (expand_expr_real_1): Likewise.
13003 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13004 * match.pd: Likewise.
13005 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13006 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13007 * tree-tailcall.c (process_assignment): Likewise.
13008 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13009 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13010 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13011 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13012 (vectorizable_assignment): Likewise.
13013 (vectorizable_shift): Likewise.
13014 (vectorizable_operation): Likewise.
13015 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13017 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13019 * match.pd: Add pow (C, x) simplification.
13021 2017-08-21 Richard Biener <rguenther@suse.de>
13023 PR tree-optimization/81900
13024 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13025 for blocks with abnormal predecessors.
13026 (compute_antic): Do not set visited flag prematurely.
13028 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13031 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13033 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13035 * stor-layout.h (vector_type_mode): Move to...
13036 * tree.h (vector_type_mode): ...here.
13037 * stor-layout.c (vector_type_mode): Move to...
13038 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13040 2017-08-21 Richard Biener <rguenther@suse.de>
13042 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13043 register_external_die hooks.
13044 (debug_false_tree_charstarstar_uhwistar): Declare.
13045 (debug_nothing_tree_charstar_uhwi): Likewise.
13046 * debug.c (do_nothing_debug_hooks): Adjust.
13047 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13048 (debug_nothing_tree_charstar_uhwi): Likewise.
13049 * dbxout.c (dbx_debug_hooks): Adjust.
13050 (xcoff_debug_hooks): Likewise.
13051 * sdbout.c (sdb_debug_hooks): Likewise.
13052 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13053 * dwarf2out.c (macinfo_label_base): New global.
13054 (dwarf2out_register_external_die): New function for the
13055 register_external_die hook.
13056 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13057 (dwarf2_debug_hooks): Use them.
13058 (dwarf2_lineno_debug_hooks): Adjust.
13059 (struct die_struct): Add with_offset flag.
13060 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13061 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13062 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13063 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13064 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13065 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13066 defining section names for the early LTO debug variants.
13067 (reset_indirect_string): New helper.
13068 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13069 (print_dw_val): Add support for offsetted symbol references.
13070 (get_ultimate_context): Split out from is_cxx.
13071 (is_cxx): Use get_ultimate_context.
13072 (is_fortran): Add decl overload.
13073 (compute_comp_unit_symbol): Split out worker from
13074 compute_section_prefix.
13075 (compute_section_prefix): Call compute_comp_unit_symbol and
13076 set comdat_type_p here.
13077 (output_die): Skip DIE symbol output for the LTO added one.
13078 Handle DIE symbol references with offset.
13079 (output_comp_unit): Guard section name mangling properly.
13080 For LTO debug sections emit a symbol at the section beginning
13081 which we use to refer to its DIEs.
13082 (add_abstract_origin_attribute): For DIEs registered via
13083 dwarf2out_register_external_die directly refer to the early
13084 DIE rather than indirectly through the shadow one we created.
13085 Remove obsolete call to dwarf2out_abstract_function for
13086 non-function/block origins.
13087 (gen_array_type_die): When generating early LTO debug do
13088 not emit DW_AT_string_length.
13089 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13090 late when in LTO. As suggested place a gcc_unreachable for
13091 the DECL_ABSTRACT_P case.
13092 (gen_subprogram_die): Avoid another specification DIE
13093 for early built declarations/definitions for the late LTO case.
13094 (gen_variable_die): Add type references for late duplicated VLA dies
13096 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13097 we have the abstract instance already.
13098 (process_scope_var): Adjust decl DIE contexts in LTO which
13099 first puts them in limbo.
13100 (gen_decl_die): Do not generate type DIEs late apart from
13101 types for VLAs or for decls we do not yet have a DIE. Do not
13102 call dwarf2out_abstract_function late.
13103 (dwarf2out_early_global_decl): Make sure to create DIEs
13104 for abstract instances of a decl first.
13105 (dwarf2out_late_global_decl): Adjust comment.
13106 (output_macinfo_op): With multiple macro sections use
13107 macinfo_label_base to distinguish labels.
13108 (output_macinfo): Likewise. Update macinfo_label_base.
13109 Pass in the line info label.
13110 (note_variable_value_in_expr): When generating LTO resolve
13111 all variable values here by generating DIEs as needed.
13112 (init_sections_and_labels): Add early LTO debug flag parameter
13113 and generate different sections and names if set. Add generation
13114 counter for the labels so we can have multiple of them.
13115 (reset_dies): Helper to allow DIEs to be output multiple times.
13116 (dwarf2out_finish): When outputting DIEs to the fat part of an
13117 LTO object first reset DIEs.
13118 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13119 (modified_type_die): Check for decl_ultimate_origin being self
13121 (gen_type_die_with_usage): Likewise.
13122 (gen_typedef_die): Allow decl_ultimate_origin being self.
13123 (set_decl_abstract_flags): Remove.
13124 (set_block_abstract_flags): Likewise.
13125 (dwarf2out_abstract_function): Treat the early generated DIEs
13126 as the abstract copy and only add DW_AT_inline and
13127 DW_AT_artificial here and call set_decl_origin_self.
13128 If the DIE has an abstract origin don't do anything.
13129 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13130 if we have none yet (Go fails to build one, PR78628).
13131 (variably_modified_type_p): Prevent endless recursion for Ada
13132 cyclic pointer types.
13133 * lto-streamer-in.c: Include debug.h.
13134 (dref_queue): New global.
13135 (lto_read_tree_1): Stream in DIE references.
13136 (lto_input_tree): Register DIE references.
13137 (input_function): Stream DECL_DEBUG_ARGS.
13138 * lto-streamer-out.c: Include debug.h.
13139 (lto_write_tree_1): Output DIE references.
13140 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13141 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13142 (output_function): Stream DECL_DEBUG_ARGS.
13143 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13144 Stream DECL_ABSTRACT_ORIGIN.
13145 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13146 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13147 DECL_CONTEXT for file-scope decls.
13148 * lto-streamer.h (struct dref_entry): Declare.
13149 (dref_queue): Likewise.
13150 * cfgexpand.c (pass_expand::execute): Do not call the
13151 outlining_inline_function hook here.
13152 * lto-wrapper.c (debug_obj): New global.
13153 (tool_cleanup): Unlink it if required.
13154 (debug_objcopy): New function.
13155 (run_gcc): Handle early debug sections in the IL files by
13156 extracting them to separate files, partially linkin them and
13157 feeding the result back as result to the linker.
13158 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13159 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13160 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13161 sections into a separate segment.
13162 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13164 (darwin_asm_dwarf_section): Likewise.
13165 (darwin_asm_output_dwarf_offset): Likewise.
13166 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13168 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13169 Alan Hayward <alan.hayward@arm.com>
13170 David Sherwood <david.sherwood@arm.com>
13172 * read-md.h (md_reader::record_potential_iterator_use): Replace
13173 pointer argument with an rtx and an index.
13174 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13175 (apply_mode_iterator): Likewise.
13176 (apply_code_iterator): Likewise.
13177 (apply_int_iterator): Likewise.
13178 (apply_subst_iterator): Likewise.
13179 (record_iterator_use): Likewise.
13180 (record_attribute_use): Likewise.
13181 (md_reader::record_potential_iterator_use): Likewise. Update calls
13182 to record_iterator_use and apply_iterator.
13183 (iterator_use): Replace ptr with x and index.
13184 (attribute_use): Likewise.
13185 (apply_attribute_uses): Update calls to apply_iterator.
13186 (apply_iterators): Likewise. Update initialization of iterator_use.
13187 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13188 and record_potential_iterator_use.
13189 (rtx_reader::read_rtx_operand): Likewise.
13191 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13192 Alan Hayward <alan.hayward@arm.com>
13193 David Sherwood <david.sherwood@arm.com>
13195 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13198 2017-08-21 Richard Biener <rguenther@suse.de>
13200 PR middle-end/81884
13201 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13202 at struct end conservatively when comparing common bases.
13204 2017-08-21 Richard Biener <rguenther@suse.de>
13206 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13207 (mem_ref_in_stmt): Remove.
13208 (determine_max_movement): Use ref index to get at the reference.
13209 (invariantness_dom_walker::before_dom_children): Deal with
13210 lim data already initialized.
13211 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13213 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13215 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13216 (TARGET_ISA_ROUND): Ditto.
13217 (TARGET_ROUND): Ditto.
13218 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13219 * config/i386/i386.md: Ditto.
13220 * config/i386/sse.md: Ditto.
13221 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13222 with OPTION_MASK_ISA_SSE4_1.
13224 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13227 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13228 __builtin_ia32_lzcnt_u16.
13230 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13233 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13234 (rs6000_set_current_function): Rewrite function to use it.
13236 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13239 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13240 and TYPE_WARN_IF_NOT_ALIGN.
13241 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13242 (handle_warn_if_not_align): New.
13243 (place_union_field): Call handle_warn_if_not_align.
13244 (place_field): Call handle_warn_if_not_align.
13245 Copy TYPE_WARN_IF_NOT_ALIGN.
13246 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13247 (layout_type): Likewise.
13248 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13250 (tree_decl_common): Add warn_if_not_align.
13251 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13252 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13253 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13254 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13255 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13256 * doc/extend.texi: Document warn_if_not_aligned attribute.
13257 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13259 2017-08-17 Martin Liska <mliska@suse.cz>
13262 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13263 (get_data_dependence): Use it as pointer type.
13264 (distribute_loop): Likewise.
13266 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13268 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13269 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13270 (p8_vmrgow_<mode>_direct): New define_insn.
13271 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13272 handle endianness for vmrgew and vmrgow permute patterns.
13274 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13276 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13277 * config/rs6000/rs6000-cpus.def: Remove comment.
13278 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13279 (POWERPC_MASKS): Likewise.
13280 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13281 use of TARGET_VSX_TIMODE.
13282 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13283 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13284 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13285 (rs6000_option_override_internal): Remove dead code.
13286 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13287 (rs6000_legitimize_reload_address): Likewise.
13288 (rs6000_legitimate_address_p): Likewise.
13289 (rs6000_opt_masks): Delete "vsx-timode".
13290 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13291 from function comment.
13292 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13293 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13294 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13296 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13297 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13298 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13299 (VSX_TI): Likewise.
13301 (define_peephole2): Likewise.
13303 2017-08-17 Martin Sebor <msebor@redhat.com>
13306 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13307 past the end of an array.
13308 (test_pp_format): Add test cases.
13310 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13312 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13313 missing ECF_NOTHROW flags.
13315 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13318 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13319 operands residing in integer registers.
13320 (*vsx_le_perm_load_<mode>): Likewise.
13321 (*vsx_le_perm_store_<mode>): Likewise.
13322 (define_peephole2): Add peepholes to optimize the above.
13324 2017-08-17 Marek Polacek <polacek@redhat.com>
13326 PR middle-end/81814
13327 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13328 to mimic what shorten_compare did. Change the return type to bool.
13329 (fold_cond_expr_with_comparison): Update call to
13330 operand_equal_for_comparison_p.
13331 (fold_ternary_loc): Likewise.
13333 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13335 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13337 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13339 2017-08-17 Richard Biener <rguenther@suse.de>
13341 * tree-ssa-structalias.c (solve_graph): When propagating
13342 to successors update the graphs succ edges and avoid duplicate work.
13344 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13347 * config/i386/i386.c (ix86_option_override_internal): Save target
13348 specific options after ix86_stack_protector_guard_reg was changed.
13350 2017-08-17 Richard Biener <rguenther@suse.de>
13352 PR tree-optimization/81827
13353 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13354 (new_var_info): Initialize it conservatively.
13355 (get_call_vi): Mark register vars.
13356 (new_scalar_tmp_constraint_exp): Likewise.
13357 (handle_rhs_call): Likewise.
13358 (handle_const_call): Likewise.
13359 (create_function_info_for): Likewise.
13360 (solve_constraints): Sort varinfos to separate register from
13361 non-register vars to pack points-to solution bitmaps during
13364 2017-08-17 Marek Polacek <polacek@redhat.com>
13366 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13368 2017-08-17 Richard Biener <rguenther@suse.de>
13370 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13371 to true when overflow is undefined and we saturated the result.
13373 2017-08-17 Alan Modra <amodra@gmail.com>
13376 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13377 Don't use store multiple if only one reg needs saving.
13378 (interesting_frame_related_regno): New function.
13379 (rs6000_frame_related): Don't emit frame info for regs that
13381 (rs6000_emit_epilogue): Likewise.
13383 2017-08-16 Nathan Sidwell <nathan@acm.org>
13385 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13386 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13387 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13388 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13389 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13390 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13391 (verify_type): Adjust for TYPE_BINFO move.
13392 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13393 process TYPE_BINFO directly.
13394 (hash_tree): Likewise.
13395 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13397 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13400 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13402 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13404 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13407 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13408 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13409 Add ix86_binary_operator_ok to insn constraint.
13410 (*iordi_1_bts): Ditto.
13411 (*xordi_1_btc): Ditto.
13412 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13413 Update corresponding peephole2 pattern.
13417 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13419 PR tree-optimization/81832
13420 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13421 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13423 2017-08-16 Marek Polacek <polacek@redhat.com>
13426 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13427 regarding size_zero_node.
13429 2017-08-16 Martin Liska <mliska@suse.cz>
13432 * config.gcc: Respect previously set extra_objs in case
13435 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13437 PR tree-optimization/81835
13438 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13439 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13440 not depend on the phi.
13442 2017-08-16 Alan Modra <amodra@gmail.com>
13444 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13447 2017-08-16 Alan Modra <amodra@gmail.com>
13449 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13450 (save_reg_p): ..into this. Update all callers.
13451 (first_reg_to_save): Simplify.
13453 2017-08-16 Alan Modra <amodra@gmail.com>
13455 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13458 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13462 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13463 constant count if that count is less than 32.
13465 2017-08-15 Nathan Sidwell <nathan@acm.org>
13467 * gcc.c (execute): Emit friendlier message if inferior is killed
13468 by an external cause.
13470 2017-08-15 Richard Biener <rguenther@suse.de>
13472 PR tree-optimization/81790
13473 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13474 CONSTRUCTORs from simplifying and VN.
13476 2017-08-14 Martin Sebor <msebor@redhat.com>
13478 * builtin-attrs.def: Add comments.
13480 2017-08-14 Martin Sebor <msebor@redhat.com>
13483 * doc/extend.texi (attribute nonstring): Document new attribute.
13485 2017-08-14 Martin Sebor <msebor@redhat.com>
13488 * tree-diagnostic.c (default_tree_printer): Handle %G.
13489 * gimple-pretty-print.h (percent_G_format): Declare new function.
13490 * gimple-pretty-print.c (percent_G_format): Define.
13491 * tree-pretty-print.c (percent_K_format): Add argument.
13493 2017-08-14 Martin Sebor <msebor@redhat.com>
13495 PR translation/79998
13496 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13497 Remove a stray space.
13499 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13502 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13503 (*iordi_1_bts): Ditto.
13504 (*xordi_1_btc): Ditto.
13506 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13509 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13510 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13512 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13513 quoted strings, and make more translator-friendly.
13514 (darwin_rs6000_override_options): Likewise.
13515 (rs6000_option_override_internal): Likewise.
13516 (rs6000_return_in_memory): Fix overlong line.
13517 (init_cmulative_args): Use quoted strings, and make more
13518 translator-friendly.
13519 (rs6000_pass_by_reference): Fix overlong line.
13520 (def_builtin): Use quoted strings.
13521 (altivec_expand_predicate_builtin): Use quoted strings, and make
13522 more translator-friendly.
13523 (htm_expand_builtin): Use quoted strings.
13524 (cpu_expand_builtin): Use quoted strings, and make more
13525 translator-friendly.
13526 (altivec_expand_builtin): Likewise.
13527 (paired_expand_predicate_builtin): Likewise.
13528 (rs6000_invalid_builtin): Likewise.
13529 (builtin_function_type): Use quoted strings.
13530 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13531 more translator-friendly.
13532 (rs6000_trampoline_init): Likewise.
13533 (rs6000_handle_altivec_attribute): Likewise.
13534 (rs6000_inner_target_options): Use quoted strings.
13535 (rs6000_disable_incompatible_switches): Likewise.
13536 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13537 strings, and make more translator-friendly.
13538 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13540 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13542 PR tree-optimization/81799
13543 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13544 cond_expr to simple gimple operand.
13546 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13548 PR middle-end/46932
13549 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13551 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13555 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13556 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13557 TARGET_GASISR_PROLOGUES.
13558 * config/avr/avr.c (avr_option_override): Same.
13559 (avr_pass_pre_proep::execute): Same.
13561 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13564 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13565 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13567 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13569 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13570 define_insn_and_split. Split to a memory load from 0 in
13571 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13572 using PTR mode iterator.
13573 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13574 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13575 (*add_tp_<mode>): Redefine as define_insn_and_split.
13576 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13577 address space. Merge with *add_tp_x32 using PTR mode iterator.
13578 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13579 Split to an add with a memory load from 0 in
13580 DEFAULT_TLS_SEG_REG address space.
13582 2017-08-12 Andrew Pinski <apinski@cavium.com>
13584 * config/aarch64/aarch64-option-extensions.def (rdma):
13585 Fix feature string to what Linux prints out in /proc/cpuinfo.
13587 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13590 * dwarf2out.c (modified_type_die): For C typedef types that have
13591 an ultimate origin, process the ultimate origin instead of the
13593 (gen_typedef_die): Assert that input DECLs have no ultimate
13595 (gen_type_die_with_usage): For typedef variants that have an
13596 ultimate origin, just call gen_decl_die on the original DECL.
13597 (process_scope_var): Avoid creating DIEs for local typedefs and
13598 concrete static variables.
13600 2017-08-12 Alan Modra <amodra@gmail.com>
13604 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13605 match gnu-user.h startfile.
13606 (ENDFILE_LINUX_SPEC): Similarly.
13608 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13611 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13613 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13615 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13616 * config/aarch64/aarch64.md (mov<mode>): Change.
13617 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13618 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13619 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13621 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13623 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13624 for storage order barriers.
13626 2017-08-11 Martin Liska <mliska@suse.cz>
13629 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13630 variables of void type.
13632 2017-08-11 Martin Liska <mliska@suse.cz>
13634 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13635 TARGET_SUPPORTS_ALIASES.
13636 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13637 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13638 (optimize_weakref): Likewise.
13639 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13640 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13641 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13643 2017-08-11 Martin Liska <mliska@suse.cz>
13646 * config/i386/i386.c (make_resolver_func): Do complete
13647 refactoring of the function.
13649 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13652 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13653 * config/i386/i386.c (ix86_stack_protect_guard): Use
13654 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13655 * doc/invoke.texi (x86 Options): Document
13656 -mstack-protector-guard-symbol= option.
13658 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13660 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13661 * config/i386/i386.c (ix86_split_stack_guard): New function.
13662 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13663 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13664 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13665 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13666 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13667 (split_stack_space_check): Call ix86_split_stack_guard.
13669 2017-08-10 Martin Sebor <msebor@redhat.com>
13671 * print-tree.c (print_node): Print location using the established
13673 Replace spaces with colons.
13674 (debug_raw, debug): Ditto.
13676 2017-08-10 Martin Sebor <msebor@redhat.com>
13679 * pretty-print.c (pp_format): Correct the handling of %s precision.
13681 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13684 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13686 (ix86_finalize_stack_frame_flags): This. Also clear
13687 frame_pointer_needed if -fno-omit-frame-pointer is used without
13689 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13690 with ix86_finalize_stack_frame_flags.
13691 (ix86_expand_epilogue): Likewise.
13692 (ix86_expand_split_stack_prologue): Likewise.
13693 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13695 2017-08-10 Martin Liska <mliska@suse.cz>
13698 * c-attribs.c (handle_target_attribute):
13699 Report warning for an empty string argument of target attribute.
13701 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13704 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13706 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13707 or DECL_NONLOCAL labels.
13708 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13709 or DECL_NONLOCAL labels here.
13711 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13713 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13714 to indicate when early gimple folding has been disabled.
13715 (rs6000_gimple_fold_builtin): Add debug content.
13716 (rs6000_invalid_builtin): Fix whitespace.
13717 (rs6000_expand_builtin): Fix whitespace.
13718 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13720 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13723 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13724 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13726 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13728 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13729 (qdf24xx): Likewise.
13730 * config/aarch64/aarch64.md: Include falkor.md.
13731 * config/aarch64/falkor.md: New.
13733 2017-08-09 Marek Polacek <polacek@redhat.com>
13736 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13737 * diagnostic.c (emit_diagnostic): Add a comment.
13738 (emit_diagnostic_valist): New function.
13740 2017-08-09 Marek Polacek <polacek@redhat.com>
13743 * input.c (make_location): New overload.
13744 * input.h (make_location): Declare.
13746 2017-08-08 Alan Modra <amodra@gmail.com>
13747 H.J. Lu <hongjiu.lu@intel.com>
13750 * gcc.c (NO_PIE_SPEC): Delete.
13751 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13753 (LINK_PIE_SPEC): ..to here.
13754 (LINK_COMMAND_SPEC): Support -no-pie.
13755 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13756 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13758 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13759 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13760 (ENDFILE_CRTEND_SPEC): Similarly.
13762 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13765 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13766 (mstack-protector-guard-offset=): Ditto.
13767 * config/i386/i386.c (ix86_option_override): Handle
13768 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13770 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13771 ix86_stack_protect_guard_offset variables.
13772 (TARGET_STACK_PROTECT_GUARD): Always define.
13773 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13774 and -mstack-protector-guard-offset= options.
13776 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13778 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13779 boundary case for the last candidate.
13781 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13783 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13784 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13786 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13788 PR middle-end/19706
13789 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13790 * config/aarch64/aarch64-builtins.c
13791 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13792 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13793 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13795 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13796 Andrew Pinski <pinskia@gmail.com>
13798 PR middle-end/19706
13799 * internal-fn.def (XORSIGN): New.
13800 * optabs.def (xorsign_optab): New.
13801 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13802 (convert_expand_mult_copysign): New.
13803 (pass_optimize_widening_mul::execute): Call
13804 convert_expand_mult_copysign.
13806 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13808 PR tree-optimization/81354
13809 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13810 Insert on edges rather than explicitly creating landing pads.
13811 (analyze_candidates_and_replace): Commit edge inserts.
13813 2017-08-08 Richard Biener <rguenther@suse.de>
13815 PR middle-end/81719
13816 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13817 (expand_simple_operations): Also look through ADDR_EXPRs with
13818 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13820 2017-08-08 Richard Biener <rguenther@suse.de>
13822 PR tree-optimization/81723
13823 * tree-vect-slp.c (struct bst_traits): New hash traits.
13824 (bst_fail): New global.
13825 (vect_build_slp_tree_2): New worker, split out from ...
13826 (vect_build_slp_tree): ... this now wrapping it with using
13827 bst_fail set to cache SLP tree build fails. Properly handle
13829 (vect_analyze_slp_instance): Allocate and free bst_fail.
13831 2017-08-08 Martin Liska <mliska@suse.cz>
13834 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13835 LABEL_DECLs that can be from a different function.
13837 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13839 PR tree-optimization/81744
13840 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13841 loop's number of iterations.
13843 2017-08-08 Martin Liska <mliska@suse.cz>
13845 * asan.c: Include header files.
13846 * attribs.c (build_decl_attribute_variant): New function moved
13848 (build_type_attribute_qual_variant): Likewise.
13849 (cmp_attrib_identifiers): Likewise.
13850 (simple_cst_list_equal): Likewise.
13851 (omp_declare_simd_clauses_equal): Likewise.
13852 (attribute_value_equal): Likewise.
13853 (comp_type_attributes): Likewise.
13854 (build_type_attribute_variant): Likewise.
13855 (lookup_ident_attribute): Likewise.
13856 (remove_attribute): Likewise.
13857 (merge_attributes): Likewise.
13858 (merge_type_attributes): Likewise.
13859 (merge_decl_attributes): Likewise.
13860 (merge_dllimport_decl_attributes): Likewise.
13861 (handle_dll_attribute): Likewise.
13862 (attribute_list_equal): Likewise.
13863 (attribute_list_contained): Likewise.
13864 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13865 (lookup_attribute_by_prefix): Likewise.
13866 * bb-reorder.c: Include header files.
13867 * builtins.c: Likewise.
13868 * calls.c: Likewise.
13869 * cfgexpand.c: Likewise.
13870 * cgraph.c: Likewise.
13871 * cgraphunit.c: Likewise.
13872 * convert.c: Likewise.
13873 * dwarf2out.c: Likewise.
13874 * final.c: Likewise.
13875 * fold-const.c: Likewise.
13876 * function.c: Likewise.
13877 * gimple-expr.c: Likewise.
13878 * gimple-fold.c: Likewise.
13879 * gimple-pretty-print.c: Likewise.
13880 * gimple.c: Likewise.
13881 * gimplify.c: Likewise.
13882 * hsa-common.c: Likewise.
13883 * hsa-gen.c: Likewise.
13884 * internal-fn.c: Likewise.
13885 * ipa-chkp.c: Likewise.
13886 * ipa-cp.c: Likewise.
13887 * ipa-devirt.c: Likewise.
13888 * ipa-fnsummary.c: Likewise.
13889 * ipa-inline.c: Likewise.
13890 * ipa-visibility.c: Likewise.
13892 * lto-cgraph.c: Likewise.
13893 * omp-expand.c: Likewise.
13894 * omp-general.c: Likewise.
13895 * omp-low.c: Likewise.
13896 * omp-offload.c: Likewise.
13897 * omp-simd-clone.c: Likewise.
13898 * opts-global.c: Likewise.
13899 * passes.c: Likewise.
13900 * predict.c: Likewise.
13901 * sancov.c: Likewise.
13902 * sanopt.c: Likewise.
13903 * symtab.c: Likewise.
13904 * toplev.c: Likewise.
13905 * trans-mem.c: Likewise.
13906 * tree-chkp.c: Likewise.
13907 * tree-eh.c: Likewise.
13908 * tree-into-ssa.c: Likewise.
13909 * tree-object-size.c: Likewise.
13910 * tree-parloops.c: Likewise.
13911 * tree-profile.c: Likewise.
13912 * tree-ssa-ccp.c: Likewise.
13913 * tree-ssa-live.c: Likewise.
13914 * tree-ssa-loop.c: Likewise.
13915 * tree-ssa-sccvn.c: Likewise.
13916 * tree-ssa-structalias.c: Likewise.
13917 * tree-ssa.c: Likewise.
13918 * tree-streamer-in.c: Likewise.
13919 * tree-vectorizer.c: Likewise.
13920 * tree-vrp.c: Likewise.
13921 * tsan.c: Likewise.
13922 * ubsan.c: Likewise.
13923 * varasm.c: Likewise.
13924 * varpool.c: Likewise.
13925 * tree.c: Remove functions moved to attribs.[ch].
13926 * tree.h: Likewise.
13927 * config/aarch64/aarch64.c: Add attrs.h header file.
13928 * config/alpha/alpha.c: Likewise.
13929 * config/arc/arc.c: Likewise.
13930 * config/arm/arm.c: Likewise.
13931 * config/avr/avr.c: Likewise.
13932 * config/bfin/bfin.c: Likewise.
13933 * config/c6x/c6x.c: Likewise.
13934 * config/cr16/cr16.c: Likewise.
13935 * config/cris/cris.c: Likewise.
13936 * config/darwin.c: Likewise.
13937 * config/epiphany/epiphany.c: Likewise.
13938 * config/fr30/fr30.c: Likewise.
13939 * config/frv/frv.c: Likewise.
13940 * config/ft32/ft32.c: Likewise.
13941 * config/h8300/h8300.c: Likewise.
13942 * config/i386/winnt.c: Likewise.
13943 * config/ia64/ia64.c: Likewise.
13944 * config/iq2000/iq2000.c: Likewise.
13945 * config/lm32/lm32.c: Likewise.
13946 * config/m32c/m32c.c: Likewise.
13947 * config/m32r/m32r.c: Likewise.
13948 * config/m68k/m68k.c: Likewise.
13949 * config/mcore/mcore.c: Likewise.
13950 * config/microblaze/microblaze.c: Likewise.
13951 * config/mips/mips.c: Likewise.
13952 * config/mmix/mmix.c: Likewise.
13953 * config/mn10300/mn10300.c: Likewise.
13954 * config/moxie/moxie.c: Likewise.
13955 * config/msp430/msp430.c: Likewise.
13956 * config/nds32/nds32-isr.c: Likewise.
13957 * config/nds32/nds32.c: Likewise.
13958 * config/nios2/nios2.c: Likewise.
13959 * config/nvptx/nvptx.c: Likewise.
13960 * config/pa/pa.c: Likewise.
13961 * config/pdp11/pdp11.c: Likewise.
13962 * config/powerpcspe/powerpcspe.c: Likewise.
13963 * config/riscv/riscv.c: Likewise.
13964 * config/rl78/rl78.c: Likewise.
13965 * config/rx/rx.c: Likewise.
13966 * config/s390/s390.c: Likewise.
13967 * config/sh/sh.c: Likewise.
13968 * config/sol2.c: Likewise.
13969 * config/sparc/sparc.c: Likewise.
13970 * config/spu/spu.c: Likewise.
13971 * config/stormy16/stormy16.c: Likewise.
13972 * config/tilegx/tilegx.c: Likewise.
13973 * config/tilepro/tilepro.c: Likewise.
13974 * config/v850/v850.c: Likewise.
13975 * config/vax/vax.c: Likewise.
13976 * config/visium/visium.c: Likewise.
13977 * config/xtensa/xtensa.c: Likewise.
13979 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13982 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13983 constraints since the -mupper-regs-* switches have been
13985 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13986 into a vector from a double word element that was extracted from
13987 another vector, and eliminate extra XXPERMDI instructions.
13988 (vsx_concat_<mode>_2): Likewise.
13989 (vsx_concat_<mode>_3): Likewise.
13990 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13991 concat to allow optimizing inserts from previous extracts.
13993 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13995 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13996 memory reference to a SSP offset in TLS address space.
13997 (ix86_print_operand) <case '@'>: Remove.
13998 (ix86_print_operand_punct_valid_p): Remove '@' code.
13999 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14000 UNSPEC_SP_TLS_TEST.
14001 (stack_tls_protect_set_<mode>): Remove.
14002 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14003 (stack_tls_protect_test_<mode>): Remove.
14004 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14006 2017-08-07 Olivier Hainque <hainque@adacore.com>
14009 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14011 2017-08-07 Douglas Rupp <rupp@adacore.com>
14013 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14014 variable was referenced as multidir in command.
14016 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14019 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14022 2017-08-07 Martin Liska <mliska@suse.cz>
14024 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14025 * config/rl78/rl78.c: Add include of attribs.h.
14026 * config/sh/sh.c: Likewise.
14027 * config/v850/v850.c: Likewise.
14029 2017-08-07 Tom de Vries <tom@codesourcery.com>
14031 PR middle-end/78266
14032 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14034 2017-08-07 Martin Liska <mliska@suse.cz>
14036 * config/mips/mips.c: Include attribs.h.
14038 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14041 * doc/invoke.texi: Document change in behvaior for -Ofast for
14044 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14046 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14048 (aarch64_pop_regs): Likewise.
14049 (aarch64_gen_load_pair): Likewise.
14050 (aarch64_save_callee_saves): Likewise.
14051 (aarch64_restore_callee_saves): Likewise.
14053 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14055 * config/i386/i386.c: Revert the last change.
14057 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14060 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14062 (ix86_finalize_stack_frame_flags): This. Also clear
14063 frame_pointer_needed if -fno-omit-frame-pointer is used without
14065 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14066 with ix86_finalize_stack_frame_flags.
14067 (ix86_expand_epilogue): Likewise.
14068 (ix86_expand_split_stack_prologue): Likewise.
14070 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14073 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14074 to P_AES for Westmere.
14076 2017-08-07 Jonathan Yong <10walls@gmail.com>
14078 * config/i386/mingw.opt (fset-stack-executable): Removed.
14079 * config/i386/cygming.opt (fset-stack-executable): Moved
14081 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14083 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14085 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14087 2017-08-07 Marek Polacek <polacek@redhat.com>
14089 PR middle-end/81737
14090 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14092 2017-08-07 Martin Liska <mliska@suse.cz>
14094 * attribs.h (canonicalize_attr_name): New function.
14095 (cmp_attribs): Move from c-format.c and adjusted.
14096 (is_attribute_p): Moved from tree.h.
14097 * tree-inline.c: Add new includes.
14098 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14099 (private_is_attribute_p): Remove.
14100 (private_lookup_attribute): Likewise.
14101 (private_lookup_attribute_by_prefix): Simplify.
14102 (remove_attribute): Use is_attribute_p.
14103 * tree.h: Remove removed declarations.
14105 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14107 PR middle-end/81698
14108 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14109 instead of computing it in the function. Formatting fix.
14110 (expand_case): Don't rely on default_edge being the first edge,
14111 clear it if removing it, pass default_edge to
14112 emit_case_dispatch_table.
14113 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14116 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14118 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14119 insn in the function, emit NOP after the insn.
14121 2017-08-06 Tom de Vries <tom@codesourcery.com>
14123 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14126 2017-08-06 Tom de Vries <tom@codesourcery.com>
14128 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14131 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14133 PR tree-optimization/57371
14134 * match.pd: New pattern.
14136 2017-08-04 Marek Polacek <polacek@redhat.com>
14138 PR middle-end/81695
14139 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14140 perform the computation in offset_int.
14142 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14144 PR tree-optimization/81136
14145 * tree-vectorizer.h: Include tree-hash-traits.h.
14146 (vec_base_alignments): New typedef.
14147 (vec_info): Add a base_alignments field.
14148 (vect_record_base_alignments): Declare.
14149 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14151 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14152 (create_data_ref): Add an is_conditional_in_stmt argument.
14153 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14154 the is_conditional_in_stmt field.
14155 (data_ref_loc): Add an is_conditional_in_stmt field.
14156 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14157 (find_data_references_in_stmt): Update call to create_data_ref.
14158 (graphite_find_data_references_in_stmt): Likewise.
14159 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14160 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14161 (vect_record_base_alignment): New function.
14162 (vect_record_base_alignments): Likewise.
14163 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14164 for nested statements even if we fail to compute a misalignment.
14165 Use pooled base alignments for unconditional references.
14166 (vect_find_same_alignment_drs): Compare base addresses instead
14168 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14169 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14171 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14173 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14174 Add an explicit name for the enum. Use auto_vec for slp_instances
14175 and grouped_stores.
14176 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14178 (_bb_vec_info): Add a constructor and destructor.
14179 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14180 (destroy_loop_vec_info): Delete.
14181 (vect_destroy_datarefs): Likewise.
14182 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14183 (vec_info::vec_info): New function.
14184 (vec_info::~vec_info): Likewise.
14185 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14186 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14187 destroy_loop_vec_info.
14188 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14189 (_loop_vec_info::_loop_vec_info): ...this.
14190 (destroy_loop_vec_info): Replace with...
14191 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14192 the stmt_vec_infos. Leave handling of vec_info information to its
14193 destructor. Remove explicit vector releases.
14194 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14195 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14196 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14197 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14198 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14199 (destroy_bb_vec_info): Replace with...
14200 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14201 information to its destructor.
14202 (vect_slp_analyze_bb_1): Use new and delete instead of
14203 new_bb_vec_info and destroy_bb_vec_info.
14204 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14207 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14209 * tree-data-ref.h (subscript): Add access_fn field.
14210 (data_dependence_relation): Add could_be_independent_p.
14211 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14212 (same_access_functions): Move to tree-data-ref.c.
14213 * tree-data-ref.c (ref_contains_union_access_p): New function.
14214 (access_fn_component_p): Likewise.
14215 (access_fn_components_comparable_p): Likewise.
14216 (dr_analyze_indices): Add a reference to access_fn_component_p.
14217 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14219 (constant_access_functions): Likewise.
14220 (add_other_self_distances): Likewise.
14221 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14222 (initialize_data_dependence_relation): Use XCNEW and remove
14223 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14224 of access functions that have the same type. Allow the
14225 subsequence to end with different bases in some circumstances.
14226 Record the chosen access functions in SUB_ACCESS_FN.
14227 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14228 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14229 (subscript_dependence_tester_1): Likewise dra and drb.
14230 (build_classic_dist_vector): Update calls accordingly.
14231 (subscript_dependence_tester): Likewise.
14232 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14233 DDR_COULD_BE_INDEPENDENT_P.
14234 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14235 comp_alias_ddrs instead of may_alias_ddrs.
14236 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14238 (vect_analyze_data_ref_dependence): Use it if
14239 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14240 distance vectors if that fails.
14241 (dependence_distance_ge_vf): New function.
14242 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14243 LOOP_VINFO_MAY_ALIAS_DDRS.
14245 2017-08-04 Richard Biener <rguenther@suse.de>
14247 PR middle-end/81705
14248 * fold-const.c (fold_binary_loc): Properly restrict
14249 minus_var0 && minus_var1 case when associating undefined overflow
14252 2017-08-04 Tom de Vries <tom@codesourcery.com>
14254 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14256 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14258 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14259 Don't start diagnostic messages with a capital letter.
14260 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14262 (rs6000_invalid_builtin): Likewise.
14263 (rs6000_trampoline_init): Likewise.
14265 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14268 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14269 after setting changeable df flags.
14271 2017-08-03 Richard Biener <rguenther@suse.de>
14273 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14274 up if the use is in USE - X.
14276 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14278 * toplev.c (dumpfile.h): New include.
14279 (internal_error_reentered): New static function. Use it...
14280 (internal_error_function): ...here to handle reentered internal_error.
14282 2017-08-03 Richard Biener <rguenther@suse.de>
14284 PR middle-end/81148
14285 * fold-const.c (split_tree): Add minus_var and minus_con
14286 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14287 here but always use minus_*.
14288 (associate_trees): Assert we never associate with MINUS_EXPR
14289 and NULL first operand. Do not recurse for PLUS_EXPR operands
14290 when associating as MINUS_EXPR either.
14291 (fold_binary_loc): Track minus_var and minus_con.
14293 2017-08-03 Tom de Vries <tom@codesourcery.com>
14296 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14297 ACCEL_COMPILER, apply finish_options on
14298 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14300 2017-08-03 Tom de Vries <tom@codesourcery.com>
14303 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14304 function_entry_patch_area_size > 0.
14306 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14309 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14310 instead of 10??LU, perform unit multiplication in wide_int,
14311 don't change alloc_object_size_limit if the limit is larger
14314 PR tree-optimization/81655
14315 PR tree-optimization/81588
14316 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14317 the case when ranges[i].low and high are 1 for unsigned type with
14320 PR middle-end/81052
14321 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14322 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14324 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14326 * tree-vrp.h: Add include guard.
14328 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14331 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14332 (ud2): New insn pattern.
14333 * config/i386/i386.c (ix86_expand_epilogue):
14334 For naked functions, generate ud2 instead of trap insn.
14336 2017-08-02 Marek Polacek <polacek@redhat.com>
14339 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14341 2017-08-02 Tom de Vries <tom@codesourcery.com>
14342 Cesar Philippidis <cesar@codesourcery.com>
14344 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14345 Add missing edge probabilities.
14347 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14349 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14350 Correct endianness.
14352 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14354 PR middle-end/79499
14355 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14356 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14357 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14358 prologue_seq sequences - if any.
14360 2017-08-02 Richard Biener <rguenther@suse.de>
14362 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14363 via vectors if supported, integer extracts via punning if supported
14364 or otherwise vector extracts.
14366 2017-08-02 Richard Biener <rguenther@suse.de>
14368 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14370 (bitmap_insert_into_set): ... this.
14372 2017-08-02 Richard Biener <rguenther@suse.de>
14374 PR tree-optimization/81633
14376 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14378 PR tree-optimization/71752
14379 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14381 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14383 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14384 (machine_function::call_ms2sysv_pad_out): Remove field.
14385 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14386 (ix86_compute_frame_layout): Likewise.
14388 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14391 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14392 attribute with interrupt attribute.
14394 2017-08-01 Andrew Pinski <apinski@cavium.com>
14396 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14397 BIT_INSERT_EXPR's operand 1
14398 to see if the types precision matches.
14400 2017-08-01 Martin Liska <mliska@suse.cz>
14402 PR middle-end/70140
14403 * builtins.c (expand_builtin_memcpy_args): Remove.
14404 (expand_builtin_memcpy): Call newly added function
14405 expand_builtin_memory_copy_args.
14406 (expand_builtin_memcpy_with_bounds): Likewise.
14407 (expand_builtin_mempcpy): Remove last argument.
14408 (expand_builtin_mempcpy_with_bounds): Likewise.
14409 (expand_builtin_memory_copy_args): New function created from
14410 expand_builtin_mempcpy_args with small modifications.
14411 (expand_builtin_mempcpy_args): Remove.
14412 (expand_builtin_stpcpy): Remove unused argument.
14413 (expand_builtin): Likewise.
14414 (expand_builtin_with_bounds): Likewise.
14416 2017-08-01 Martin Liska <mliska@suse.cz>
14419 Make mempcpy more optimal (PR middle-end/70140).
14421 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14424 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14425 __builtin_vec_cmpne verify both arguments are compatible vectors
14426 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14427 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14428 move computation of aligned to after checking the argument types.
14432 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14433 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14436 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14437 Doug Rupp <rupp@adacore.com>
14438 Olivier Hainque <hainque@adacore.com>
14440 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14441 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14443 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14444 for TARGET_OS_CPP_BUILTIN.
14445 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14446 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14448 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14449 passing required abi options to the assembler for EABI configurations.
14450 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14451 of .text.hot and .text.unlikely sections for kernel modules when
14452 using ARM style exceptions.
14453 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14454 options. Add EXTRA_CC1_SPEC.
14455 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14457 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14459 (ARM_TARGET2_DWARF_FORMAT): Define.
14460 * config/arm/t-vxworks: Adjust multilib control to removal of the
14461 Diab command line options.
14463 2017-08-01 Martin Liska <mliska@suse.cz>
14465 PR gcov-profile/81561
14466 * gcov.c (unblock): Make unblocking safe as we need to preserve
14467 index correspondence of blocks and block_lists.
14469 2017-08-01 Richard Biener <rguenther@suse.de>
14471 PR tree-optimization/81181
14472 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14473 (compute_antic): ... end of iteration here.
14475 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14477 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14478 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14479 (ftree-slp-vectorize): Likewise.
14480 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14481 can no longer be set independent of flag_tree_loop_vectorize.
14482 * omp-general.c (emp_max_vf): Likewise.
14483 * opts.c (enable_fdo_optimizations): Remove references to
14484 flag_tree_vectorize, these are now implicit.
14485 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14486 and leave it for the options machinery.
14488 2017-08-01 Martin Liska <mliska@suse.cz>
14490 PR middle-end/70140
14491 * builtins.c (expand_builtin_memcpy_args): Remove.
14492 (expand_builtin_memcpy): Call newly added function
14493 expand_builtin_memory_copy_args.
14494 (expand_builtin_memcpy_with_bounds): Likewise.
14495 (expand_builtin_mempcpy): Remove last argument.
14496 (expand_builtin_mempcpy_with_bounds): Likewise.
14497 (expand_builtin_memory_copy_args): New function created from
14498 expand_builtin_mempcpy_args with small modifications.
14499 (expand_builtin_mempcpy_args): Remove.
14500 (expand_builtin_stpcpy): Remove unused argument.
14501 (expand_builtin): Likewise.
14502 (expand_builtin_with_bounds): Likewise.
14504 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14507 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14508 print "ds:" only for immediates in generic address space.
14510 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14513 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14514 (ix86_function_ok_for_sibcall): Return false for naked functions.
14516 2017-08-01 Richard Biener <rguenther@suse.de>
14518 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14519 (compute_antic): Seed worklist with exit block predecessors.
14520 * cfganal.c (dfs_find_deadend): For a cycle return the source
14521 of the edge closing it.
14523 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14525 * config/aarch64/aarch64.c
14526 (aarch64_can_const_movi_rtx_p): Move 0 check.
14528 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14530 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14531 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14533 * match.pd: Ditto in address comparison pattern.
14535 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14537 PR tree-optimization/81627
14538 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14539 closed ssa form for store-store chain.
14541 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14543 PR tree-optimization/81620
14544 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14545 for store-store chain.
14547 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14549 PR tree-optimization/81588
14550 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14551 ranges[i].in_p, invert comparison code ccode. For >/>=,
14552 swap rhs1 and rhs2 and comparison code unconditionally,
14553 for </<= don't do that. Don't swap rhs1/rhs2 again if
14554 ranges[i].in_p, instead invert comparison code ccode if
14555 opcode or oe->rank is BIT_IOR_EXPR.
14558 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14559 a direct optab to conversion optab.
14560 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14561 with GET_MODE_INNER as last argument instead of optab_handler.
14562 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14563 vector extraction if possible and optab is available.
14564 * expr.c (store_constructor): Use convert_optab_handler instead
14565 of optab_handler. Use vector initialization from smaller
14566 vectors if possible and optab is available.
14567 * tree-vect-stmts.c (vectorizable_load): Likewise.
14568 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14569 now have two modes.
14570 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14571 of vec_init from half-sized vectors with the same element mode.
14572 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14573 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14574 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14575 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14576 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14577 after mode in gen_vec_extract* calls.
14578 (vec_extract<mode>): Renamed to ...
14579 (vec_extract<mode><ssescalarmodelower>): ... this.
14580 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14581 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14582 element mode after mode in gen_vec_init* calls.
14583 (VEC_INIT_HALF_MODE): New mode iterator.
14584 (vec_init<mode>): Renamed to ...
14585 (vec_init<mode><ssescalarmodelower>): ... this.
14586 (vec_init<mode><ssehalfvecmodelower>): New expander.
14587 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14588 (vec_extractv2sfsf): ... this.
14589 (vec_initv2sf): Renamed to ...
14590 (vec_initv2sfsf): ... this.
14591 (vec_extractv2si): Renamed to ...
14592 (vec_extractv2sisi): ... this.
14593 (vec_initv2si): Renamed to ...
14594 (vec_initv2sisi): ... this.
14595 (vec_extractv4hi): Renamed to ...
14596 (vec_extractv4hihi): ... this.
14597 (vec_initv4hi): Renamed to ...
14598 (vec_initv4hihi): ... this.
14599 (vec_extractv8qi): Renamed to ...
14600 (vec_extractv8qiqi): ... this.
14601 (vec_initv8qi): Renamed to ...
14602 (vec_initv8qiqi): ... this.
14603 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14604 (vec_init<mode>): Renamed to ...
14605 (vec_init<mode><VEC_base_l>): ... this.
14606 (vec_extract<mode>): Renamed to ...
14607 (vec_extract<mode><VEC_base_l>): ... this.
14608 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14609 (vec_initv2sfsf): ... this.
14610 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14611 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14612 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14613 element mode after mode in gen_vec_init* calls.
14614 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14615 (vec_init<mode><Vel>): ... this.
14616 (vec_extract<mode>): Renamed to ...
14617 (vec_extract<mode><Vel>): ... this.
14618 * config/aarch64/iterators.md (Vel): New mode attribute.
14619 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14620 Add element mode after mode in gen_vec_extract* calls.
14621 * config/s390/vector.md (non_vec_l): New mode attribute.
14622 (vec_extract<mode>): Renamed to ...
14623 (vec_extract<mode><non_vec_l>): ... this.
14624 (vec_init<mode>): Renamed to ...
14625 (vec_init<mode><non_vec_l>): ... this.
14626 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14627 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14629 * config/arm/iterators.md (V_elem_l): New mode attribute.
14630 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14631 (vec_extract<mode><V_elem_l>): ... this.
14632 (vec_extractv2di): Renamed to ...
14633 (vec_extractv2didi): ... this.
14634 (vec_init<mode>): Renamed to ...
14635 (vec_init<mode><V_elem_l>): ... this.
14636 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14637 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14638 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14639 Add element mode after gen_vec_extract* calls.
14640 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14641 (vec_init<mode><unitmode>): ... this.
14642 (vec_extract<mode>): Renamed to ...
14643 (vec_extract<mode><unitmode>): ... this.
14644 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14645 (vec_init<mode><unitmode>): ... this.
14646 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14647 (vec_initv2sfsf): ... this.
14648 (vec_extractv2sf): Renamed to ...
14649 (vec_extractv2sfsf): ... this.
14650 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14651 Add element mode after gen_vec_extract* calls.
14652 * config/mips/mips.md (unitmode): New mode iterator.
14653 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14654 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14655 * config/spu/spu.md (inner_l): New mode attribute.
14656 (vec_init<mode>): Renamed to ...
14657 (vec_init<mode><inner_l>): ... this.
14658 (vec_extract<mode>): Renamed to ...
14659 (vec_extract<mode><inner_l>): ... this.
14660 * config/sparc/sparc.md (veltmode): New mode iterator.
14661 (vec_init<VMALL:mode>): Renamed to ...
14662 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14663 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14664 (vec_initv2sisi): ... this.
14665 (vec_initv2sf): Renamed to ...
14666 (vec_initv2sfsf): ... this.
14667 (vec_extractv2sf): Renamed to ...
14668 (vec_extractv2sfsf): ... this.
14669 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14670 (vec_init<mode>): Renamed to ...
14671 (vec_init<mode><VEC_base_l>): ... this.
14672 (vec_extract<mode>): Renamed to ...
14673 (vec_extract<mode><VEC_base_l>): ... this.
14674 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14675 (vec_initv2sfsf): ... this.
14676 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14677 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14678 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14679 gen_vec_init* calls.
14681 2017-08-01 Richard Biener <rguenther@suse.de>
14683 PR tree-optimization/81297
14684 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14685 TREE_OVERFLOW from INTEGER_CSTs.
14687 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14689 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14691 2017-07-31 Carl Love <cel@us.ibm.com>
14693 * config/rs6000/rs6000-c: Add support for built-in functions
14694 vector signed char vec_xl_be (signed long long, signed char *);
14695 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14696 vector signed int vec_xl_be (signed long long, signed int *);
14697 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14698 vector signed long long vec_xl_be (signed long long, signed long long *);
14699 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14700 vector signed short vec_xl_be (signed long long, signed short *);
14701 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14702 vector double vec_xl_be (signed long long, double *);
14703 vector float vec_xl_be (signed long long, float *);
14704 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14705 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14706 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14708 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14709 (altivec_expand_builtin): Add switch statement to call
14710 altivec_expand_xl_be for each builtin.
14711 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14712 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14713 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14714 __builtin_vsx_le_be_v16qi.
14715 * doc/extend.texi: Update the built-in documentation file for the
14716 new built-in functions.
14718 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14721 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14723 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14725 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14727 * config.gcc: Add z14.
14728 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14729 CPU model numbers for z13s and z14.
14730 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14732 * config/s390/s390-opts.h (enum processor_type): Rename
14733 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14734 * config/s390/s390.c (processor_table): Add field for CPU name to
14735 be passed to Binutils.
14736 (s390_asm_output_machine_for_arch): Use the new field in
14737 processor_table for Binutils.
14738 (s390_expand_builtin): Replace arch12 with z14.
14739 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14740 (s390_get_sched_attrmask): Likewise.
14741 (s390_get_unit_mask): Likewise.
14742 * config/s390/s390.opt: Add z14 to processor_type enum.
14744 2017-07-31 Martin Jambor <mjambor@suse.cz>
14747 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14748 regardless of optimization level.
14750 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14751 Martin Liska <mliska@suse.cz>
14753 * predict.def: Remove old comment and adjust probability.
14754 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14755 PREDICT statements.
14757 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14760 * config/i386/i386.c (ix86_function_naked): New function.
14761 (ix86_can_use_return_insn_p): Return false for naked functions.
14762 (ix86_expand_prologue): Skip prologue for naked functions.
14763 (ix86_expand_epilogue): Skip epilogue for naked functions
14764 and emit trap instruction.
14765 (ix86_warn_func_return): New function.
14766 (ix86_attribute_table): Add "naked" attribute specification.
14767 (TARGET_WARN_FUNC_RETURN): Define.
14768 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14770 2017-07-31 Martin Liska <mliska@suse.cz>
14772 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14773 (dump_gimple_bb_header): Always dump BB info.
14774 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14776 2017-07-31 Martin Liska <mliska@suse.cz>
14779 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14780 also with current_function_decl non-null equality.
14782 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14785 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14786 change type to the element type, instead add eltype variable and
14787 use it where we are interested in the element type.
14789 PR tree-optimization/81603
14790 * ipa-polymorphic-call.c
14791 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14792 offset arithmetic in offset_int, bail out if the resulting bit offset
14793 doesn't fit into shwi.
14795 2017-07-31 Martin Liska <mliska@suse.cz>
14797 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14798 (gimplify_save_expr): Fix comment.
14800 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14803 * config/i386/i386.c (ix86_function_arg): Update arguments for
14805 (ix86_compute_frame_layout): Set the initial stack offset to
14806 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14807 INCOMING_FRAME_SP_OFFSET.
14808 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14809 stack before exception handler returns.
14810 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14811 the 'ERROR_CODE' for exception handler.
14813 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14815 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14816 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14817 (ASM_OUTPUT_REG_POP): Ditto.
14818 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14819 instead of asm_fprintf to output pure string.
14821 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14823 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14824 to imported_module_or_decl hook.
14825 (debug_nothing_tree_tree_tree_bool): Remove.
14826 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14827 * debug.c (do_nothing_debug_hooks): Use
14828 debug_nothing_tree_tree_tree_bool_bool instead of
14829 debug_nothing_tree_tree_tree_bool.
14830 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14831 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14832 * sdbout.c (sdb_debug_hooks): Likewise.
14833 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14834 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14836 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14837 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14838 attribute, don't add anything.
14840 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14842 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14843 (fold_build2_stat_loc): Likewise.
14844 (fold_build3_stat_loc): Likewise.
14845 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14846 (fold_build1_loc): Remove macro.
14847 (fold_build2_loc): Likewise.
14848 (fold_build3_loc): Likewise.
14850 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14852 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14853 (gimple_build_debug_bind_source_stat): Likewise.
14854 * gimple.h (gimple_build_debug_bind): Remove macro.
14855 (gimple_build_debug_bind_source): Likewise.
14857 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14859 * bitmap.c (bitmap_alloc): Adjust.
14860 (bitmap_gc_alloc): Likewise.
14861 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14863 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14865 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14866 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14867 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14868 (bitmap_gc_alloc_stat): Likewise.
14869 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14871 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14873 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14874 * rtl.h (shallow_copy_rtx): Remove macro.
14876 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14878 * emit-rtl.c (gen_raw_REG): Adjust.
14879 * gengenrtl.c (gendef): Likewise.
14880 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14881 * rtl.h (rtx_alloc): Remove macro.
14883 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14885 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14886 (build_tree_list_stat): Likewise.
14887 * tree.h (build_tree_list): Remove macro.
14888 (build_tree_list_vec): Likewise.
14890 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14892 * tree.c (make_vector_stat): Remove _stat from name.
14893 (build_vector_stat): Likewise.
14894 * tree.h (make_vector_stat): Remove macro.
14895 (build_vector_stat): Likewise.
14897 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14899 * tree.h (build_var_debug_value): Remove prototype.
14901 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14903 * tree.c (tree_cons_stat): Remove _stat from name.
14904 * tree.h (tree_cons): Remove macro.
14906 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14908 * tree.c (build_vl_exp_stat): Remove _stat from name.
14909 * tree.h (build_vl_exp): Remove macro.
14911 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14913 * tree.c (build_decl_stat): Remove _stat from name.
14914 * tree.h (build_decl): Remove macro.
14916 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14918 * gimple.c (gimple_build_with_ops_stat): Adjust.
14919 (gimple_alloc_stat): Remove _stat from name.
14920 * gimple.h (gimple_alloc): Remove macro.
14922 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14924 * tree.c (make_tree_vec_stat): Remove _stat from name.
14925 (grow_tree_vec_stat): Likewise.
14926 * tree.h (make_tree_vec_stat): Adjust prototype.
14927 (grow_tree_vec_stat): Likewise.
14928 (make_tree_vec): Remove macro.
14929 (grow_tree_vec): Likewise.
14931 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14933 * fold-const.c (fold_build1_stat_loc): Adjust.
14934 (fold_build2_stat_loc): Likewise.
14935 (fold_build3_stat_loc): Likewise.
14936 * tree.c (build0_stat): Remove _stat from name.
14937 (build1_stat): Likewise.
14938 (build2_stat): Likewise.
14939 (build3_stat): Likewise.
14940 (build4_stat): Likewise.
14941 (build5_stat): Likewise.
14942 * tree.h (build1_loc): Remove macro, and rename _stat function
14944 (build2_loc): Likewise.
14945 (build3_loc): Likewise.
14946 (build4_loc): Likewise.
14947 (build5_loc): Likewise.
14949 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14951 * tree.c (make_int_cst_stat): Remove _stat from name.
14952 * tree.h (make_int_cst_stat): Adjust prototype.
14953 (make_int_cst): Remove macro.
14955 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14957 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14958 * tree.h (make_tree_binfo_stat): Adjust prototype.
14959 (make_tree_binfo): Remove.
14961 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14963 * tree.c (copy_node_stat): Rename to copy_node.
14964 (build_distinct_type_copy): Adjust.
14965 * tree.h (copy_node_stat): Adjust prototype.
14966 (copy_node): Remove macro.
14968 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14970 * tree.c (make_node_stat): rename to make_node.
14971 (build_tree_list_stat): Adjust.
14972 (build0_stat): Likewise.
14973 (build2_stat): Likewise.
14974 (build3_stat): Likewise.
14975 (build4_stat): Likewise.
14976 (build5_stat): Likewise.
14977 (build_decl_stat): Likewise.
14978 * tree.h (make_node_stat): Adjust prototype.
14979 (make_node): remove macro.
14981 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14983 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14984 (PPC_FEATURE2_SCV): Likewise.
14985 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14987 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14989 * config/aarch64/aarch64.c
14990 (aarch64_internal_mov_immediate): Add new special pattern.
14991 * config/aarch64/aarch64.md (*movdi_aarch64):
14992 Add reg/32bit const mov case.
14994 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14995 Richard Sandiford <richard.sandiford@linaro.org>
14997 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14998 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14999 Add integer and movi cases.
15000 (movi-split-hf-df-sf split, fp16): New.
15001 (enabled): Added TARGET_FP_F16INST.
15002 * config/aarch64/iterators.md (GPF_HF): New.
15003 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15005 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15007 * config/aarch64/aarch64.c
15008 (aarch64_simd_container_mode): Add prototype.
15009 (aarch64_expand_mov_immediate): Add HI support.
15010 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15011 (aarch64_can_const_movi_rtx_p): New.
15012 (aarch64_preferred_reload_class):
15013 Remove restrictions of using FP registers for certain SIMD operations.
15014 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15015 (aarch64_valid_floating_const): Add integer move validation.
15016 (aarch64_simd_imm_scalar_p): Remove.
15017 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15018 (aarch64_legitimate_constant_p): Expand list of supported cases.
15019 * config/aarch64/aarch64-protos.h
15020 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15021 (aarch64_reinterpret_float_as_int): New.
15022 (aarch64_simd_imm_scalar_p): Remove.
15023 * config/aarch64/constraints.md (Uvi): New.
15024 (Dd): Split into Ds and new Dd.
15025 * config/aarch64/aarch64.md (*movsi_aarch64):
15027 (*movdi_aarch64): Add SIMD mov case.
15029 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15031 * tree-predcom.c: (struct chain): Handle store-store chain in which
15032 stores for elimination only store loop invariant values.
15033 (execute_pred_commoning_chain): Ditto.
15034 (prepare_initializers_chain_store_elim): Ditto.
15035 (prepare_finalizers): Ditto.
15036 (is_inv_store_elimination_chain): New function.
15037 (initialize_root_vars_store_elim_1): New function.
15039 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15041 * tree-predcom.c: Revise general description of the pass.
15042 (enum chain_type): New enum type for store elimination.
15043 (struct chain): New field supporting store elimination.
15044 (struct component): Ditto.
15045 (dump_chain): Dump store-stores chain.
15046 (release_chain): Release resources.
15047 (split_data_refs_to_components): Compute and create component
15048 contains only stores for elimination.
15049 (get_chain_last_ref_at): New function.
15050 (make_invariant_chain): Initialization.
15051 (make_rooted_chain): Specify chain type in parameter and record it.
15052 (add_looparound_copies): Skip for store-stores chain.
15053 (determine_roots_comp): Compute type of chain and pass it to
15055 (initialize_root_vars_store_elim_2): New function.
15056 (finalize_eliminated_stores): New function.
15057 (remove_stmt): Handle store for elimination.
15058 (execute_pred_commoning_chain): Execute predictive commoning on
15059 store-store chains.
15060 (determine_unroll_factor): Skip unroll for store-stores chain.
15061 (prepare_initializers_chain_store_elim): New function.
15062 (prepare_initializers_chain): Hanlde store-store chain.
15063 (prepare_finalizers_chain, prepare_finalizers): New function.
15064 (tree_predictive_commoning_loop): Return integer value indicating
15065 if loop is unrolled or lcssa form is corrupted.
15066 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15068 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15070 * tree-predcom.c (initialize_root): Delete.
15071 (execute_pred_commoning_chain): Initialize root vars and replace
15072 reference of non-combined chain directly, rather than call above
15075 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15077 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15078 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15080 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15082 * tree-predcom.c (struct chain): New field init_seq.
15083 (release_chain): Release init_seq.
15084 (prepare_initializers_chain): Record intialization stmts in above
15086 (insert_init_seqs): New function.
15087 (tree_predictive_commoning_loop): Call insert_init_seqs.
15089 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15091 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15093 2017-07-28 Richard Biener <rguenther@suse.de>
15095 * match.pd: Remove superfluous :c.
15096 * genmatch.c (simplify::id): Add member.
15097 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15099 (current_id): New global.
15100 (dt_node::parent): Move from ...
15101 (dt_operand::parent): ... here. Add for_id member.
15102 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15103 (decision_tree::find_node): Relax order requirement when
15104 merging DT_TRUE nodes to ones inbetween the current simplify
15105 and the one we try to merge with. Add diagnostic whenever
15106 we need to enforce pattern order by not merging.
15107 (decision_tree::insert): Set current_id.
15108 (decision_tree::print_node): Dump parent node and for_id.
15109 (parser::last_id): Add member.
15110 (parser::push_simplify): Assign unique id.
15111 (parser::parser): Initialize last_id.
15113 2017-07-28 Martin Liska <mliska@suse.cz>
15116 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15117 gimple_build_debug_bind.
15119 2017-07-28 Richard Biener <rguenther@suse.de>
15121 PR tree-optimization/81502
15122 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15124 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15126 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15127 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15128 for BIT_FIELD_REF args.
15129 * fold-const.c (make_bit_field_ref): Likewise.
15130 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15132 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15135 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15136 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15137 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15138 Or it into SANITIZER_UNDEFINED.
15139 * ubsan.c: Include gimple-fold.h and varasm.h.
15140 (ubsan_expand_ptr_ifn): New function.
15141 (instrument_pointer_overflow): New function.
15142 (maybe_instrument_pointer_overflow): New function.
15143 (instrument_object_size): Formatting fix.
15144 (pass_ubsan::execute): Call instrument_pointer_overflow
15145 and maybe_instrument_pointer_overflow.
15146 * internal-fn.c (expand_UBSAN_PTR): New function.
15147 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15148 * sanitizer.def (__ubsan_handle_pointer_overflow,
15149 __ubsan_handle_pointer_overflow_abort): New builtins.
15150 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15151 * internal-fn.def (UBSAN_PTR): New internal function.
15152 * opts.c (sanitizer_opts): Add pointer-overflow.
15153 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15154 * fold-const.c (build_range_check): Compute pointer range check in
15155 integral type if pointer arithmetics would be needed. Formatting
15158 2017-07-28 Martin Liska <mliska@suse.cz>
15161 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15162 parameters that are of a variable-length.
15164 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15166 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15168 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15169 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15170 (CRT_CALL_STATIC_FUNCTION): Likewise.
15171 (ASM_DEFAULT_SPEC): New define.
15172 (ASM_SPEC32): Likewise.
15173 (ASM_SPEC64): Likewise.
15174 (ASM_SPEC_COMMON): Likewise.
15175 (ASM_SPEC): Likewise.
15176 (INVALID_64BIT): Likewise.
15177 (LINK_OS_DEFAULT_SPEC): Likewise.
15178 (LINK_OS_SPEC32): Likewise.
15179 (LINK_OS_SPEC64): Likewise.
15180 (POWERPC_LINUX): Likewise.
15181 (PTRDIFF_TYPE): Likewise.
15182 (RESTORE_FP_PREFIX): Likewise.
15183 (RESTORE_FP_SUFFIX): Likewise.
15184 (SAVE_FP_PREFIX): Likewise.
15185 (SAVE_FP_SUFFIX): Likewise.
15186 (SIZE_TYPE): Likewise.
15187 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15188 (TARGET_64BIT): Likewise.
15189 (TARGET_64BIT): Likewise.
15190 (TARGET_AIX): Likewise.
15191 (WCHAR_TYPE_SIZE): Likewise.
15192 (WCHAR_TYPE): Undefine.
15193 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15194 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15195 (CPP_OS_RTEMS_SPEC): Delete.
15196 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15197 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15199 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15201 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15203 PR tree-optimization/81578
15204 * tree-parloops.c (build_new_reduction): Bail out if
15205 reduction_code isn't one of the standard OpenMP reductions.
15206 Move the details printing after that decision.
15208 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15210 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15211 related to reload_in_progress.
15212 (splat_input_operand): Likewise.
15213 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15215 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15217 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15218 (TARGET_INSTANTIATE_DECLS): Likewise.
15219 (legitimate_indexed_address_p): Delete reload_in_progress code.
15220 (rs6000_debug_legitimate_address_p): Likewise.
15221 (rs6000_eliminate_indexed_memrefs): Likewise.
15222 (rs6000_emit_le_vsx_store): Likewise.
15223 (rs6000_emit_move_si_sf_subreg): Likewise.
15224 (rs6000_emit_move): Likewise.
15225 (register_to_reg_type): Likewise.
15226 (rs6000_pre_atomic_barrier): Likewise.
15227 (rs6000_machopic_legitimize_pic_address): Likewise.
15228 (rs6000_allocate_stack_temp): Likewise.
15229 (rs6000_address_for_fpconvert): Likewise.
15230 (rs6000_address_for_altivec): Likewise.
15231 (rs6000_secondary_memory_needed_rtx): Delete function.
15232 (rs6000_check_sdmode): Likewise.
15233 (rs6000_alloc_sdmode_stack_slot): Likewise.
15234 (rs6000_instantiate_decls): Likewise.
15235 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15236 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15237 Delete reload_in_progress.
15238 (*vec_reload_and_plus_<mptrsize>): Likewise.
15239 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15240 (vsx_div_v2di): Likewise.
15241 (vsx_udiv_v2di): Likewise.
15243 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15245 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15246 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15247 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15248 (rs6000_debug_reg_global): Delete print of LRA status.
15249 (rs6000_option_override_internal): Delete dead LRA related code.
15250 (rs6000_lra_p): Delete function.
15251 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15253 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15255 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15256 * config/riscv/rtems.h: New file.
15258 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15259 Sudakshina Das <sudi.das@arm.com>
15261 * config/aarch64/aarch64.md
15262 (define_split for and<mode>3nr_compare): Move
15263 non aarch64_logical_operand to a register.
15264 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15265 register immediate operand to a register.
15266 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15268 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15270 PR middle-end/81564
15271 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15273 2017-07-27 Richard Biener <rguenther@suse.de>
15275 PR tree-optimization/81573
15276 PR tree-optimization/81494
15277 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15278 multi defuse cycle case.
15280 2017-07-27 Richard Biener <rguenther@suse.de>
15282 PR tree-optimization/81571
15283 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15286 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15288 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15289 earlier and only if MASK_FPU is set. Adjust formatting.
15291 2017-07-27 Martin Liska <mliska@suse.cz>
15293 * opt-functions.awk: Add validation of value of Init.
15294 * optc-gen.awk: Pass new argument.
15296 2017-07-27 Martin Liska <mliska@suse.cz>
15298 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15299 Fix wrong condition.
15301 2017-07-27 Martin Liska <mliska@suse.cz>
15303 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15304 BBs and edges seen by autoFDO.
15306 2017-07-27 Richard Biener <rguenther@suse.de>
15308 PR tree-optimization/81502
15309 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15310 with incompatible but same sized type.
15311 (execute_update_addresses_taken): Likewise.
15313 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15315 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15316 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15318 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15321 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15322 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15323 Change s_operand to memory_operand.
15325 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15327 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15328 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15329 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15330 Emit instructions rather than returning an expression. Handle TFmode
15331 and KFmode by casting to TImode.
15332 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15333 (rs6000_emit_le_vsx_store): Likewise.
15334 * config/rs6000/vsx.md (VSX_TI): New iterator.
15335 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15336 (*vsx_le_undo_permute_<mode>): Likewise.
15337 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15338 emit the split sequence.
15339 (*vsx_le_perm_store_<mode>): Likewise.
15341 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15343 PR tree-optimization/81555
15344 PR tree-optimization/81556
15345 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15346 if true, force CHANGED for the recursive invocation.
15347 (reassociate_bb): Remember original length of ops array, pass
15348 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15350 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15351 attributes for noipa attribute. For naked attribute use
15352 lookup_attribute first before lookup_attribute_spec.
15353 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15355 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15356 for functions with noipa attribute.
15357 (cgraph_externally_visible_p): Return true for functions with noipa
15359 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15360 for functions with noipa attribute.
15361 * doc/extend.texi: Document noipa function attribute.
15362 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15363 also for functions with noipa attribute.
15364 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15366 2017-07-26 Andrew Pinski <apinski@cavium.com>
15368 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15369 vec_unalign_load_cost and vec_unalign_store_cost.
15371 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15373 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15374 -mvsx-small-integer option.
15375 (ISA_3_0_MASKS_IEEE): Likewise.
15376 (OTHER_VSX_VECTOR_MASKS): Likewise.
15377 (POWERPC_MASKS): Likewise.
15378 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15379 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15380 code, only testing for DImode being allowed in non-VSX floating
15382 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15383 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15385 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15386 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15387 TARGET_P8_VECTOR test.
15388 (rs6000_secondary_reload_simple_move): Likewise.
15389 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15390 since TARGET_P9_VECTOR was already tested.
15391 (rs6000_opt_masks): Remove -mvsx-small-integer.
15392 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15393 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15395 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15396 test for TARGET_VEXTRACTUB was used, and that uses
15398 (p9 extract splitter): Likewise.
15399 (vsx_extract_<mode>_di_p9): Likewise.
15400 (vsx_extract_<mode>_store_p9): Likewise.
15401 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15402 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15403 the elimination of TARGET_VSX_SMALL_INTEGER.
15404 (vsx_extract_<mode>_p8): Likewise.
15405 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15406 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15407 (vsx_set_<mode>_p9): Likewise.
15408 (vsx_set_v4sf_p9): Likewise.
15409 (vsx_set_v4sf_p9_zero): Likewise.
15410 (vsx_insert_extract_v4sf_p9): Likewise.
15411 (vsx_insert_extract_v4sf_p9_2): Likewise.
15412 * config/rs6000/rs6000.md (sign extend splitter): Change
15413 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15414 (floatsi<mode>2_lfiwax_mem): Likewise.
15415 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15416 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15417 since a test for TARGET_P9_VECTOR was used.
15418 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15419 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15420 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15421 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15422 TARGET_P8_VECTOR test.
15423 (fix_trunc<mode>si2_stfiwx): Likewise.
15424 (fix_trunc<mode>si2_internal): Likewise.
15425 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15426 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15428 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15429 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15430 TARGET_P8_VECTOR test.
15431 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15432 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15433 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15435 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15436 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15437 since a test for TARGET_P9_VECTOR was used.
15438 (splitter for loading small constants): Likewise.
15440 2017-07-26 Andrew Pinski <apinski@cavium.com>
15442 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15445 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15448 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15449 (fp_valid_at): Likewise.
15451 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15453 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15454 (qdf24xx_addrcost_table): Likewise.
15455 (cortexa57_tunings): Update to use generic_branch_cost.
15456 (cortexa72_tunings): Likewise.
15457 (cortexa73_tunings): Likewise.
15458 (qdf24xx_tunings): Likewise.
15460 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15462 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15463 (thunderx2t99_branch_cost): Likewise.
15464 (cortexa35_tunings): Update to use generic_branch_cost.
15465 (cortexa53_tunings): Likewise.
15466 (cortexa57_tunings): Likewise.
15467 (cortexa72_tunings): Likewise.
15468 (cortexa73_tunings): Likewise.
15469 (thunderx2t99_tunings): Likewise.
15471 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15473 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15474 (sparc_option_override): Honour MASK_FSMULD.
15475 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15476 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15477 * config/sparc/sparc.opt (mfsmuld): New option.
15478 * doc/invoke.texi (mfsmuld): Document option.
15480 2017-07-26 Marek Polacek <polacek@redhat.com>
15482 PR middle-end/70992
15483 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15485 2017-07-26 Richard Biener <rguenther@suse.de>
15487 * gimple-match-head.c (do_valueize): Return OP if valueize
15489 (get_def): New helper to get at the def stmt of a SSA name
15490 if valueize allows.
15491 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15492 do_valueize to get at the def stmt.
15493 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15495 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15497 PR middle-end/46932
15498 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15500 2017-07-26 Martin Liska <mliska@suse.cz>
15503 * function.c (expand_function_start): Make expansion of
15504 nonlocal_goto_save_area after parm_birth_insn.
15506 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15508 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15509 from all CPU target flags enable members.
15511 2017-07-26 Richard Biener <rguenther@suse.de>
15513 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15514 (decision_tree::gen): Make 'type' const.
15515 (write_predicate): Likewise.
15517 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15519 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15520 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15521 (rs6000_option_override_internal): Likewise.
15522 (rs6000_expand_vector_set): Likewise.
15523 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15524 (TARGET_UPPER_REGS_SF): Likewise.
15525 (TARGET_UPPER_REGS_DI): Likewise.
15526 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15527 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15528 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15529 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15530 (Splitters for DI constants in Altivec registers): Likewise.
15531 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15532 (vsx_set_v4sf_p9): Likewise.
15533 (vsx_set_v4sf_p9_zero): Likewise.
15534 (vsx_insert_extract_v4sf_p9): Likewise.
15535 (vsx_insert_extract_v4sf_p9_2): Likewise.
15537 2017-07-25 Carl Love <cel@us.ibm.com>
15539 * doc/extend.texi: Update the built-in documentation file for the
15540 existing built-in functions
15541 vector signed char vec_cnttz (vector signed char);
15542 vector unsigned char vec_cnttz (vector unsigned char);
15543 vector signed short vec_cnttz (vector signed short);
15544 vector unsigned short vec_cnttz (vector unsigned short);
15545 vector signed int vec_cnttz (vector signed int);
15546 vector unsigned int vec_cnttz (vector unsigned int);
15547 vector signed long long vec_cnttz (vector signed long long);
15548 vector unsigned long long vec_cnttz (vector unsigned long long);
15550 2017-07-25 Andrew Pinski <apinski@cavium.com>
15552 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15553 accesses where the use is for the first operand of a BIT_INSERT.
15555 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15558 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15559 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15561 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15563 * config/i386/gstabs.h: Delete.
15564 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15566 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15568 * config/i386/i386.c (ix86_decompose_address): Do not check for
15569 register RTX when looking at index_reg or base_reg.
15570 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15572 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15574 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15575 to update EH info here.
15577 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15579 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15581 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15583 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15585 2017-07-25 Torsten Duwe <duwe@suse.de>
15587 * common.opt: Introduce -fpatchable-function-entry
15588 command line option, and its variables function_entry_patch_area_size
15589 and function_entry_patch_area_start.
15590 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15591 including a two-value parser.
15592 * target.def (print_patchable_function_entry): New target hook.
15593 * targhooks.h (default_print_patchable_function_entry): New function.
15594 * targhooks.c (default_print_patchable_function_entry): Likewise.
15595 * toplev.c (process_options): Switch off IPA-RA if
15596 patchable function entries are being generated.
15597 * varasm.c (assemble_start_function): Look at the
15598 patchable-function-entry command line switch and current
15599 function attributes and maybe generate NOP instructions by
15600 calling the print_patchable_function_entry hook.
15601 * doc/extend.texi: Document patchable_function_entry attribute.
15602 * doc/invoke.texi: Document -fpatchable_function_entry
15603 command line option.
15604 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15606 * doc/tm.texi: Re-generate.
15608 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15611 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15612 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15614 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15616 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15618 2017-07-25 Richard Biener <rguenther@suse.de>
15620 PR tree-optimization/81455
15621 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15622 not walk in cycles when looking for guards.
15624 2017-07-25 Richard Biener <rguenther@suse.de>
15626 PR tree-optimization/81529
15627 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15628 when optimizing backedge uses.
15630 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15632 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15634 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15635 to dl_section_ref. On AIX, append an expression to subtract
15636 the size of the section length to dl_section_ref.
15638 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15640 * configure.ac: If any of the config.* scripts fail, exit 1.
15641 * configure: Regenerate.
15643 2017-07-25 Richard Biener <rguenther@suse.de>
15645 PR middle-end/81546
15646 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15647 of immediate uses, be more verbose on errors.
15649 2017-07-25 Richard Biener <rguenther@suse.de>
15651 PR tree-optimization/81510
15652 * tree-vect-loop.c (vect_is_simple_reduction): When the
15653 reduction stmt is not inside the loop bail out.
15655 2017-07-25 Richard Biener <rguenther@suse.de>
15657 PR tree-optimization/81303
15658 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15659 profitability check against LOOP_VINFO_NITERSM1.
15661 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15663 * domwalk.c (cmp_bb_postorder): Simplify.
15664 (sort_bbs_postorder): New function. Use it...
15665 (dom_walker::walk): ...here to optimize common cases.
15667 2017-07-25 Martin Liska <mliska@suse.cz>
15670 * ipa-visibility.c (function_and_variable_visibility): Make the
15671 redirection just on target that supports aliasing.
15672 Fix GNU coding style.
15674 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15677 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15679 * config/arm/rtems.h: Likewise.
15680 * config/bfin/rtems.h: Likewise.
15681 * config/i386/rtemself.h: Likewise.
15682 * config/lm32/rtems.h: Likewise.
15683 * config/m32c/rtems.h: Likewise.
15684 * config/m68k/rtemself.h: Likewise.
15685 * config/microblaze/rtems.h: Likewise.
15686 * config/mips/rtems.h: Likewise.
15687 * config/moxie/rtems.h: Likewise.
15688 * config/nios2/rtems.h: Likewise.
15689 * config/powerpcspe/rtems.h: Likewise.
15690 * config/rs6000/rtems.h: Likewise.
15691 * config/rtems.h: Likewise.
15692 * config/sh/rtems.h: Likewise.
15693 * config/sh/rtemself.h: Likewise.
15694 * config/sparc/rtemself.h: Likewise.
15696 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15699 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15700 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15701 * tree-ssa-structalias.c (alias_get_name): Same.
15703 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15706 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15707 instructions if no du chain is found.
15709 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15711 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15713 2017-07-25 Richard Biener <rguenther@suse.de>
15715 PR middle-end/81505
15716 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15719 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15721 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15722 upper-regs options.
15723 (ISA_2_7_MASKS_SERVER): Likewise.
15724 (ISA_3_0_MASKS_IEEE): Likewise.
15725 (OTHER_P8_VECTOR_MASKS): Likewise.
15726 (OTHER_VSX_VECTOR_MASKS): Likewise.
15727 (POWERPC_MASKS): Likewise.
15728 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15729 duplicate list of options.
15730 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15731 explicit -mupper-regs options.
15732 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15733 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15734 alias for -mupper-regs-df.
15735 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15736 (rs6000_init_hard_regno_mode_ok): Likewise.
15737 (rs6000_option_override_internal): Likewise.
15738 (rs6000_opt_masks): Likewise.
15739 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15740 options in terms of whether -mvsx or -mpower8-vector was used.
15741 (TARGET_UPPER_REGS_DI): Likewise.
15742 (TARGET_UPPER_REGS_SF): Likewise.
15743 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15744 -mupper-regs-* options.
15746 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15748 * passes.c (emergency_dump_function): Print some empty lines and a
15749 header before the RTL dump.
15751 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15753 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15755 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15758 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15759 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15761 2017-07-24 Carl Love <cel@us.ibm.com>
15763 * config/rs6000/rs6000-c.c: Add support for built-in functions
15764 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15765 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15766 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15767 vec_extract_fp_from_shortl): Add defines for the two builtins.
15768 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15769 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15771 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15772 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15773 * doc/extend.texi: Update the built-in documentation file for the
15774 new built-in function.
15776 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15779 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15781 * doc/generic.texi: Likewise.
15782 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15783 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15785 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15787 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15788 (aarch64_mls_elt_merge<mode>): Likewise.
15790 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15792 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15793 having __cxa_atexit.
15795 2017-07-23 Michael Collison <michael.collison@arm.com>
15797 * config/arm/arm.c (arm_option_override): Deprecate
15798 use of -mstructure-size-boundary.
15799 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15800 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15802 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15805 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15806 Reduce cost estimate for direct moves.
15808 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15811 * config/i386/i386.c (ix86_option_override_internal): Disable
15812 BMI, BMI2 and TBM instructions for -m16.
15814 2017-07-21 Carl Love <cel@us.ibm.com>
15816 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15817 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15818 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15819 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15820 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15821 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15822 VMULOSW): New enum "unspec" values.
15823 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15824 altivec_vmulosw): New patterns.
15825 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15826 VMULOSW): Add definitions.
15828 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15830 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15831 (qdf24xx): Likewise.
15832 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15833 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15834 (AARCH64_FL_V8_1): Renumber.
15835 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15836 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15837 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15838 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15839 rdma to feature modifiers list.
15841 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15843 PR middle-end/56727
15844 * ipa-visibility (function_and_variable_visibility): Convert
15845 recursive PLT call to direct call if appropriate.
15847 2017-07-21 Andrew Pinski <apinski@cavium.com>
15849 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15850 operand 1 to see if the types precision matches.
15851 * fold-const.c (operand_equal_p): Likewise.
15853 2017-07-21 Richard Biener <rguenther@suse.de>
15855 PR tree-optimization/81303
15856 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15857 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15858 (vect_peeling_hash_get_lowest_cost): Adjust.
15859 (vect_enhance_data_refs_alignment): Likewise. Use
15860 vect_get_peeling_costs_all_drs to compute the penalty for no
15861 peeling to match up costs.
15863 2017-07-21 Richard Biener <rguenther@suse.de>
15865 PR tree-optimization/81500
15866 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15867 we didn't identify a reduction path.
15869 2017-07-21 Tom de Vries <tom@codesourcery.com>
15870 Cesar Philippidis <cesar@codesourcery.com>
15872 PR gcov-profile/81442
15873 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15876 2017-07-21 Tom de Vries <tom@codesourcery.com>
15879 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15881 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15882 nvptx_override_options_after_change.
15884 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15886 * dwarf2out.c (output_file_names): Avoid double testing for
15887 dwarf_version >= 5.
15889 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15891 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15893 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15895 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15897 (try_crossjump_to_edge): Do not punt on partitioned functions.
15899 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15901 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15902 Put all BBs reachable only via paths crossing cold region to cold
15904 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15906 2016-07-21 Richard Biener <rguenther@suse.de>
15908 PR tree-optimization/81303
15909 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15910 into account prologue and epilogue iterations when raising
15911 min_profitable_iters to sth at least covering one vector iteration.
15913 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15915 * config/arm/arm.c (arm_test_cpu_arch_dat):
15918 2017-07-20 Nathan Sidwell <nathan@acm.org>
15920 Remove TYPE_METHODS.
15921 * tree.h (TYPE_METHODS): Delete.
15922 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15923 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15924 (dbxout_type_methods): Scan TYPE_FIELDS.
15925 (dbxout_type): Don't check TYPE_METHODS here.
15926 * function.c (use_register_for_decl): Always ignore register for
15927 class types when not optimizing.
15928 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15929 * tree.c (free_lang_data_in_type): Stitch out member functions and
15930 templates from TYPE_FIELDS.
15931 (build_distinct_type_copy, verify_type_variant,
15932 verify_type): Member fns are on TYPE_FIELDS.
15933 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15934 * tree-pretty-print.c (dump_generic_node): Likewise.
15936 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15939 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15940 V2TImode and V4TImode.
15941 (ix86_expand_vector_extract): Likewise.
15942 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15943 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15944 (ssescalarmode): Handle V4TImode and V2TImode.
15945 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15946 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15947 (VEXTRACTI128_MODE): New mode iterator.
15948 (splitter for *vec_extractv?ti first element): New.
15949 (VEC_INIT_MODE): New mode iterator.
15950 (vec_init<mode>): Consolidate 3 expanders into one using
15951 VEC_INIT_MODE mode iterator.
15953 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15955 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15956 non_spilled_static_chain_regno_p.
15958 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15960 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15962 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15964 * bb-reorder.c (connect_traces): Allow copying of blocks within
15967 2017-07-20 Richard Biener <rguenther@suse.de>
15969 * gimple.h (gimple_phi_result): Add gphi * overload.
15970 (gimple_phi_result_ptr): Likewise.
15971 (gimple_phi_arg): Likewise. Adjust index assert to only
15972 allow actual argument accesses rather than all slots available
15974 (gimple_phi_arg_def): Add gphi * overload.
15975 * tree-phinodes.c (make_phi_node): Initialize only actual
15977 (resize_phi_node): Clear memory not covered by old node,
15978 do not initialize excess argument slots.
15979 (reserve_phi_args_for_new_edge): Initialize new argument slot
15982 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15984 PR tree-optimization/81388
15986 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15988 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15989 by removing computation of may_be_zero.
15991 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15992 Tom de Vries <tom@codesourcery.com>
15994 PR middle-end/81030
15995 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15996 when gimple level profile disagrees with what RTL expander did.
15998 2017-07-20 Richard Biener <rguenther@suse.de>
16000 PR tree-optimization/61171
16001 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16002 (vect_analyze_stmt): Add slp instance parameter.
16003 (vectorizable_reduction): Likewise.
16004 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16005 (vect_is_simple_reduction): Deal with chains not detected
16006 as SLP reduction chain, specifically not properly associated
16007 chains containing a mix of plus/minus.
16008 (get_reduction_op): Remove.
16009 (get_initial_defs_for_reduction): Simplify, pass in whether
16010 this is a reduction chain, pass in the SLP node for the PHIs.
16011 (vect_create_epilog_for_reduction): Get the SLP instance as
16013 (vectorizable_reduction): Get the SLP instance as arg.
16014 During analysis remember the SLP node with the PHIs in the
16015 instance. Simplify getting at the vectorized reduction PHIs.
16016 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16017 through SLP instance.
16018 (vect_slp_analyze_operations): Likewise.
16019 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16020 (vect_transform_stmt): Likewise.
16022 2017-07-20 Tom de Vries <tom@codesourcery.com>
16024 PR tree-optimization/81489
16025 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16026 read of phi arg location to before loop that modifies phi.
16028 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16030 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16033 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16035 PR middle-end/81331
16036 * except.c (execute): Fix ordering issue.
16038 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16040 PR rtl-optimization/81423
16041 * combine.c (make_compound_operation_int): Don't try to optimize
16042 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16044 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16046 PR rtl-optimization/81423
16047 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16048 with a constant that is -1 in the truncated to mode.
16050 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16052 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16053 (determine_unlikely_bbs): ... here.
16054 * predict.h (propagate_unlikely_bbs_forward): Declare.
16055 * cfgexpand.c (pass_expand::execute): Use it.
16056 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16058 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16059 propagate_unlikely_bbs_forward.
16061 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16063 PR middle-end/81331
16064 * except.c (maybe_add_nop_after_section_switch): New function.
16067 2017-07-19 Tom de Vries <tom@codesourcery.com>
16069 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16071 2017-07-19 Tom de Vries <tom@codesourcery.com>
16073 * gimple.h (gimple_phi_arg): Make assert more strict.
16075 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16077 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16078 * config/rs6000/mmintrin.h: New file.
16079 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16081 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16083 PR tree-optimization/81346
16084 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16086 2017-07-19 Tom de Vries <tom@codesourcery.com>
16088 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16090 2017-07-19 Tom de Vries <tom@codesourcery.com>
16092 * config/nvptx/nvptx-modes.def: Add V2DImode.
16093 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16094 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16095 (nvptx_output_mov_insn): Handle lack of mov.b128.
16096 (nvptx_print_operand): Handle 'H' and 'L' codes.
16097 (nvptx_vector_mode_supported): Allow V2DImode.
16098 (nvptx_preferred_simd_mode): New function.
16099 (nvptx_data_alignment): New function.
16100 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16101 nvptx_preferred_simd_mode.
16102 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16104 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16106 2017-07-19 Tom de Vries <tom@codesourcery.com>
16108 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16109 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16110 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16111 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16112 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16113 (mov<VECIM>_insn): New define_insn.
16114 (define_expand "mov<VECIM>): New define_expand.
16116 2017-07-19 Tom de Vries <tom@codesourcery.com>
16118 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16120 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16122 PR tree-optimization/81346
16123 * fold-const.h (fold_div_compare, range_check_type): Declare.
16124 * fold-const.c (range_check_type): New function.
16125 (build_range_check): Use range_check_type.
16126 (fold_div_compare): No longer static, rewritten into
16127 a match.pd helper function.
16128 (fold_comparison): Don't call fold_div_compare here.
16129 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16130 as helper function.
16132 2017-07-19 Nathan Sidwell <nathan@acm.org>
16134 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16135 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16136 * tree.c (find_decls_types_r, verify_type): Use
16137 TYPE_{MIN,MAX}_VALUE_RAW.
16138 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16139 (hash_tree): Likewise.
16140 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16142 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16145 2017-07-18 Tom de Vries <tom@codesourcery.com>
16147 PR middle-end/81464
16148 * omp-expand.c (expand_omp_for_static_chunk): Handle
16149 equal-argument loop exit phi.
16151 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16154 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16155 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16156 operand 2 predicate.
16157 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16158 operand 2 predicate.
16159 (ror,rol -> rorx splitters): Use const_int_operand as
16160 operand 2 predicate.
16162 2017-06-18 Richard Biener <rguenther@suse.de>
16164 PR tree-optimization/81410
16165 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16166 the gap in the ! slp_perm SLP case after each group.
16168 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16170 PR middle-end/81463
16171 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16174 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16176 PR middle-end/81462
16177 * predict.c (set_even_probabilities): Cleanup; do not affect
16178 probabilities that are already known.
16179 (combine_predictions_for_bb): Call even when count is set.
16181 2017-07-18 Nathan Sidwell <nathan@acm.org>
16183 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16186 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16189 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16190 optimization for loop niter analysis.
16192 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16195 * config/avr/avr.c (avr_optimize_casesi): Don't use
16196 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16198 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16200 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16201 body_cost_vec from _vect_peel_extended_info.
16202 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16203 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16206 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16208 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16210 2017-07-18 Richard Biener <rguenther@suse.de>
16212 PR tree-optimization/80620
16213 PR tree-optimization/81403
16214 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16215 info when re-using a VN table entry.
16217 2017-07-18 Richard Biener <rguenther@suse.de>
16219 PR tree-optimization/81418
16220 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16221 vectype_in. Verify that with lane-reducing reduction operations
16222 we have a single def-use cycle.
16224 2017-07-17 Carl Love <cel@us.ibm.com>
16226 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16228 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16229 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16230 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16231 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16232 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16233 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16234 VMULOSW): New enum "unspec" values.
16235 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16236 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16237 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16238 altivec_vmulosw): New patterns.
16239 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16240 VMULOSW): Add definitions.
16241 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16243 * config/alpha/alpha.c: Include predict.h.
16245 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16247 * tree-vrp.c (compare_assert_loc): Fix comparison function
16248 to return predictable results.
16250 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16252 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16254 (subdi3): Likewise.
16255 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16256 * doc/invoke.texi (mexpand-adddi): Update text.
16258 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16260 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16261 that also clobbers the CC register. The old expand code is moved
16263 (*arc_clzsi2): ... here.
16264 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16265 the CC register. The old expand code is moved to ...
16266 (arc_ctzsi2): ... here.
16268 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16270 * config/arc/arc.opt (mindexed-loads): Use initial value
16271 TARGET_INDEXED_LOADS_DEFAULT.
16272 (mauto-modify-reg): Use initial value
16273 TARGET_AUTO_MODIFY_REG_DEFAULT.
16274 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16275 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16276 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16277 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16279 2017-07-17 Martin Liska <mliska@suse.cz>
16282 * opts.c (finish_options): Do not allow -fgnu-tm
16283 w/ -fsanitize={kernel-,}address. Say sorry.
16285 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16288 * tree-loop-distribution.c (classify_partition): Only assert on
16289 numer of iterations.
16290 (merge_dep_scc_partitions): Delete prameter. Update function call.
16291 (distribute_loop): Remove code handling loop with unknown niters.
16292 (pass_loop_distribution::execute): Skip loop with unknown niters.
16294 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16297 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16298 function sort_partitions_by_post_order.
16300 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16302 PR tree-optimization/81374
16303 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16304 the max index of basic blocks, rather than number of basic blocks.
16306 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16308 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16310 (arc_legitimate_pic_operand_p): Likewise.
16311 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16313 (arc_needs_pcl_p): Likewise.
16314 (arc_legitimate_pc_offset_p): Likewise.
16315 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16316 function is also used in constrains.md.
16317 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16318 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16319 PLUS. Only return true/false in known cases, otherwise assert.
16320 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16321 is already called in arc_legitimate_constant_p.
16322 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16324 (LEGITIMATE_PIC_OPERAND_P): Use
16325 arc_raw_symbolic_reference_mentioned_p function.
16326 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16331 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16332 Andrew Burgess <andrew.burgess@embecosm.com>
16334 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16335 (arc_return_address_register): New function.
16336 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16337 (arc_handle_fndecl_attribute): Add naked attribute.
16338 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16339 (TARGET_WARN_FUNC_RETURN): Likewise.
16340 (arc_allocate_stack_slots_for_args): New function.
16341 (arc_warn_func_return): Likewise.
16342 (machine_function): Change type fn_type.
16343 (arc_compute_function_type): Consider new naked function type,
16344 change function return type.
16345 (arc_must_save_register): Adapt to handle new
16346 arc_compute_function_type's return type.
16347 (arc_expand_prologue): Likewise.
16348 (arc_expand_epilogue): Likewise.
16349 (arc_return_address_regs): Delete.
16350 (arc_return_address_register): New function.
16351 (arc_epilogue_uses): Use above function.
16352 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16353 (arc_function_type): Change encoding, add naked type.
16354 (ARC_INTERRUPT_P): Change to handle the new encoding.
16355 (ARC_FAST_INTERRUPT_P): Likewise.
16356 (ARC_NORMAL_P): Define.
16357 (ARC_NAKED_P): Likewise.
16358 (arc_compute_function_type): Delete prototype.
16359 * config/arc/arc.md (in_ret_delay_slot): Use
16360 arc_return_address_register function.
16361 (simple_return): Likewise.
16362 (p_return_i): Likewise.
16364 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16366 PR tree-optimization/81428
16367 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16368 can't be built for those types.
16370 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16372 Remove stuff dead since r239246.
16374 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16375 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16376 (avr_inform_devices): Remove dead stuff.
16378 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16380 * config/arm/arm_neon.h: Fix softp typo.
16382 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16384 PR tree-optimization/81365
16385 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16386 aggregate moves onto bb predecessor edges, make sure there are no
16387 loads that could alias the lhs in between the start of bb and the
16390 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16393 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16394 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16395 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16397 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16399 PR tree-optimization/81396
16400 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16401 (init_symbolic_number): Initialize it to 1.
16402 (perform_symbolic_merge): Add n_ops from both operands into the new
16404 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16405 without base_addr as useless if they need more than one operation.
16406 (bswap_replace): Handle !bswap case for NULL base_addr.
16408 2017-07-17 Tom de Vries <tom@codesourcery.com>
16411 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16414 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16416 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16417 conditional builtin define __FIX_LEON3FT_B2BST.
16419 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16421 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16422 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16425 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16427 PR rtl-optimization/81424
16428 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16429 to remove potential trapping from operands if -fnon-call-exceptions.
16431 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16433 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16434 profile_proability for scalling.
16435 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16437 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16439 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16441 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16443 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16444 fixpoint arithmetics.
16446 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16448 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16449 fixpoint arithmetics.
16451 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16453 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16454 fixpoint arithmetics.
16456 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16458 * profile-count.h (profile_probability::from_reg_br_prob_note,
16459 profile_probability::to_reg_br_prob_note): New functions.
16460 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16461 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16462 * predict.c (probability_reliable_p): Update.
16463 (edge_probability_reliable_p): Update.
16464 (br_prob_note_reliable_p): Update.
16465 (invert_br_probabilities): Update.
16466 (add_reg_br_prob_note): New function.
16467 (combine_predictions_for_insn): Update.
16468 * asan.c (asan_clear_shadow): Update.
16469 * cfgbuild.c (compute_outgoing_frequencies): Update.
16470 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16471 (update_br_prob_note): Update.
16472 (rtl_verify_edges): Update.
16473 (purge_dead_edges): Update.
16474 (fixup_reorder_chain): Update.
16475 * emit-rtl.c (try_split): Update.
16476 * ifcvt.c (cond_exec_process_insns): Update.
16477 (cond_exec_process_if_block): Update.
16478 (dead_or_predicable): Update.
16479 * internal-fn.c (expand_addsub_overflow): Update.
16480 (expand_neg_overflow): Update.
16481 (expand_mul_overflow): Update.
16482 * loop-doloop.c (doloop_modify): Update.
16483 * loop-unroll.c (compare_and_jump_seq): Update.
16484 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16485 * predict.h: Update.
16486 * reorg.c (mostly_true_jump): Update.
16488 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16489 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16490 * config/arc/arc.c: (emit_unlikely_jump): Update.
16491 * config/arm/arm.c: (emit_unlikely_jump): Update.
16492 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16493 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16494 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16495 (ix86_print_operand): Update.
16496 (ix86_split_fp_branch): Update.
16497 (predict_jump): Update.
16498 * config/ia64/ia64.c (ia64_print_operand): Update.
16499 * config/mmix/mmix.c (mmix_print_operand): Update.
16500 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16501 (rs6000_expand_split_stack_prologue): Update.
16502 * config/rs6000/rs6000.c: Update.
16503 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16504 (s390_expand_vec_movstr): Update.
16505 (s390_expand_cs_tdsi): Update.
16506 (s390_expand_split_stack_prologue): Update.
16507 * config/sh/sh.c (sh_print_operand): Update.
16508 (expand_cbranchsi4): Update.
16509 (expand_cbranchdi4): Update.
16510 * config/sparc/sparc.c (output_v9branch): Update.
16511 * config/spu/spu.c (get_branch_target): Update.
16512 (ea_load_store_inline): Update.
16513 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16514 * config/tilepro/tilepro.c: Update.
16516 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16518 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16519 (gimplify_save_expr): Likewise.
16521 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16523 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16525 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16527 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16529 * timevar.def (TV_IPA_FNSUMMARY): Define.
16531 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16533 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16534 to back store errata sensitive sequence from being generated.
16535 (sqrtdf2_fix): Likewise.
16537 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16539 * tree-ssa-threadupdate.c (compute_path_counts,
16540 update_joiner_offpath_counts): Use profile_probability.
16542 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16545 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16547 * config/arm/arm-c.c (arm_cpu_builtins): Define
16548 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16550 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16552 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16553 array entries to represent __ieee128 versions of the
16554 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16555 scalar_extract_sig, and scalar_insert_exp built-in functions.
16556 (altivec_resolve_overloaded_builtin): Add special case handling
16557 for the __builtin_scalar_insert_exp function, as represented by
16558 the P9V_BUILTIN_VEC_VSIEDP constant.
16559 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16560 exponent support for __ieee128 argument.
16561 (VSESQP): Add scalar extract signature support for __ieee128
16563 (VSTDCNQP): Add scalar test negative support for __ieee128
16565 (VSIEQP): Add scalar insert exponent support for __int128 argument
16566 with __ieee128 result.
16567 (VSIEQPF): Add scalar insert exponent support for __ieee128
16568 argument with __ieee128 result.
16569 (VSTDCQP): Add scalar test data class support for __ieee128
16571 (VSTDCNQP): Add overload support for scalar test negative with
16572 __ieee128 argument.
16573 (VSTDCQP): Add overload support for scalar test data class
16574 __ieee128 argument.
16575 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16576 UNSPEC_VSX_SXSIGDP.
16577 (UNSPEC_VSX_SIEXPQP): New constant.
16578 (xsxexpqp): New insn for VSX scalar extract exponent quad
16580 (xsxsigqp): New insn for VSX scalar extract significand quad
16582 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16583 precision with floating point argument.
16584 (xststdcqp): New expand for VSX scalar test data class quad
16586 (xststdcnegqp): New expand for VSX scalar test negative quad
16588 (xststdcqp): New insn to match expansions for VSX scalar test data
16589 class quad precision and VSX scalar test negative quad precision.
16590 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16591 special case operand checking to enforce that second operand of
16592 VSX scalar test data class with quad precision argument is a 7-bit
16594 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16595 prototypes and descriptions of __ieee128 versions of
16596 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16597 scalar_test_data_class, and scalar_test_neg built-in functions.
16599 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16601 PR tree-optimization/81162
16602 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16603 replace a negate with an add.
16605 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16607 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16609 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16611 * config/arm/arm-c.c (arm_cpu_builtins): Define
16612 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16614 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16616 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16617 (armv8-r): Set ARM Cortex-R52 as default CPU.
16618 * config/arm/arm-tables.opt: Regenerate.
16619 * config/arm/arm-tune.md: Regenerate.
16620 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16622 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16623 extension for -mcpu=cortex-r52.
16625 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16627 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16628 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16629 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16630 (fp-armv8): Define it as FP_ARMv8 only.
16631 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16632 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16634 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16635 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16636 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16637 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16638 than TARGET_FPU_ARMV8.
16639 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16640 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16641 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16643 * config/arm/neon.md (neon_vrint): Likewise.
16644 (neon_vcvt): Likewise.
16645 (neon_<fmaxmin_op><mode>): Likewise.
16646 (<fmaxmin><mode>3): Likewise.
16647 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16648 * config/arm/predicates.md (arm_cond_move_operator): Check against
16649 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16651 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16653 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16654 to top of function.
16656 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16658 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16659 loop in comment with memset.
16661 2017-07-14 Martin Liska <mliska@suse.cz>
16663 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16664 * dwarf2out.c (is_java): Remove the function.
16665 (output_pubname): Remove usage of the function.
16666 (lower_bound_default): Remove usage of DW_LANG_Java.
16667 (gen_compile_unit_die): Likewise.
16668 * gcc.c: Remove compiler defaults for .java and .zip files.
16669 * gimple-expr.c (remove_suffix): Change as there's no longer
16670 extension than 4-letter one.
16671 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16672 (gimplify_save_expr): Likewise.
16673 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16674 as it's possible even for other languages than Java.
16675 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16676 * lto-opts.c (lto_write_options): Remove reference to Java.
16677 * opts.c (strip_off_ending): Update file extension handling.
16678 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16679 * tree-eh.c (lower_resx): Likewise.
16680 * tree.c (free_lang_data_in_type): Remove dead code.
16681 (find_decls_types_r): Likewise.
16682 (build_common_builtin_nodes): Remove Java from a comment.
16683 (verify_type): Remove dead code.
16684 * varasm.c (assemble_external): Remove Java from a comment.
16686 2017-07-14 Martin Liska <mliska@suse.cz>
16688 * opts.c (finish_options): Add quotes.
16689 (common_handle_option): Likewise.
16691 2017-07-14 Martin Liska <mliska@suse.cz>
16693 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16694 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16695 Remove N_SO_PASCAL.
16696 * dwarf2out.c (lower_bound_default): Do not handle
16698 (gen_compile_unit_die): Likewise.
16699 * gcc.c: Remove default extension binding for GNU Pascal.
16700 * stmt.c: Remove Pascal language from a comment.
16701 * xcoffout.c: Likewise.
16703 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16706 * diagnostic-show-locus.c (fixit_cmp): New function.
16707 (layout::layout): Sort m_fixit_hints.
16708 (column_range::column_range): Assert that the values are valid.
16709 (struct char_span): New struct.
16710 (correction::overwrite): New method.
16711 (struct source_line): New struct.
16712 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16713 calls in terms of classes source_line and char_span, and
16714 correction::overwrite.
16715 (selftest::test_overlapped_fixit_printing_2): New function.
16716 (selftest::diagnostic_show_locus_c_tests): Call it.
16718 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16720 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16721 early if there is no lhs.
16723 2017-07-13 Martin Liska <mliska@suse.cz>
16725 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16726 (gen_reference_type_die): Likewise.
16727 * stor-layout.c: Remove Pascal-related comment.
16729 2017-07-13 Martin Liska <mliska@suse.cz>
16731 * opts.c (finish_options): Add quotes to error messages.
16732 (parse_sanitizer_options): Likewise.
16734 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16736 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16738 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16740 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16742 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16744 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16746 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16748 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16751 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16752 provides the hardware capability bits, define the macro
16753 __BUILTIN_CPU_SUPPORTS__.
16754 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16755 if GLIBC does not provide the hardware capability bits. Add a
16756 gcc_unreachable call if the built-in cpu function is neither
16757 __builtin_cpu_is nor __builtin_cpu_supports.
16758 (rs6000_get_function_versions_dispatcher): Change the warning
16759 that an old GLIBC is used which does not export the capability
16760 bits to be an error.
16761 * doc/extend.texi (target_clones attribute): Document the
16762 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16763 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16764 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16765 the macros defined by GCC if the newer GLIBC is available.
16767 2017-07-12 Jeff Law <law@redhat.com>
16769 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16770 remaining includes slightly.
16771 * config/riscv/riscv-builtins.c: Include profile-count.h.
16773 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16776 * config/avr/avr.c (avr_set_current_function): In diagnostic
16777 messages: Quote keywords and (parts of) identifiers.
16778 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16781 2017-07-12 Carl Love <cel@us.ibm.com>
16783 * config/rs6000/rs6000-c.c: Add support for built-in functions
16784 vector bool char vec_revb (vector bool char);
16785 vector bool short vec_revb (vector short char);
16786 vector bool int vec_revb (vector bool int);
16787 vector bool long long vec_revb (vector bool long long);
16788 * doc/extend.texi: Update the built-in documentation file for the
16789 new built-in functions.
16791 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16793 * config/s390/s390.md: Remove movcc splitter.
16795 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16797 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16798 load/store on condition.
16800 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16803 * config/avr/avr.c (avr_encode_section_info)
16804 [progmem && !TREE_READONLY]: Error if progmem object needs
16807 2017-07-11 Michael Collison <michael.collison@arm.com>
16809 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16812 2017-07-11 Carl Love <cel@us.ibm.com>
16814 * config/rs6000/rs6000-c.c: Add support for builtins
16815 vector unsigned int vec_parity_lsbb (vector signed int);
16816 vector unsigned int vec_parity_lsbb (vector unsigned int);
16817 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16818 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16819 vector unsigned long long vec_parity_lsbb (vector signed long long);
16820 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16821 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16822 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16823 * doc/extend.texi: Update the built-in documentation file for the
16824 new built-in functions.
16826 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16828 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16829 (layout::m_primary_loc): New field.
16830 (layout::layout): Initialize new field. Move location filtering
16831 logic from here to...
16832 (layout::maybe_add_location_range): ...this new method. Add
16833 support for filtering to just the lines already specified by other
16835 (layout::will_show_line_p): New method.
16836 (gcc_rich_location::add_location_if_nearby): New method.
16837 (selftest::test_add_location_if_nearby): New test function.
16838 (selftest::diagnostic_show_locus_c_tests): Call it.
16839 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16842 2017-07-11 Tom de Vries <tom@codesourcery.com>
16844 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16845 (bb_first_real_insn): New function.
16846 (nvptx_single): Add extra initialization of broadcasted condition
16849 2017-07-11 Nathan Sidwell <nathan@acm.org>
16851 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16853 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16855 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16856 Binutils doc as TEXI will mess them up.
16857 * doc/invoke.texi (AVR Options): Same here.
16859 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16861 * config/sparc/sparc.opt (mfix-ut700): New option.
16862 (mfix-gr712rc): Likewise.
16863 (sparc_fix_b2bst): New variable.
16864 * doc/invoke.texi (SPARC options): Document them.
16865 (ARM options): Fix warnings.
16866 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16867 instructions to prevent sequences that can trigger the store-store
16868 errata for certain LEON3FT processors.
16869 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16870 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16871 * config/sparc/sparc.md (fix_b2bst): New attribute.
16872 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16874 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16877 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16879 (*rsqrtsf2_sse): Ditto.
16881 (div<mode>3): Macroize insn from divdf3 and divsf3
16882 using MODEF mode iterator.
16884 2017-07-10 Martin Sebor <msebor@redhat.com>
16886 PR tree-optimization/80397
16887 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16888 instead of testing for equality to INTEGER_TYPE.
16890 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16892 * config.gcc: Remove uclibc from arc target spec.
16894 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16896 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16898 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16901 * lto-wrapper.c (remove_option): New function.
16902 (merge_and_complain): Merge PIC/PIE options more realistically.
16904 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16906 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16910 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16911 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16912 * config.in: Regenerate.
16913 * configure: Regenerate.
16914 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16915 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16916 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16917 (TARGET_GASISR_PROLOGUES): ...target mask.
16918 * common/config/avr/avr-common.c
16919 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16920 Set -mgas-isr-prologues.
16921 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16922 INSERT_PASS_BEFORE for it.
16923 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16924 * config/avr/avr.c (avr_option_override)
16925 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16926 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16927 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16928 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16929 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16931 (make_avr_pass_pre_proep): New function.
16932 (emit_push_sfr) <treg>: Add argument to function and use it
16933 instead of TMP_REG.
16934 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16935 and set machine->gasisr.yes.
16936 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16937 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16938 __gcc_isr.n_pushed to .L__stack_usage.
16939 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16940 (avr_asm_final_postscan_insn): ...this new static function.
16941 * config/avr/avr.h (machine_function)
16942 <is_no_gccisr, use_L__stack_usage>: New fields.
16943 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16944 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16945 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16946 (gasisr, *gasisr): New expander and insn.
16947 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16948 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16949 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16951 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16953 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16956 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16958 Move jump-tables out of .text again.
16961 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16962 (ASM_OUTPUT_ADDR_VEC): New function.
16963 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16964 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16965 INSN_ADDRESSes as asm comment.
16966 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16967 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16968 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16969 * config/avr/avr.md (*tablejump): Adjust comment.
16970 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16971 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16973 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16974 (avr_output_addr_vec): New proto.
16975 (avr_log_t) <insn_addresses>: New field.
16977 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16980 * config/i386/i386.c (ix86_function_arg_advance): Set
16981 outgoing_args_on_stack to true if there are outgoing arguments
16983 (ix86_function_arg): Likewise.
16984 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16985 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16986 * config/i386/i386.h (machine_function): Add
16987 outgoing_args_on_stack.
16989 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16991 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16992 supporting pthreds.
16993 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16995 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16997 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16998 (REAL_H): Remove $(MACHMODE_H).
16999 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17001 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17002 $(MACHMODE_H) and double-int.h.
17003 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17005 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17008 2017-07-07 Andrew Pinski <apinski@cavium.com>
17010 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17011 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17013 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17015 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17016 Add warning if GCC was not configured to link against a GLIBC that
17017 exports the hardware capability bits.
17018 (make_resolver_func): Make resolver function private and not a
17019 COMDAT function. Create the name with clone_function_name instead
17020 of make_unique_name.
17023 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17024 correct operand in doing the split.
17026 2017-07-07 Carl Love <cel@us.ibm.com>
17028 * config/rs6000/rs6000-c: Add support for built-in function
17029 vector unsigned short vec_pack_to_short_fp32 (vector float,
17031 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17032 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17033 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17034 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17035 (convert_4f32_8i16): Add define_expand.
17036 * doc/extend.texi: Update the built-in documentation file for the
17037 new built-in function.
17039 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17041 * config/sparc/m8.md: New file.
17042 * config/sparc/sparc.md: Include m8.md.
17044 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17046 * config/sparc/sparc.opt: New option -mvis4b.
17047 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17048 (sparc_option_override): Handle VIS4B.
17049 (enum sparc_builtins): Define
17050 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17051 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17052 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17053 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17054 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17055 (check_constant_argument): New function.
17056 (sparc_vis_init_builtins): Define builtins
17057 __builtin_vis_dictunpack{8,16,32},
17058 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17059 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17060 __builtin_vis_fpcmpde{8,16,32}shl and
17061 __builtin_vis_fpcmpur{8,16,32}shl.
17062 (sparc_expand_builtin): Check that the constant operands to
17063 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17064 constant and in range.
17065 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17067 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17068 (SPARC_IMM5_P): Likewise.
17069 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17070 (enabled): Handle vis4b.
17071 (UNSPEC_DICTUNPACK): New unspec.
17072 (UNSPEC_FPCMPSHL): Likewise.
17073 (UNSPEC_FPUCMPSHL): Likewise.
17074 (UNSPEC_FPCMPDESHL): Likewise.
17075 (UNSPEC_FPCMPURSHL): Likewise.
17076 (cpu_feature): New CPU feature `vis4b'.
17077 (dictunpack{8,16,32}): New insns.
17078 (FPCSMODE): New mode iterator.
17079 (fpcscond): New code iterator.
17080 (fpcsucond): Likewise.
17081 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17082 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17083 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17084 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17085 * config/sparc/constraints.md: Define constraints `q' for unsigned
17086 2-bit integer constants and `t' for unsigned 5-bit integer
17088 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17090 (imm5_operand_dictunpack16): Likewise.
17091 (imm5_operand_dictunpack32): Likewise.
17092 (imm2_operand): Likewise.
17093 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17094 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17095 ditunpack* and fpcmp*shl builtins.
17097 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17099 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17100 * config.in: Add HAVE_AS_SPARC6 define.
17101 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17103 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17105 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17106 (CPP_CPU_SPEC): Handle m8.
17107 (ASM_CPU_SPEC): Likewise.
17108 * config/sparc/sparc-opts.h (enum processor_type): Add
17110 * config/sparc/sparc.c (m8_costs): New struct.
17111 (sparc_option_override): Handle TARGET_CPU_m8.
17112 (sparc32_initialize_trampoline): Likewise.
17113 (sparc64_initialize_trampoline): Likewise.
17114 (sparc_issue_rate): Likewise.
17115 (sparc_register_move_cost): Likewise.
17116 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17117 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17118 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17119 (CPP_CPU_SPEC): Handle M8.
17120 (ASM_CPU_SPEC): Likewise.
17121 (AS_M8_FLAG): Define.
17122 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17123 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17124 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17126 * configure: Regenerate.
17127 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17130 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17132 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17134 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17135 ("*movdi_insn_sp32"): Do not set v3pipe.
17136 ("*movsi_insn"): Likewise.
17137 ("*movdi_insn_sp64"): Likewise.
17138 ("*movsf_insn"): Likewise.
17139 ("*movdf_insn_sp32"): Likewise.
17140 ("*movdf_insn_sp64"): Likewise.
17141 ("*zero_extendsidi2_insn_sp64"): Likewise.
17142 ("*sign_extendsidi2_insn"): Likewise.
17143 ("*mov<VM32:mode>_insn"): Likewise.
17144 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17145 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17146 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17147 ("<vlop:code><VL:mode>3"): Likewise.
17148 ("*not_<vlop:code><VL:mode>3"): Likewise.
17149 ("*nand<VL:mode>_vis"): Likewise.
17150 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17151 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17152 ("one_cmpl<VL:mode>2"): Likewise.
17153 ("faligndata<VM64:mode>_vis"): Likewise.
17154 ("alignaddrsi_vis"): Likewise.
17155 ("alignaddrdi_vis"): Likweise.
17156 ("alignaddrlsi_vis"): Likewise.
17157 ("alignaddrldi_vis"): Likewise.
17158 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17159 ("bmaskdi_vis"): Likewise.
17160 ("bmasksi_vis"): Likewise.
17161 ("bshuffle<VM64:mode>_vis"): Likewise.
17162 ("cmask8<P:mode>_vis"): Likewise.
17163 ("cmask16<P:mode>_vis"): Likewise.
17164 ("cmask32<P:mode>_vis"): Likewise.
17165 ("pdistn<P:mode>_vis"): Likewise.
17166 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17168 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17170 * config/sparc/sparc.md ("subtype"): New insn attribute.
17171 ("*wrgsr_sp64"): Set insn subtype.
17172 ("*rdgsr_sp64"): Likewise.
17173 ("alignaddrsi_vis"): Likewise.
17174 ("alignaddrdi_vis"): Likewise.
17175 ("alignaddrlsi_vis"): Likewise.
17176 ("alignaddrldi_vis"): Likewise.
17177 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17178 ("fexpand_vis"): Likewise.
17179 ("fpmerge_vis"): Likewise.
17180 ("faligndata<VM64:mode>_vis"): Likewise.
17181 ("bshuffle<VM64:mode>_vis"): Likewise.
17182 ("cmask8<P:mode>_vis"): Likewise.
17183 ("cmask16<P:mode>_vis"): Likewise.
17184 ("cmask32<P:mode>_vis"): Likewise.
17185 ("fchksm16_vis"): Likewise.
17186 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17187 ("fmean16_vis"): Likewise.
17188 ("fp<plusminus_insn>64_vis"): Likewise.
17189 ("<plusminus_insn>v8qi3"): Likewise.
17190 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17191 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17192 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17193 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17194 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17195 ("*movqi_insn"): Likewise.
17196 ("*movhi_insn"): Likewise.
17197 ("*movsi_insn"): Likewise.
17198 ("movsi_pic_gotdata_op"): Likewise.
17199 ("*movdi_insn_sp32"): Likewise.
17200 ("*movdi_insn_sp64"): Likewise.
17201 ("movdi_pic_gotdata_op"): Likewise.
17202 ("*movsf_insn"): Likewise.
17203 ("*movdf_insn_sp32"): Likewise.
17204 ("*movdf_insn_sp64"): Likewise.
17205 ("*zero_extendhisi2_insn"): Likewise.
17206 ("*zero_extendqihi2_insn"): Likewise.
17207 ("*zero_extendqisi2_insn"): Likewise.
17208 ("*zero_extendqidi2_insn"): Likewise.
17209 ("*zero_extendhidi2_insn"): Likewise.
17210 ("*zero_extendsidi2_insn_sp64"): Likewise.
17211 ("ldfsr"): Likewise.
17212 ("prefetch_64"): Likewise.
17213 ("prefetch_32"): Likewise.
17214 ("tie_ld32"): Likewise.
17215 ("tie_ld64"): Likewise.
17216 ("*tldo_ldub_sp32"): Likewise.
17217 ("*tldo_ldub1_sp32"): Likewise.
17218 ("*tldo_ldub2_sp32"): Likewise.
17219 ("*tldo_ldub_sp64"): Likewise.
17220 ("*tldo_ldub1_sp64"): Likewise.
17221 ("*tldo_ldub2_sp64"): Likewise.
17222 ("*tldo_ldub3_sp64"): Likewise.
17223 ("*tldo_lduh_sp32"): Likewise.
17224 ("*tldo_lduh1_sp32"): Likewise.
17225 ("*tldo_lduh_sp64"): Likewise.
17226 ("*tldo_lduh1_sp64"): Likewise.
17227 ("*tldo_lduh2_sp64"): Likewise.
17228 ("*tldo_lduw_sp32"): Likewise.
17229 ("*tldo_lduw_sp64"): Likewise.
17230 ("*tldo_lduw1_sp64"): Likewise.
17231 ("*tldo_ldx_sp64"): Likewise.
17232 ("*mov<VM32:mode>_insn"): Likewise.
17233 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17234 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17236 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17238 * config/sparc/sparc.md ("type"): New insn type viscmp.
17239 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17241 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17242 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17243 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17244 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17246 ("n7_vis_logical_11cycle"): Likewise.
17247 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17248 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17249 * config/sparc/niagara.md ("niag_vis"): Likewise.
17250 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17251 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17253 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17255 * config/sparc/sparc.md: New instruction type `bmask'.
17256 (bmaskdi_vis): Use the `bmask' type.
17257 (bmasksi_vis): Likewise.
17258 * config/sparc/ultra3.md (us3_array): Likewise.
17259 * config/sparc/niagara7.md (n7_array): Likewise.
17260 * config/sparc/niagara4.md (n4_array): Likewise.
17261 * config/sparc/niagara2.md (niag2_vis): Likewise.
17262 (niag3_vis): Likewise.
17263 * config/sparc/niagara.md (niag_vis): Likewise.
17265 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17267 * ipa-comdats.c: Remove optimize check from gate.
17268 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17269 for functions not optimized.
17270 (ipa_fn_summary_read): Skip optimize check.
17271 (ipa_fn_summary_write): Likewise.
17272 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17274 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17276 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17277 (check_callers): Give up on caller which is not optimized.
17278 (inline_small_functions): Likewise.
17279 (ipa_inline): Do not give up when not optimizing.
17280 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17281 away unoptimizes cdtors.
17282 (whole_program_function_and_variable_visibility): Do
17283 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17284 * ipa.c (process_references): Do not check optimize.
17285 (symbol_table::remove_unreachable_nodes): Update optimize check.
17286 (set_writeonly_bit): Update optimize check.
17287 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17288 (pass_ipa_single_use::gate): Remove.
17290 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17292 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17293 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17294 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17295 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17296 permute_load, permute_store, adjust_extract, adjust_splat,
17297 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17298 replace_swap_with_copy, dump_swap_insn_table,
17299 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17300 recombine_lvx_pattern, recombine_stvx_pattern,
17301 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17302 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17303 to file rs6000-p8swap.c.
17304 * config/rs6000/rs6000-p8swap.c: New file.
17305 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17306 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17307 and rs6000*-*-* targets.
17309 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17311 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17313 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17315 * lto-wrapper.c (merge_and_complain): Do not merge
17316 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17317 fsigned_zeros, ftrapping_math, fwrapv.
17318 (append_compiler_options): Do not track these options.
17319 (append_linker_options): Likewie
17321 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17323 * cgraphunit.c (cgraph_node::finalize_function): When
17324 !flag_toplevel_reorde set no_reorder flag.
17325 (varpool_node::finalize_decl): Likewise.
17326 (symbol_table::compile): Drop no toplevel reorder path.
17328 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17330 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17331 edges; zero probability is not better than uninitialized.
17333 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17335 * asan.h (asan_sanitize_allocas_p): Declare.
17336 * asan.c (asan_sanitize_allocas_p): New function.
17337 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17338 (handle_builtin_alloca): Likewise.
17339 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17340 if !asan_sanitize_allocas_p.
17341 * params.def (asan-instrument-allocas): Add new option.
17342 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17343 * opts.c (common_handle_option): Disable allocas sanitization for
17346 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17348 * asan.c: Include gimple-fold.h.
17349 (get_last_alloca_addr): New function.
17350 (handle_builtin_stackrestore): Likewise.
17351 (handle_builtin_alloca): Likewise.
17352 (asan_emit_allocas_unpoison): Likewise.
17353 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17354 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17355 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17356 (instrument_builtin_call): Pass gimple iterator to
17357 get_mem_refs_of_builtin_call.
17358 (last_alloca_addr): New global.
17359 * asan.h (asan_emit_allocas_unpoison): Declare.
17360 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17361 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17362 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17363 if function calls alloca.
17364 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17365 * gimple-fold.h (replace_call_with_value): Declare.
17366 * internal-fn.c: Include asan.h.
17367 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17368 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17370 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17372 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17373 (C_SELFTEST_FLAGS): New.
17374 (CPP_SELFTEST_FLAGS): New.
17375 (SELFTEST_DEPS): New, from deps of s-selftest.
17376 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17377 (CPP_SELFTEST_DEPS): New.
17378 (selftest): Add dependency on s-selftest-c++.
17379 (s-selftest): Rename to...
17380 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17381 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17382 than SELFTEST_FLAGS.
17383 (selftest-gdb): Rename to...
17384 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17386 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17387 (selftest-valgrind): Rename to...
17388 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17390 (selftest-valgrind): Reintroduce as an alias for
17391 selftest-c-valgrind.
17392 (s-selftest-c++): New.
17393 (selftest-c++-gdb): New.
17394 (selftest-c++-valgrind): New.
17396 2017-07-06 Olivier Hainque <hainque@adacore.com>
17398 * gcc.c (process_command): When deciding if undefined variables
17399 should be ignored when processing specs, accept "gcc -v" as well.
17401 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17403 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17404 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17406 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17408 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17409 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17410 * config/arm/arm-tables.opt: Regenerate.
17411 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17413 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17415 2017-07-06 Carl Love <cel@us.ibm.com>
17417 * ChangeLog: Clean up from mid air collision
17419 2017-07-06 Carl Love <cel@us.ibm.com>
17421 * config/rs6000/rs6000-c.c: Add support for built-in functions
17422 vector signed int vec_subc (vector signed int, vector signed int);
17423 vector signed __int128 vec_subc (vector signed __int128,
17424 vector signed __int128);
17425 vector unsigned __int128 vec_subc (vector unsigned __int128,
17426 vector unsigned __int128);
17427 vector signed int vec_sube (vector signed int, vector signed int,
17428 vector signed int);
17429 vector unsigned int vec_sube (vector unsigned int,
17430 vector unsigned int,
17431 vector unsigned int);
17432 vector signed __int128 vec_sube (vector signed __int128,
17433 vector signed __int128,
17434 vector signed__int128);
17435 vector unsigned __int128 vec_sube (vector unsigned __int128,
17436 vector unsigned __int128,
17437 vector unsigned __int128);
17438 vector signed int vec_subec (vector signed int, vector signed int,
17439 vector signed int);
17440 vector unsigned int vec_subec (vector unsigned int,
17441 vector unsigned int,
17442 vector unsigned int);
17443 vector signed __int128 vec_subec (vector signed __int128,
17444 vector signed __int128,
17445 vector signed__int128);
17446 vector unsigned __int128 vec_subec (vector unsigned __int128,
17447 vector unsigned __int128,
17448 vector unsigned __int128);
17449 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17450 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17451 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17452 BU_ALTIVEC_OVERLOAD_X definitions.
17453 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17454 * doc/extend.texi: Update the built-in documentation file for the new
17455 built-in functions.
17457 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17460 * diagnostic-show-locus.c (layout::layout): Use start and finish
17461 spelling location for the start and finish of each range.
17462 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17463 Add unused aspect param.
17464 * input.c (expand_location_1): Add "aspect" param, and use it
17465 to access the correct part of the location.
17466 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17468 (expand_location_to_spelling_point): Likewise.
17469 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17470 param, and pass it to expand_location_1.
17472 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17474 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17475 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17476 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17477 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17478 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17479 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17480 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17481 _mm_maskz_getmant_ss): New intrinsics.
17482 (__builtin_ia32_getexpss128_mask): Changed to ...
17483 __builtin_ia32_getexpss128_round ... this.
17484 (__builtin_ia32_getexpsd128_mask): Changed to ...
17485 __builtin_ia32_getexpsd128_round ... this.
17486 * config/i386/i386-builtin-types.def
17487 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17488 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17489 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17490 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17491 __builtin_ia32_getmantss_mask_round): New builtins.
17492 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17493 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17494 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17495 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17496 * config/i386/sse.md
17497 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17498 avx512f_sgetexp<mode><mask_scalar_name>
17499 <round_saeonly_scalar_name> ... this.
17500 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17501 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17502 vgetexp<ssescalarmodesuffix>
17503 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17504 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17505 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17506 avx512f_vgetmant<mode><mask_scalar_name>
17507 <round_saeonly_scalar_name> ... this.
17508 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17509 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17510 vgetmant<ssescalarmodesuffix>
17511 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17512 %0<mask_scalar_operand4>, %1, %2
17513 <round_saeonly_scalar_mask_op4>, %3} ... this.
17514 * config/i386/subst.md (mask_scalar_operand4,
17515 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17516 round_saeonly_scalar_nimm_predicate): New subst attributes.
17518 2017-07-06 Julia Koval <julia.koval@intel.com>
17520 * config/i386/i386.c (ix86_erase_embedded_rounding):
17521 Remove code for old rounding pattern.
17523 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17525 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17527 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17529 * doc/sourcebuild.texi (Test Directives, Variants of
17530 dg-require-support): Add documentation for dg-require-stack-check.
17532 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17534 * config/i386/subst.md (mask_scalar, round_scalar,
17535 round_saeonly_scalar): New meta-templates.
17536 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17537 round_scalar_mask_operand3, round_scalar_mask_op3,
17538 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17539 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17540 round_saeonly_scalar_constraint,
17541 round_saeonly_scalar_prefix): New subst attribute.
17542 * config/i386/sse.md
17543 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17544 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17545 <round_scalar_name> ... this.
17546 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17547 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17548 <round_scalar_name> ... this.
17549 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17550 <sse>_vm<code><mode>3<mask_scalar_name>
17551 <round_saeonly_scalar_name> ... this.
17552 (v<plusminus_mnemonic><ssescalarmodesuffix>
17553 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17554 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17555 v<plusminus_mnemonic><ssescalarmodesuffix>
17556 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17557 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17558 (v<multdiv_mnemonic><ssescalarmodesuffix>
17559 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17560 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17561 v<multdiv_mnemonic><ssescalarmodesuffix>
17562 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17563 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17564 (v<maxmin_float><ssescalarmodesuffix>
17565 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17566 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17567 v<maxmin_float><ssescalarmodesuffix>
17568 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17569 %0<mask_scalar_operand3>, %1, %<iptr>2
17570 <round_saeonly_scalar_mask_op3>} ... this.
17572 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17574 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17575 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17577 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17578 Alan Hayward <alan.hayward@arm.com>
17579 David Sherwood <david.sherwood@arm.com>
17581 * combine.c (simplify_if_then_else): Remove "enum" before
17583 * compare-elim.c (can_eliminate_compare): Likewise.
17584 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17586 (aarch64_lookup_simd_builtin_type): Likewise.
17587 (aarch64_simd_builtin_type): Likewise.
17588 (aarch64_init_simd_builtin_types): Likewise.
17589 (aarch64_simd_expand_args): Likewise.
17590 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17592 (aarch64_reverse_mask): Likewise.
17593 (aarch64_simd_emit_reg_reg_move): Likewise.
17594 (aarch64_gen_adjusted_ldpstp): Likewise.
17595 (aarch64_ccmp_mode_to_code): Likewise.
17596 (aarch64_operands_ok_for_ldpstp): Likewise.
17597 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17598 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17600 (aarch64_min_divisions_for_recip_mul): Likewise.
17601 (aarch64_reassociation_width): Likewise.
17602 (aarch64_get_condition_code_1): Likewise.
17603 (aarch64_simd_emit_reg_reg_move): Likewise.
17604 (aarch64_simd_attr_length_rglist): Likewise.
17605 (aarch64_reverse_mask): Likewise.
17606 (aarch64_operands_ok_for_ldpstp): Likewise.
17607 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17608 (aarch64_gen_adjusted_ldpstp): Likewise.
17609 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17611 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17612 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17613 (arm_lookup_simd_builtin_type): Likewise.
17614 (arm_simd_builtin_type): Likewise.
17615 (arm_init_simd_builtin_types): Likewise.
17616 (arm_expand_builtin_args): Likewise.
17617 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17618 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17619 (ft32_setup_incoming_varargs): Likewise.
17620 (ft32_function_arg): Likewise.
17621 (ft32_function_arg_advance): Likewise.
17622 (ft32_pass_by_reference): Likewise.
17623 (ft32_arg_partial_bytes): Likewise.
17624 (ft32_valid_pointer_mode): Likewise.
17625 (ft32_addr_space_pointer_mode): Likewise.
17626 (ft32_addr_space_legitimate_address_p): Likewise.
17627 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17629 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17630 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17631 (iamcu_alignment): Likewise.
17632 (canonicalize_vector_int_perm): Likewise.
17633 (ix86_noce_conversion_profitable_p): Likewise.
17634 (ix86_mpx_bound_mode): Likewise.
17635 (ix86_operands_ok_for_move_multiple): Likewise.
17636 * config/microblaze/microblaze-protos.h
17637 (microblaze_expand_conditional_branch_reg): Likewise.
17638 * config/microblaze/microblaze.c
17639 (microblaze_expand_conditional_branch_reg): Likewise.
17640 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17642 (rs6000_reassociation_width): Likewise.
17643 (rs6000_invalid_binary_op): Likewise.
17644 (fusion_p9_p): Likewise.
17645 (emit_fusion_p9_load): Likewise.
17646 (emit_fusion_p9_store): Likewise.
17647 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17649 (riscv_hard_regno_mode_ok_p): Likewise.
17650 (riscv_address_insns): Likewise.
17651 (riscv_split_symbol): Likewise.
17652 (riscv_legitimize_move): Likewise.
17653 (riscv_function_value): Likewise.
17654 (riscv_hard_regno_nregs): Likewise.
17655 (riscv_expand_builtin): Likewise.
17656 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17657 (riscv_build_integer): Likewise.
17658 (riscv_split_integer): Likewise.
17659 (riscv_legitimate_constant_p): Likewise.
17660 (riscv_cannot_force_const_mem): Likewise.
17661 (riscv_regno_mode_ok_for_base_p): Likewise.
17662 (riscv_valid_base_register_p): Likewise.
17663 (riscv_valid_offset_p): Likewise.
17664 (riscv_valid_lo_sum_p): Likewise.
17665 (riscv_classify_address): Likewise.
17666 (riscv_legitimate_address_p): Likewise.
17667 (riscv_address_insns): Likewise.
17668 (riscv_load_store_insns): Likewise.
17669 (riscv_force_binary): Likewise.
17670 (riscv_split_symbol): Likewise.
17671 (riscv_force_address): Likewise.
17672 (riscv_legitimize_address): Likewise.
17673 (riscv_move_integer): Likewise.
17674 (riscv_legitimize_const_move): Likewise.
17675 (riscv_legitimize_move): Likewise.
17676 (riscv_address_cost): Likewise.
17677 (riscv_subword): Likewise.
17678 (riscv_output_move): Likewise.
17679 (riscv_canonicalize_int_order_test): Likewise.
17680 (riscv_emit_int_order_test): Likewise.
17681 (riscv_function_arg_boundary): Likewise.
17682 (riscv_pass_mode_in_fpr_p): Likewise.
17683 (riscv_pass_fpr_single): Likewise.
17684 (riscv_pass_fpr_pair): Likewise.
17685 (riscv_get_arg_info): Likewise.
17686 (riscv_function_arg): Likewise.
17687 (riscv_function_arg_advance): Likewise.
17688 (riscv_arg_partial_bytes): Likewise.
17689 (riscv_function_value): Likewise.
17690 (riscv_pass_by_reference): Likewise.
17691 (riscv_setup_incoming_varargs): Likewise.
17692 (riscv_print_operand): Likewise.
17693 (riscv_elf_select_rtx_section): Likewise.
17694 (riscv_save_restore_reg): Likewise.
17695 (riscv_for_each_saved_reg): Likewise.
17696 (riscv_register_move_cost): Likewise.
17697 (riscv_hard_regno_mode_ok_p): Likewise.
17698 (riscv_hard_regno_nregs): Likewise.
17699 (riscv_class_max_nregs): Likewise.
17700 (riscv_memory_move_cost): Likewise.
17701 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17702 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17703 (rl78_addr_space_address_mode): Likewise.
17704 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17706 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17707 (rs6000_reassociation_width): Likewise.
17708 (rs6000_invalid_binary_op): Likewise.
17709 (fusion_p9_p): Likewise.
17710 (emit_fusion_p9_load): Likewise.
17711 (emit_fusion_p9_store): Likewise.
17712 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17713 (ok_for_simple_move_operands): Likewise.
17714 (ok_for_simple_move_strict_operands): Likewise.
17715 (ok_for_simple_arith_logic_operands): Likewise.
17716 (visium_legitimize_reload_address): Likewise.
17717 (visium_select_cc_mode): Likewise.
17718 (output_cbranch): Likewise.
17719 (visium_split_double_move): Likewise.
17720 (visium_expand_copysign): Likewise.
17721 (visium_expand_int_cstore): Likewise.
17722 (visium_expand_fp_cstore): Likewise.
17723 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17724 (visium_function_arg): Likewise.
17725 (visium_function_arg_advance): Likewise.
17726 (visium_libcall_value): Likewise.
17727 (visium_setup_incoming_varargs): Likewise.
17728 (visium_legitimate_constant_p): Likewise.
17729 (visium_legitimate_address_p): Likewise.
17730 (visium_legitimize_address): Likewise.
17731 (visium_secondary_reload): Likewise.
17732 (visium_register_move_cost): Likewise.
17733 (visium_memory_move_cost): Likewise.
17734 (prepare_move_operands): Likewise.
17735 (ok_for_simple_move_operands): Likewise.
17736 (ok_for_simple_move_strict_operands): Likewise.
17737 (ok_for_simple_arith_logic_operands): Likewise.
17738 (visium_function_value_1): Likewise.
17739 (rtx_ok_for_offset_p): Likewise.
17740 (visium_legitimize_reload_address): Likewise.
17741 (visium_split_double_move): Likewise.
17742 (visium_expand_copysign): Likewise.
17743 (visium_expand_int_cstore): Likewise.
17744 (visium_expand_fp_cstore): Likewise.
17745 (visium_split_cstore): Likewise.
17746 (visium_select_cc_mode): Likewise.
17747 (visium_split_cbranch): Likewise.
17748 (output_cbranch): Likewise.
17749 (visium_print_operand_address): Likewise.
17750 * expmed.c (flip_storage_order): Likewise.
17751 * expmed.h (emit_cstore): Likewise.
17752 (flip_storage_order): Likewise.
17753 * genrecog.c (validate_pattern): Likewise.
17754 * hsa-gen.c (gen_hsa_addr): Likewise.
17755 * internal-fn.c (expand_arith_overflow): Likewise.
17756 * ira-color.c (allocno_copy_cost_saving): Likewise.
17757 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17758 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17759 (process_invariant_for_inheritance): Likewise.
17760 * lra-eliminations.c (move_plus_up): Likewise.
17761 * omp-low.c (lower_oacc_reductions): Likewise.
17762 * simplify-rtx.c (simplify_subreg): Likewise.
17763 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17764 (TARGET_CHKP_BOUND_MODE): Likewise..
17765 * targhooks.c (default_chkp_bound_mode): Likewise.
17766 (default_setup_incoming_vararg_bounds): Likewise.
17767 * targhooks.h (default_chkp_bound_mode): Likewise.
17768 (default_setup_incoming_vararg_bounds): Likewise.
17769 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17770 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17771 (have_whole_vector_shift): Likewise.
17772 * tree-vect-stmts.c (vectorizable_load): Likewise.
17773 * doc/tm.texi: Regenerate.
17775 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17777 Graceful degrade if Binutils PR21472 is not available.
17780 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17781 .rodata in flash test fails.
17782 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17783 * confgure: Regenerate.
17784 * config.in: Regenerate.
17785 * config/avr/avr.c (avr_asm_named_section)
17786 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17787 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17788 (avr_asm_init_sections): Same.
17790 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17792 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17793 (fma<VH:mode>4_intrinsic): Likewise.
17794 (*fmsub<VCVTF:mode>4): Likewise.
17795 (*fmsub<VH:mode>4_intrinsic): Likewise.
17797 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17800 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17801 Don't depend on "optimize > 0".
17802 (out_movhi_r_mr, out_movqi_mr_r): Same.
17803 (out_movhi_mr_r, out_movqi_r_mr): Same.
17804 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17805 io_address_operand on "optimize > 0".
17807 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17809 * tree-loop-distribution.c: Add general explanantion on the pass.
17810 (generate_loops_for_partition): Mark distributed loop.
17811 (pg_add_dependence_edges): New parameter. Handle alias data
17812 dependence specially and record it in the parameter if asked.
17813 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17814 (init_partition_graph_vertices, add_partition_graph_edge): New.
17815 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17816 (free_partition_graph_vdata, build_partition_graph): New.
17817 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17818 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17819 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17820 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17821 (version_for_distribution_p, finalize_partitions): New.
17822 (distribute_loop): Handle alias data dependence specially. Factor
17823 out loop fusion code as functions and call these functions.
17825 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17827 * tree-loop-distribution.c (classify_partition): New parameter and
17828 better handle reduction statement.
17829 (rdg_build_partitions): Revise comment.
17830 (distribute_loop): Compute statements in all partitions and pass it
17831 to classify_partition.
17833 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17835 * tree-loop-distribution.c (enum partition_type): New.
17836 (struct partition): New field type.
17837 (partition_merge_into): Add parameter. Update partition type.
17838 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17839 (build_rdg_partition_for_vertex): Compute partition type.
17840 (rdg_build_partitions): Dump partition type.
17841 (distribute_loop): Update calls to partition_merge_into.
17843 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17845 * tree-loop-distribution.c (struct ddr_hasher): New.
17846 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17848 (classify_partition): Call get_data_dependence.
17849 (pg_add_dependence_edges): Ditto.
17850 (distribute_loop): Release data dependence hash table.
17852 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17854 * tree-loop-distribution.c (ref_base_address): Delete.
17855 (similar_memory_accesses): Rename ...
17856 (share_memory_accesses): ... to this. Check if partitions access
17857 the same memory reference.
17858 (distribute_loop): Call share_memory_accesses.
17860 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17862 * tree-loop-distribution.c (struct partition): New field recording
17863 its data reference.
17864 (partition_alloc, partition_free): Init and release data refs.
17865 (partition_merge_into): Merge data refs.
17866 (build_rdg_partition_for_vertex): Collect data refs for partition.
17867 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17869 (distribute_loop): Remve data refs from vertice data of partition
17872 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17874 * tree-loop-distribution.c (params.h): Include header file.
17875 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17876 (datarefs_vec): New global var.
17877 (create_rdg_vertices): Use datarefs_vec directly.
17878 (free_rdg): Don't free data references.
17879 (build_rdg): Update use. Don't free data references.
17880 (distribute_loop): Compute global variable for data references.
17881 Bail out if there are too many data references.
17883 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17885 * tree-loop-distribution.c (loop_nest): New global var.
17886 (build_rdg): Use loop directly, rather than loop nest.
17887 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17889 (distribute_loop): Compute global variable loop nest. Update use.
17891 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17893 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17894 (partition_merge_into): New parameter. Dump reason for fusion.
17895 (distribute_loop): Update use of partition_merge_into.
17897 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17899 * tree-loop-distribution.c (bb_top_order_index): New.
17900 (bb_top_order_index_size, bb_top_order_cmp): New.
17901 (stmts_from_loop): Use topological order.
17902 (pass_loop_distribution::execute): Compute and release topological
17903 order for basic blocks.
17905 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17907 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17910 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17912 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17913 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17914 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17915 * internal-fn.def (LOOP_DIST_ALIAS): New.
17916 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17917 (fold_loop_internal_call): ... this.
17918 (vect_loop_dist_alias_call): New function.
17919 (set_uid_loop_bbs): Call fold_loop_internal_call.
17920 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17923 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17926 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17927 Require dead FLAGS_REG at the beginning of a peephole.
17929 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17932 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17933 arguments in the call to __builtin_ia32_sbb_u32.
17934 (_subborrow_u64): Swap _X and _Y arguments in the call to
17935 __builtin_ia32_sbb_u64.
17937 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17940 * tree-vrp.c (compare_assert_loc): Turn into a function template
17941 with stable template parameter. Only test if a->e is NULL,
17942 !a->e == !b->e has been verified already. Use e == NULL or
17943 e != NULL instead of e or ! e tests. If stable is true, don't use
17944 iterative_hash_expr, on the other side allow a or b or both NULL
17945 and sort the NULLs last.
17946 (process_assert_insertions): Sort using compare_assert_loc<false>
17947 instead of compare_assert_loc, later sort using
17948 compare_assert_loc<true> before calling process_assert_insertions_for
17949 in a loop. Use break instead of continue once seen NULL pointer.
17951 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17953 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17954 Cortex-R7 and Cortex-R8 processors.
17956 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17958 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17959 uninitialized while src is not.
17961 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17963 * common/config/arm/arm-common.c: Adjust include path for
17965 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17966 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17967 (arm-cpu-data.h): Likewise.
17968 (arm-cpu-cdata.h): Likewise.
17969 * config/arm/arm-cpu.h: Delete.
17970 * config/arm/arm-cpu-cdata.h: Delete.
17971 * config/arm/arm-cpu-data.h: Delete.
17973 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17975 * config/arm/arm-cpus.in (cortex-a55): New.
17976 (cortex-a75): Likewise.
17977 (cortex-a75.cortex-a55): Likewise.
17978 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17980 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17981 * config/arm/arm-cpu-cdata.h: Regenerate.
17982 * config/arm/arm-cpu-data.h: Regenerate.
17983 * config/arm/arm-cpu.h: Regenerate.
17984 * config/arm/arm-tables.opt: Regenerate.
17985 * config/arm/arm-tune.md: Regenerate.
17987 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17989 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17991 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17993 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17996 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17998 PR tree-optimization/81292
17999 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18000 full_string_p, also call adjust_related_strinfos if the adjustment
18001 is simple, otherwise invalidate related strinfos.
18003 2017-07-04 Martin Liska <mliska@suse.cz>
18006 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18007 newly created variable as DECL_IGNORED_P.
18009 2017-07-04 Martin Liska <mliska@suse.cz>
18012 * ipa-inline.c (inline_small_functions):
18013 Use xstrdup_for_dump.
18015 2017-07-04 Tom de Vries <tom@codesourcery.com>
18017 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18019 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18022 * config/darwin.c (darwin_function_switched_text_sections):
18025 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18027 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18029 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18031 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18033 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18035 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18036 min_profitable_iters, and th as inclusive lower bounds.
18037 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18038 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18039 for min_profitable_iters and min_profitable_estimate.
18040 (vect_transform_loop): Treat th as an inclusive lower bound.
18041 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18043 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18046 * config/darwin.c (darwin_function_switched_text_sections):
18047 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18048 in two pieces, and suppress the use of buf.
18050 2017-07-03 Nathan Sidwell <nathan@acm.org>
18052 * hash-table.h (hash_table_mod1): Fix indentation.
18054 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18056 PR middle-end/81290
18057 * predict.c (force_edge_cold): Be more careful about propagation
18059 * profile-count.h (profile_probability::guessed,
18060 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18062 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18064 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18066 * doc/invoke.texi (rcpc architecture extension): Document it.
18068 2017-07-03 Richard Biener <rguenther@suse.de>
18070 PR tree-optimization/60510
18071 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18072 the scalar reduction PHI and use it.
18073 (vectorizable_reduction): Properly guard the single_defuse_cycle
18074 path for non-SLP reduction chains where we cannot use it.
18075 Rework reduc_def/index and vector type deduction. Rework
18076 vector operand gathering during reduction op code-gen.
18077 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18078 chains dissolve the chain and leave it to non-SLP reduction
18081 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18083 * tree-data-ref.h (dr_alignment): Declare.
18084 * tree-data-ref.c (dr_alignment): New function.
18085 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18086 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18088 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18090 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18092 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18093 and base_misalignment fields.
18094 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18095 * tree-data-ref.c: Include builtins.h.
18096 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18097 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18098 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18099 * tree-vect-data-refs.c: Include tree-cfg.h.
18100 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18101 fields instead of calculating an alignment here.
18102 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18103 innermost_loop_behavior fields.
18105 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18107 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18109 (DR_STEP_ALIGNMENT): New macro.
18110 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18111 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18112 (create_data_ref): Print it.
18113 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18114 to tell whether the step preserves vector (mis)alignment.
18115 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18116 Move the check for an integer step and generalise to all INTEGER_CST.
18117 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18118 Print the outer step alignment.
18120 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18122 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18123 with offset_alignment.
18124 (DR_ALIGNED_TO): Delete.
18125 (DR_OFFSET_ALIGNMENT): New macro.
18126 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18127 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18128 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18129 (create_data_ref): Likewise.
18130 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18131 (vect_analyze_data_refs): Likewise.
18132 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18133 creating dummy innermost behavior.
18135 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18137 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18138 with a "innermost_loop_behavior *" and refeence tree.
18139 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18140 (create_data_ref): Update call accordingly.
18141 * tree-predcom.c (find_looparound_phi): Likewise.
18143 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18145 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18146 fields with dr_wrt_vec_loop.
18147 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18148 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18149 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18150 (vect_dr_behavior): New function.
18151 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18152 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18153 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18154 track whether the step preserves the misalignment.
18155 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18156 Use vect_dr_behavior.
18157 (vect_setup_realignment): Update call accordingly.
18158 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18159 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18160 call to vect_create_addr_base_for_vector_ref.
18161 (vect_create_cond_for_align_checks): Likewise.
18162 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18163 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18164 (vect_recog_mask_conversion_pattern): Likewise.
18165 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18166 (new_stmt_vec_info): Remove redundant zeroing.
18168 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18170 * common/config/arm/arm-common.c (arm_be8_option): New function.
18171 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18172 (ISA_ARMv6): Add isa_bit_be8.
18173 * config/arm/arm.h (arm_be8_option): Add prototype.
18174 (BE8_SPEC_FUNCTION): New define.
18175 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18176 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18177 (mlittle-endian): Similarly.
18178 (mbe8, mbe32): New options.
18179 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18180 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18182 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18184 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18186 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18188 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18189 (cleanup_tree_cfg_bb): Use it.
18190 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18192 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18194 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18197 * loop-doloop.c (add_test): Update profile.
18199 2017-07-03 Martin Liska <mliska@suse.cz>
18202 * sanopt.c (rewrite_usage_of_param): New function.
18203 (sanitize_rewrite_addressable_params): Likewise.
18204 (pass_sanopt::execute): Call rewrite_usage_of_param.
18206 2017-07-03 Richard Biener <rguenther@suse.de>
18208 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18209 back to using VIEW_CONVERT_EXPR.
18211 2017-07-03 Martin Liska <mliska@suse.cz>
18214 * doc/extend.texi: Document when a resolver function is
18215 generated for target_clones.
18217 2017-07-03 Martin Liska <mliska@suse.cz>
18219 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18220 and shadow memory of auto variables which are subject of
18221 use-after-scope sanitization.
18222 (asan_expand_mark_ifn): Add do set only when is_poison.
18224 2016-07-03 Richard Biener <rguenther@suse.de>
18226 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18228 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18229 (vectorizable_reduction): Perform reduction PHI creation when
18230 visiting a reduction PHI and adjust and simplify code generation
18231 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18232 (vect_transform_loop): Visit reduction PHIs.
18233 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18234 defs into the SLP tree.
18235 (vect_build_slp_tree): Reduction defs terminate the recursion.
18236 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18238 (vect_get_vec_defs_for_stmt_copy): Export.
18239 (vect_get_vec_defs): Likewise.
18240 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18242 (vect_get_vec_defs_for_stmt_copy): Declare.
18243 (vect_get_vec_defs): Likewise.
18245 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18247 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18248 parameter with a "loop" parameter and use it instead of the
18249 loop containing DR_STMT. Don't check simple_iv when doing
18250 BB analysis. Describe the two analysis modes in the comment.
18252 2017-07-03 Tom de Vries <tom@codesourcery.com>
18254 PR tree-optimization/69468
18255 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18256 (find_same_succ_bb): Handle ignore_edge_flags.
18258 2017-07-03 Tom de Vries <tom@codesourcery.com>
18260 PR tree-optimization/81192
18261 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18263 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18265 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18267 2017-07-03 Tom de Vries <tom@codesourcery.com>
18269 PR tree-optimization/81192
18270 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18271 BB_SAME_SUCC (bb) == NULL.
18273 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18275 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18278 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18280 * dumpfile.c: Include profile-count.h
18281 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18283 (insert_cond_bb): Update profile.
18284 * tree-cfg.h (insert_cond_bb): Update prototype.
18285 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18286 * tree-dump.c: Do not include tree-cfg.
18288 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18290 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18292 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18294 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18297 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18299 * tree-complex.c (expand_complex_div_wide): update profile.
18301 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18302 Alan Hayward <alan.hayward@arm.com>
18303 David Sherwood <david.sherwood@arm.com>
18305 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18306 (CORETYPES_H): New define.
18307 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18308 (insn-modes-inline.h, s-modes-inline-h): New rules.
18309 (generated_files): Add insn-modes-inline.h.
18310 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18311 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18312 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18313 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18314 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18315 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18316 (build/gencodes.o, build/genconditions.o): Likewise.
18317 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18318 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18319 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18320 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18321 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18322 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18323 * coretypes.h: Include everything up to real.h for generators.
18324 Include insn-modes.h first. Include wide-int-print.h after
18325 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18326 * machmode.h: Don't include insn-modes.h here.
18327 * function-tests.c: Remove includes of signop.h, machmode.h,
18328 double-int.h and wide-int.h.
18330 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18332 * optc-save-gen.awk: Likewise.
18333 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18334 * godump.c: Remove include of wide-int-print.h.
18335 * pretty-print.h: Likewise.
18336 * wide-int-print.cc: Likewise.
18337 * wide-int.cc: Likewise.
18338 * hash-map-tests.c: Remove include of signop.h.
18339 * hash-set-tests.c: Likewise.
18340 * rtl-tests.c: Likewise.
18341 * mkconfig.sh: Remove include of machmode.h.
18342 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18344 (emit_insn_modes_inline_h): ...this new function. Emit the code
18345 into an insn-modes-inline.h header file, adding appropriate
18346 include guards and end comments.
18347 (emit_insn_modes_c_header): Remove include of machmode.h.
18348 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18350 (main): Handle -i flag and call emit_insn_modes_inline_h when
18353 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18355 * tree-ssa-strlen.c (strinfo): Rename the length field to
18356 nonzero_chars. Add a full_string_p field.
18357 (compare_nonzero_chars, zero_length_string_p): New functions.
18358 (get_addr_stridx): Add an offset_out parameter.
18359 Use compare_nonzero_chars.
18360 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18361 (new_strinfo): Update after above changes to strinfo.
18362 (set_endptr_and_length): Set full_string_p.
18363 (get_string_length): Update after above changes to strinfo.
18364 (unshare_strinfo): Update call to new_strinfo.
18365 (maybe_invalidate): Likewise.
18366 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18367 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18368 as a uhwi instead of an shwi. Update after above changes to
18369 strinfo and new_strinfo.
18370 (zero_length_string): Assert that chainsi contains full strings.
18371 Use zero_length_string_p. Update call to new_strinfo.
18372 (adjust_related_strinfos): Update after above changes to strinfo.
18373 Copy full_string_p from origsi.
18374 (adjust_last_stmt): Use zero_length_string_p.
18375 (handle_builtin_strlen): Update after above changes to strinfo and
18376 new_strinfo. Install the lhs as the string length if the previous
18377 entry didn't describe a full string.
18378 (handle_builtin_strchr): Update after above changes to strinfo
18380 (handle_builtin_strcpy): Likewise.
18381 (handle_builtin_strcat): Likewise.
18382 (handle_builtin_malloc): Likewise.
18383 (handle_pointer_plus): Likewise.
18384 (handle_builtin_memcpy): Likewise. Track nonzero characters
18385 that aren't necessarily followed by a nul terminator.
18386 (handle_char_store): Likewise.
18388 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18390 PR tree-optimization/80769
18391 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18392 for malloc and calloc. Document the new invariant that all related
18393 strinfos have delayed lengths or none do.
18394 (verify_related_strinfos): Move earlier in file.
18395 (set_endptr_and_length): New function, split out from...
18396 (get_string_length): ...here. Also set the lengths of related
18398 (zero_length_string): Assert that chainsi has known (rather than
18400 (adjust_related_strinfos): Likewise.
18402 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18404 PR tree-optimization/81136
18405 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18406 assert that two references with the same misalignment have the same
18407 compile-time misalignment if those compile-time misalignments
18410 2017-07-01 Andi Kleen <ak@linux.intel.com>
18412 * print-tree.c (print_node): Print all attributes.
18414 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18416 * cfg.c (scale_bbs_frequencies): New function.
18417 * cfg.h (scale_bbs_frequencies): Declare it.
18418 * cfgloopanal.c (single_likely_exit): Cleanup.
18419 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18421 (scale_loop_profile): Likewise.
18422 (loop_version): Likewise.
18423 (create_empty_loop_on_edge): Update.
18424 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18425 scale_loop_frequencies, scale_loop_profile, loopify,
18426 loop_version): Update prototypes.
18427 * modulo-sched.c (sms_schedule): Update.
18428 * predict.c (unlikely_executed_edge_p): Also check probability.
18429 (probably_never_executed_edge_p): Fix typo.
18430 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18431 * tree-parloops.c (gen_parallel_loop): Update.
18432 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18433 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18434 * tree-ssa-loop-split.c (split_loop): Update.
18435 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18436 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18437 (vect_loop_versioning): Update.
18438 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18440 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18442 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18444 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18446 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18447 to keep profile consistent.
18449 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18451 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18452 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18453 * profile-count.h (max_safe_multiplier): Make unsigned.
18454 (profile_count::guessed_zero): New.
18456 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18458 * bb-reorder.c (fix_up_crossing_landing_pad,
18459 fix_crossing_conditional_branches): Use make_single_succ_edge
18460 to keep profile consistent.
18462 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18464 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18467 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18470 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18471 the right scopes, make sure cond_jump isn't preserved between multiple
18472 iterations. Search for fallthru edge whenever there are 3+ edges and
18473 use find_fallthru_edge for it.
18475 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18477 Patch by Alexander Monakov <amonakov@ispras.ru>
18478 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18479 probabilities consistently.
18481 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18483 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18484 emit_cmp_and_jump_insns.
18486 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18489 * tree-inline.c (expand_call_inline): Combine profile statuses.
18491 2017-06-30 Andrew Pinski <apinski@cavium.com>
18493 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18494 fold_stmt returned true.
18496 2017-06-30 Nathan Sidwell <nathan@acm.org>
18498 * ggc.h (empty_string): Delete.
18499 * cfgexpand.c (expand_asm_stmt): Use plain "".
18500 * optabs.c (expand_asm_memory_barrier): Likewise.
18501 * stringpool.c (empty_string): Delete.
18502 (digit_vector, digit_string): Delete.
18503 (ggc_alloc_string): Use plain "", don't optimize single digit
18504 strings. Use ggc_alloc_atomic.
18506 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18508 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18509 comparison set and one other set, use the cost of the non-comparison
18512 2017-06-30 Nathan Sidwell <nathan@acm.org>
18514 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18517 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18519 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18520 loops. Remove now unneeded calls to gimple_switch_set_label() that
18521 just set removed labels to NULL_TREE.
18523 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18525 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18526 (set_range_info): ...here. Only call set_range_info_raw if domain
18528 (set_nonzero_bits): Call set_range_info_raw.
18529 * tree-ssanames.h (set_range_info_raw): New.
18531 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18534 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18535 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18536 of nonimmediate_operand and <store_mask_constraint> instead of m
18537 for the input operand. For V8FI iterator, always split if input
18538 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18539 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18540 <store_mask_predicate> instead of register_operand and
18541 <store_mask_constraint> instead of v for the input operand. Make
18542 sure both operands aren't MEMs for if not <mask_applied>.
18544 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18546 * lto-wrapper.c (copy_file) Close both file descriptors before
18549 2017-06-30 Martin Liska <mliska@suse.cz>
18552 * multiple_target.c (create_dispatcher_calls): Make ifunc
18553 also for function that don't have calls or are not referenced.
18555 2017-06-30 Richard Biener <rguenther@suse.de>
18557 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18558 analyze the first scalar stmt. Move vector type computation
18559 for the BB case here from ...
18560 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18561 live operation processing in the SLP case properly.
18563 2017-06-30 Richard Biener <rguenther@suse.de>
18565 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18567 2017-06-30 Martin Liska <mliska@suse.cz>
18570 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18571 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18573 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18575 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18578 2017-06-30 Richard Biener <rguenther@suse.de>
18580 PR tree-optimization/81249
18581 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18582 condition reduction result to original scalar type.
18584 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18586 * profile-count.h (enum profile_quality): Fix typos and whitespace
18589 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18591 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18592 type for branch probabilities.
18594 2017-06-29 Julian Brown <julian@codesourcery.com>
18595 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18597 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18598 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18599 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18600 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18602 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18604 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18605 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18606 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18607 CC usage from generic code to here.
18608 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18609 CC usage into the target macros.
18611 2017-06-29 Maya Rashish <coypu@sdf.org>
18613 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18616 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18618 * arm/arm-builtins.c: Include profile-count.h
18619 * except.c (sjlj_emit_function_enter): Use
18620 profile_probability::unlikely.
18622 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18624 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18625 and tocrel_offset be pointer args rather than implicitly using
18627 (legitimate_constant_pool_address_p, rs6000_emit_move,
18628 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18629 tocrel_offset and use in toc_relative_expr_p call.
18630 (print_operand, print_operand_address): Use static tocrel_base_oac
18631 and tocrel_offset_oac.
18632 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18635 2017-06-29 Maya Rashish <coypu@sdf.org>
18637 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18639 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18641 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18642 objects, take into account only the alignment of 'op0' and 'mode1' if
18645 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18647 * ccmp.c (ccmp_tree_comparison_p): New function.
18648 (ccmp_candidate_p): Update to use above function.
18649 (get_compare_parts): New function.
18650 (expand_ccmp_next): Update to use new functions.
18651 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18653 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18654 take mode as argument.
18655 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18656 * expr.c (expand_expr_real_1): Pass mode as argument.
18658 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18660 * combine.c (combine_instructions): Print insns to dump_file, together
18663 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18665 * asan.c (asan_emit_stack_protection): Update.
18666 (create_cond_insert_point): Update.
18667 * auto-profile.c (afdo_propagate_circuit): Update.
18668 * basic-block.h (struct edge_def): Turn probability to
18669 profile_probability.
18670 (EDGE_FREQUENCY): Update.
18671 * bb-reorder.c (find_traces_1_round): Update.
18672 (better_edge_p): Update.
18673 (sanitize_hot_paths): Update.
18674 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18675 (make_single_succ_edge): Update.
18676 (check_bb_profile): Update.
18677 (dump_edge_info): Update.
18678 (update_bb_profile_for_threading): Update.
18679 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18680 probabilitycount to 0.
18681 * cfgbuild.c (compute_outgoing_frequencies): Update.
18682 * cfgcleanup.c (try_forward_edges): Update.
18683 (outgoing_edges_match): Update.
18684 (try_crossjump_to_edge): Update.
18685 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18686 (expand_gimple_tailcall): Update.
18687 (construct_init_block): Use make_single_succ_edge.
18688 (construct_exit_block): Use make_single_succ_edge.
18689 * cfghooks.c (verify_flow_info): Update.
18690 (redirect_edge_succ_nodup): Update.
18691 (split_edge): Update.
18692 (account_profile_record): Update.
18693 * cfgloopanal.c (single_likely_exit): Update.
18694 * cfgloopmanip.c (scale_loop_profile): Update.
18695 (set_zero_probability): Remove.
18696 (duplicate_loop_to_header_edge): Update.
18697 * cfgloopmanip.h (loop_version): Update prototype.
18698 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18699 (force_nonfallthru_and_redirect): Update.
18700 (update_br_prob_note): Update.
18701 (rtl_verify_edges): Update.
18702 (purge_dead_edges): Update.
18703 (rtl_lv_add_condition_to_bb): Update.
18704 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18705 * cgraphunit.c (init_lowered_empty_function): Update.
18706 (cgraph_node::expand_thunk): Update.
18707 * cilk-common.c: Include profile-count.h
18708 * dojump.c (inv): Remove.
18709 (jumpifnot): Update.
18710 (jumpifnot_1): Update.
18711 (do_jump_1): Update.
18713 (do_jump_by_parts_greater_rtx): Update.
18714 (do_compare_rtx_and_jump): Update.
18715 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18716 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18717 * dwarf2cfi.c: Include profile-count.h
18718 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18719 (sjlj_emit_dispatch_table): Likewise.
18720 * explow.c: Include profile-count.h
18721 * expmed.c (emit_store_flag_force): Update.
18722 (do_cmp_and_jump): Update.
18723 * expr.c (compare_by_pieces_d::generate): Update.
18724 (compare_by_pieces_d::finish_mode): Update.
18725 (emit_block_move_via_loop): Update.
18726 (store_expr_with_bounds): Update.
18727 (store_constructor): Update.
18728 (expand_expr_real_2): Update.
18729 (expand_expr_real_1): Update.
18730 * expr.h (try_casesi, try_tablejump): Update prototypes.
18731 * gimple-pretty-print.c (dump_probability): Update.
18732 (dump_profile): New.
18733 (dump_gimple_label): Update.
18734 (dump_gimple_bb_header): Update.
18735 * graph.c (draw_cfg_node_succ_edges): Update.
18736 * hsa-gen.c (convert_switch_statements): Update.
18737 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18738 (find_if_case_1): Update.
18739 (find_if_case_2): Update.
18740 * internal-fn.c (expand_arith_overflow_result_store): Update.
18741 (expand_addsub_overflow): Update.
18742 (expand_neg_overflow): Update.
18743 (expand_mul_overflow): Update.
18744 (expand_vector_ubsan_overflow): Update.
18745 * ipa-cp.c (good_cloning_opportunity_p): Update.
18746 * ipa-split.c (split_function): Use make_single_succ_edge.
18747 * ipa-utils.c (ipa_merge_profiles): Update.
18748 * loop-doloop.c (add_test): Update.
18749 (doloop_modify): Update.
18750 * loop-unroll.c (compare_and_jump_seq): Update.
18751 (unroll_loop_runtime_iterations): Update.
18752 * lra-constraints.c (lra_inheritance): Update.
18753 * lto-streamer-in.c (input_cfg): Update.
18754 * lto-streamer-out.c (output_cfg): Update.
18755 * mcf.c (adjust_cfg_counts): Update.
18756 * modulo-sched.c (sms_schedule): Update.
18757 * omp-expand.c (expand_omp_for_init_counts): Update.
18758 (extract_omp_for_update_vars): Update.
18759 (expand_omp_ordered_sink): Update.
18760 (expand_omp_for_ordered_loops): Update.
18761 (expand_omp_for_generic): Update.
18762 (expand_omp_for_static_nochunk): Update.
18763 (expand_omp_for_static_chunk): Update.
18764 (expand_cilk_for): Update.
18765 (expand_omp_simd): Update.
18766 (expand_omp_taskloop_for_outer): Update.
18767 (expand_omp_taskloop_for_inner): Update.
18768 * omp-simd-clone.c (simd_clone_adjust): Update.
18769 * optabs.c (expand_doubleword_shift): Update.
18770 (expand_abs): Update.
18771 (emit_cmp_and_jump_insn_1): Update.
18772 (expand_compare_and_swap_loop): Update.
18773 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18774 * predict.c (predictable_edge_p): Update.
18775 (edge_probability_reliable_p): Update.
18776 (set_even_probabilities): Update.
18777 (combine_predictions_for_insn): Update.
18778 (combine_predictions_for_bb): Update.
18779 (propagate_freq): Update.
18780 (estimate_bb_frequencies): Update.
18781 (force_edge_cold): Update.
18782 * profile-count.c (profile_count::dump): Add missing space into dump.
18783 (profile_count::debug): Add newline.
18784 (profile_count::differs_from_p): Explicitly convert to unsigned.
18785 (profile_count::stream_in): Update.
18786 (profile_probability::dump): New member function.
18787 (profile_probability::debug): New member function.
18788 (profile_probability::differs_from_p): New member function.
18789 (profile_probability::differs_lot_from_p): New member function.
18790 (profile_probability::stream_in): New member function.
18791 (profile_probability::stream_out): New member function.
18792 * profile-count.h (profile_count_quality): Rename to ...
18793 (profile_quality): ... this one.
18794 (profile_probability): New.
18795 (profile_count): Update.
18796 * profile.c (compute_branch_probabilities): Update.
18797 * recog.c (peep2_attempt): Update.
18798 * sched-ebb.c (schedule_ebbs): Update.
18799 * sched-rgn.c (find_single_block_region): Update.
18800 (compute_dom_prob_ps): Update.
18801 (schedule_region): Update.
18802 * sel-sched-ir.c (compute_succs_info): Update.
18803 * stmt.c (struct case_node): Update.
18804 (do_jump_if_equal): Update.
18805 (get_outgoing_edge_probs): Update.
18806 (conditional_probability): Update.
18807 (emit_case_dispatch_table): Update.
18808 (expand_case): Update.
18809 (expand_sjlj_dispatch_table): Update.
18810 (emit_case_nodes): Update.
18811 * targhooks.c: Update.
18812 * tracer.c (better_p): Update.
18813 (find_best_successor): Update.
18814 * trans-mem.c (expand_transaction): Update.
18815 * tree-call-cdce.c: Update.
18816 * tree-cfg.c (gimple_split_edge): Upate.
18817 (move_sese_region_to_fn): Upate.
18818 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18819 * tree-eh.c (lower_resx): Upate.
18820 (cleanup_empty_eh_move_lp): Upate.
18821 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18822 * tree-inline.c (copy_edges_for_bb): Update.
18823 (copy_cfg_body): Update.
18824 * tree-parloops.c (gen_parallel_loop): Update.
18825 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18826 (gimple_gen_time_profiler): Update.
18827 * tree-ssa-dce.c (remove_dead_stmt): Update.
18828 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18829 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18830 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18831 (unloop_loops): Update.
18832 (try_peel_loop): Update.
18833 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18834 * tree-ssa-loop-split.c (connect_loops): Update.
18835 (split_loop): Update.
18836 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18837 (hoist_guard): Update.
18838 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18839 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18840 (value_replacement): Update.
18841 * tree-ssa-reassoc.c (branch_fixup): Update.
18842 * tree-ssa-tail-merge.c (replace_block_by): Update.
18843 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18844 (create_edge_and_update_destination_phis): Update.
18845 (compute_path_counts): Update.
18846 (recompute_probabilities): Update.
18847 (update_joiner_offpath_counts): Update.
18848 (freqs_to_counts_path): Update.
18849 (duplicate_thread_path): Update.
18850 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18851 (struct switch_conv_info): Update.
18852 (gen_inbound_check): Update.
18853 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18854 (vect_do_peeling): Update.
18855 (vect_loop_versioning): Update.
18856 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18857 (optimize_mask_stores): Update.
18858 * ubsan.c (ubsan_expand_null_ifn): Update.
18859 * value-prof.c (gimple_divmod_fixed_value): Update.
18860 (gimple_divmod_fixed_value_transform): Update.
18861 (gimple_mod_pow2): Update.
18862 (gimple_mod_pow2_value_transform): Update.
18863 (gimple_mod_subtract): Update.
18864 (gimple_mod_subtract_transform): Update.
18865 (gimple_ic): Update.
18866 (gimple_stringop_fixed_value): Update.
18867 (gimple_stringops_transform): Update.
18868 * value-prof.h: Update.
18870 2017-06-29 Carl Love <cel@us.ibm.com>
18872 * config/rs6000/rs6000-c.c: Add support for built-in functions
18873 vector signed int vec_signed (vector float);
18874 vector signed long long vec_signed (vector double);
18875 vector signed int vec_signed2 (vector double, vector double);
18876 vector signed int vec_signede (vector double);
18877 vector signed int vec_signedo (vector double);
18878 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18879 instruction generator.
18880 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18881 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18882 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18884 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18885 vunsignede_v2df): Add define_expands.
18886 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18887 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18888 VEC_UNSIGNEDO): Add definitions.
18889 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18890 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18891 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18892 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18893 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18894 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18895 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18896 vec_unsignede, vec_unsignedo): Add builtin defines.
18897 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18899 * doc/extend.texi: Update the built-in documentation file for the
18900 new built-in functions.
18902 2017-06-29 Richard Biener <rguenther@suse.de>
18904 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18905 reduction chains to LOOP_VINFO_REDUCTIONS.
18906 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18907 SLP reductions after processing reduction chains.
18909 2017-06-29 Nathan Sidwell <nathan@acm.org>
18911 * builtins.c (fold_builtin_FUNCTION): Use
18912 lang_hooks.decl_printable_name.
18914 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18916 PR middle-end/81194
18917 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18918 with only one label.
18919 * stmt.c (expand_case): Assert NCASES is greater than one.
18921 2017-06-29 Richard Biener <rguenther@suse.de>
18923 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18925 (group_case_labels): Likewise.
18926 (find_taken_edge): Push sanity checking on val to workers...
18927 (find_taken_edge_cond_expr): ... here
18928 (find_taken_edge_switch_expr): ... and here, handle cases
18929 with just a default label.
18930 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18931 (group_case_labels): Likewise.
18932 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18933 group_case_labels does anything cleanup the CFG again.
18935 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18937 PR tree-optimization/81196
18938 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18939 exit condition comparing two IVs.
18941 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18943 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18944 profile to the dummy entry at the end of the list of architectures.
18945 * config/arm/arm-cpu-cdata.h: Regenerated.
18947 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18948 Michael Collison <michael.collison@arm.com>
18951 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18953 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18954 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18955 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18956 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18957 of shift when the shift amount is masked with constant equal to
18958 the size of the mode.
18959 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18962 2017-06-29 Martin Liska <mliska@suse.cz>
18964 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18966 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18969 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18972 (edge_growth_cache_entry::edge_growth_cache_entry): New
18974 (reset_edge_growth_cache): Update to use constructor.
18976 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18978 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18979 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18980 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18982 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18984 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18985 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18987 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18989 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18990 (*-linux-uclibc*): Add t-uclibc tmake_file.
18991 * config/t-musl: New.
18992 * config/t-uclibc: New.
18994 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18996 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18998 (gen_comm_data): Emit architectural setting of arch_prof.
18999 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19001 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19002 (armv8-m.base, armv8-m.main): Likewise.
19003 * arm-protos.h (arm_build_target): Add profile field.
19004 (arch_option): Likewise.
19005 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19007 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19008 arm_active_target.profile.
19010 2017-06-28 Richard Biener <rguenther@suse.de>
19012 PR middle-end/81227
19013 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19014 TYPE_OVERFLOW_WRAPS.
19015 * match.pd (negate_expr_p): Likewise.
19016 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19017 fold_build2, not fold_binary.
19019 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19021 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19022 Convert memory address to Pmode.
19023 (aarch64_print_operand): Assert MEM operands are always Pmode.
19025 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19028 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19029 Remove redundant if.
19030 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19031 * config/arm/aarch-common-protos.h
19032 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19033 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19035 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19038 * multiple_target.c (create_dispatcher_calls): Set the default
19039 clone to be static, not public.
19041 2017-06-28 Richard Biener <rguenther@suse.de>
19043 * tree-vect-loop.c (vectorizable_reduction): Move special
19044 cond reduction IV var creation ...
19045 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19046 parameter. Use STMT_VINFO_VECTYPE.
19047 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19050 2017-06-28 Martin Liska <mliska@suse.cz>
19053 * ipa-visibility.c (non_local_p): Handle visibility.
19055 2017-06-28 Martin Liska <mliska@suse.cz>
19058 * common.opt: Add IntegerRange to various options.
19059 * opt-functions.awk (integer_range_info): New function.
19060 * optc-gen.awk: Add integer_range_info to cl_options struct.
19061 * opts-common.c (decode_cmdline_option): Handle
19062 CL_ERR_INT_RANGE_ARG.
19063 (cmdline_handle_error): Likewise.
19064 * opts.c (print_filtered_help): Show valid interval in
19065 when --help is provided.
19066 * opts.h (struct cl_option): Add range_min and range_max fields.
19067 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19069 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19071 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19072 (x * C EQ/NE y * C): New transformation.
19074 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19076 * genmultilib (combination_space): Accept '+' in option names.
19078 2017-06-28 Martin Liska <mliska@suse.cz>
19081 * asan.c (instrument_derefs): Bail out inner references
19082 that are hard register variables.
19084 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19087 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19088 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19090 2017-06-28 Richard Biener <rguenther@suse.de>
19092 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19093 (vect_get_slp_defs): Adjust.
19094 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19096 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19098 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19099 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19100 (vectorizable_reduction): Adjust.
19101 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19103 (vect_get_slp_defs): Likewise.
19104 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19105 (vectorizable_bswap): Adjust.
19106 (vectorizable_call): Likewise.
19107 (vectorizable_conversion): Likewise.
19108 (vectorizable_assignment): Likewise.
19109 (vectorizable_shift): Likewise.
19110 (vectorizable_operation): Likewise.
19111 (vectorizable_store): Likewise.
19112 (vectorizable_condition): Likewise.
19113 (vectorizable_comparison): Likewise.
19115 2017-06-28 Michael Collison <michael.collison@arm.com>
19118 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19120 (arm_gen_movmemqi): Removed unused variable 'i'.
19121 Convert 'for' loop into 'while' loop.
19122 (arm_expand_prologue): Remove last unnecessary set of insn.
19123 (thumb_pop): Remove unused variable 'pushed_words'.
19124 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19126 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19128 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19129 * config/s390/s390-protos.h: Add prototype of
19130 s390_rel_address_ok_p.
19131 * config/s390/s390.c (s390_got_symbol): New function.
19132 (s390_rel_address_ok_p): New function.
19133 (legitimize_pic_address): Use s390_rel_address_ok_p.
19134 (s390_load_got): Use s390_got_symbol.
19135 (s390_option_override): Issue error if
19136 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19137 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19139 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19141 2017-06-27 Andrew Pinski <apinski@cavium.com>
19143 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19144 (X * copysign (1.0, X)): New pattern.
19145 (X * copysign (1.0, -X)): New pattern.
19146 (copysign (-1.0, CST)): New pattern.
19148 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19150 * genmultilib (combination_space): Remove variable.
19151 Validate reuse rules against regular expression for any sequence
19152 of multilib options in any order.
19154 2017-06-27 Michael Collison <michael.collison@arm.com>
19156 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19157 call aarch64_split_simd_combine.
19158 * (aarch64_combine_internal<mode>): Delete pattern.
19159 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19160 Allow register and subreg operands.
19162 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19164 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19165 specific need, just fallback on defaults.
19166 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19168 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19169 Olivier Hainque <hainque@adacore.com>
19171 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19173 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19174 targets. Pick a default if no particular attempt applied.
19175 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19178 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19180 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19181 (x86_64-wrs-vxworks7): Likewise.
19183 2017-06-27 Marek Polacek <polacek@redhat.com>
19186 * ubsan.c (instrument_null): Check get_base_address's result for null.
19188 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19190 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19192 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19194 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19195 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19196 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19197 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19198 New function types.
19199 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19200 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19201 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19202 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19203 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19204 BUILT_IN_FEUPDATEENV): New builtins.
19205 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19206 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19207 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19208 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19210 (builtin_structptr_types): Adjust size.
19211 * tree.c (builtin_structptr_types): Add four entries.
19213 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19214 Olivier Hainque <hainque@adacore.com>
19216 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19217 (TLS_SYM): New local macro, forcing reference to __tls__ on
19218 link command lines for VxWorks 7 RTPs, triggering initialization
19220 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19221 OS features TLS support, true for RTPs on VxWorks 7.
19222 * config/vxworks.c (vxworks_override_options): Setup emutls
19225 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19227 * predict.c (test_prediction_value_range): Use -1U instead of -1
19228 to avoid narrowing conversion warning.
19229 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19230 to avoid narrowing conversion warning.
19231 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19233 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19235 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19237 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19238 64bit configurations.
19239 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19240 (SIZE_TYPE): Likewise.
19241 * config/vxworks.c (vxworks_emutls_var_fields): Use
19242 long_unsigned_type_node instead of unsigned_type_node as the offset
19243 field type, which is "pointer" mode in emutls.c.
19245 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19248 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19250 PR middle-end/81207
19251 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19252 gimple_vuse copying separately from gimple_vdef copying.
19254 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19256 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19259 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19260 Olivier Hainque <hainque@adacore.com>
19262 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19263 for all vxworks7 targets.
19264 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19265 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19266 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19267 variations for VX6/VX7 and 32/64bits later on in ...
19268 (VXWORKS_LIB_SPEC): Leverage new macros.
19269 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19270 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19272 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19274 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19275 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19277 2017-06-26 Carl Love <cel@us.ibm.com>
19279 * config/rs6000/rs6000-c.c: Add support for built-in functions
19280 vector bool char vec_reve (vector bool char);
19281 vector signed char vec_reve (vector signed char);
19282 vector unsigned char vec_reve (vector unsigned char);
19283 vector bool int vec_reve (vector bool int);
19284 vector signed int vec_reve (vector signed int);
19285 vector unsigned int vec_reve (vector unsigned int);
19286 vector bool long long vec_reve (vector bool long long);
19287 vector signed long long vec_reve (vector signed long long);
19288 vector unsigned long long vec_reve (vector unsigned long long);
19289 vector bool short vec_reve (vector bool short);
19290 vector signed short vec_reve (vector signed short);
19291 vector double vec_reve (vector double);
19292 vector float vec_reve (vector float);
19293 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19294 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19295 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19296 (altivec_vreve): New pattern.
19297 * config/rs6000/altivec.h (vec_reve): New define.
19298 * doc/extend.texi (vec_rev): Update the built-in documentation file
19299 for the new built-in functions.
19301 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19303 PR tree-optimization/71815
19304 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19306 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19308 (slsr_process_phi): Likewise.
19309 (replace_uncond_cands_and_profitable_phis): Don't replace a
19310 multiply candidate with a stride of 1 (copy or cast).
19311 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19313 (lowest_cost_path): Likewise.
19314 (total_savings): Likewise.
19316 2017-06-26 Richard Biener <rguenther@suse.de>
19319 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19320 Use def_builtin_pure for all gather builtins.
19322 2017-06-26 Richard Biener <rguenther@suse.de>
19324 PR tree-optimization/81203
19325 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19326 non-dominating BBs.
19328 2017-06-26 Marek Polacek <polacek@redhat.com>
19331 * doc/invoke.texi: Document -Wmultistatement-macros.
19333 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19335 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19336 arm_neon_ok_no_float_abi effective target.
19338 2017-06-26 Richard Biener <rguenther@suse.de>
19340 PR tree-optimization/80928
19341 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19342 (copy_bbs): Set BB_DUPLICATED flag early.
19343 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19345 (execute_on_shrinking_pred): Likewise.
19346 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19347 BB_DUPLICATED blocks.
19348 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19349 iterate over all PHIs considering removal of *gsi.
19351 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19353 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19356 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19358 * config/rs6000/rs6000-string.c: (expand_block_clear,
19359 do_load_for_compare, select_block_compare_mode,
19360 compute_current_alignment, expand_block_compare,
19361 expand_strncmp_align_check, expand_strn_compare,
19362 expand_block_move, rs6000_output_load_multiple)
19363 Move functions related to string/block move/compare
19364 to a separate file.
19365 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19366 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19367 for this function which is now used in two files.
19368 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19369 * config.gcc: Add rs6000-string.o to extra_objs for
19370 targets powerpc*-*-* and rs6000*-*-*.
19372 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19375 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19376 32-bit, since indexed is not valid for DImode.
19377 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19378 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19379 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19380 (define_peephole2 for Altivec d-form store): Likewise.
19383 * multiple_target.c (create_dispatcher_calls): Only create the
19384 dispatcher call if the function is the default clone of a
19385 versioned function.
19387 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19389 PR middle-end/80902
19390 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19391 a call, force the call to not be a tail call.
19393 2017-06-23 Jeff Law <law@redhat.com>
19395 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19398 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19400 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19401 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19402 handling for early expansion of vector shifts (sl,sr,sra,rl).
19403 (builtin_function_type): Add vector shift right instructions
19404 to the unsigned argument list.
19406 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19408 rtl-optimizatoin/79286
19409 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19410 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19411 trap. PIC register plus a const unspec without offset can never trap.
19413 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19415 * tree.h (builtin_structptr_type): New type.
19416 (builtin_structptr_types): Declare new array.
19417 * tree.c (builtin_structptr_types): New array.
19418 (free_lang_data, build_common_tree_nodes): Use it.
19420 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19423 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19426 2017-06-22 Matt Turner <mattst88@gmail.com>
19428 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19429 Lake models to skylake case. Assume skylake for unknown
19430 models with clflushopt.
19432 2017-06-22 Jeff Law <law@redhat.com>
19434 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19435 frame sizes that do not satisfy aarch64_uimm12_shift.
19437 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19439 * profile-count.h (apply_probability,
19440 apply_scale, probability_in): Fix checks for zero.
19442 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19444 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19445 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19447 2016-06-22 Richard Biener <rguenther@suse.de>
19449 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19450 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19451 REDUC_MAX_EXPR support.
19452 (vectorizable_reduction): Likewise.
19453 (vect_create_epilog_for_reduction): Likewise.
19455 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19457 * match.pd (A / (1 << B) -> A >> B): New.
19458 * generic-match-head.c: Include optabs-tree.h.
19459 * gimple-match-head.c: Likewise.
19460 * optabs-tree.h (target_supports_op_p): New.
19461 * optabs-tree.c (target_supports_op_p): New.
19463 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19465 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19466 $gcc_cv_ld --help output.
19467 (gcc_cv_ld_demangle): Likewise.
19468 (gcc_cv_ld_eh_frame_hdr): Likewise.
19469 (gcc_cv_ld_pie): Likewise.
19470 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19471 (gcc_cv_ld_buildid): Likewise.
19472 (gcc_cv_ld_sysroot): Likewise.
19473 (ld_bndplt_support): Likewise.
19474 (ld_pushpopstate_support): Likewise.
19475 * configure: Regenerate.
19476 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19478 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19481 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19482 operands indexes to avoid gap between operands and match_dups.
19484 2017-06-21 Andrew Pinski <apinski@cavium.com>
19486 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19487 Increment Arith_shift and Arith_shift_reg by 1.
19488 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19490 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19491 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19492 (aarch64_strip_extend): Add new argument and test for it.
19493 (aarch64_cheap_mult_shift_p): New function.
19494 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19495 add a cost if it is true.
19496 Update calls to aarch64_strip_extend.
19497 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19499 2017-06-21 Andrew Pinski <apinski@cavium.com>
19501 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19503 (thunderxt88): Likewise.
19504 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19505 (thunderx_prefetch_tune): New variable.
19506 (thunderx2t99_prefetch_tune): Update for the correct values.
19507 (thunderxt88_tunings): New variable.
19508 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19509 generic_prefetch_tune.
19510 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19512 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19514 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19515 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19516 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19517 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19518 (aarch64_atomic_cas<mode>, GPI): Likewise.
19520 2017-06-21 Martin Liska <mliska@suse.cz>
19522 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19523 statements on cold and hot labels.
19524 * predict.c (tree_estimate_probability_bb): Remove the
19525 prediction from this place.
19527 2017-06-21 Martin Liska <mliska@suse.cz>
19529 PR tree-optimization/79489
19530 * gimplify.c (maybe_add_early_return_predict_stmt): New
19532 (gimplify_return_expr): Call the function.
19533 * predict.c (tree_estimate_probability_bb): Remove handling
19535 * predict.def: Update comment about early return predictor.
19536 * gimple-predict.h (is_gimple_predict): New function.
19537 * predict.def: Change default value of early return to 66.
19538 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19540 * passes.def: Put pass_strip_predict_hints to the beginning of
19543 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19545 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19546 FUNCTION_DECL declarations.
19547 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19549 (dwaf2out_decl): Likewise.
19550 * godump.c (go_early_global_decl): Skip call to the real debug hook
19551 for FUNCTION_DECL declarations.
19552 * passes.c (rest_of_decl_compilation): Skip call to the
19553 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19554 -fdump-go-spec is passed.
19556 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19558 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19559 Reorder for compactness.
19560 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19561 (def_builtin_pure, def_builtin_pure2): New functions.
19562 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19564 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19566 * match.pd (nop_convert): New predicate.
19567 ((A +- CST1) +- CST2): Allow some NOP conversions.
19569 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19572 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19573 with ctors/dtors if GOVD_SHARED is set.
19575 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19577 * config/aarch64/aarch64.md (movti_aarch64):
19578 Emit mov rather than orr.
19579 (movtf_aarch64): Likewise.
19580 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19581 Emit mov rather than orr.
19583 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19585 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19586 Swap alternatives, make integer dup more expensive.
19588 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19590 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19591 Return true for non-tls symbols.
19593 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19595 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19596 (cortex-a75): Likewise.
19597 (cortex-a75.cortex-a55): Likewise.
19598 * config/aarch64/aarch64-tune.md: Regenerate.
19599 * doc/invoke.texi (-mtune): Document new values for -mtune.
19601 2017-06-21 Tom de Vries <tom@codesourcery.com>
19603 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19604 stack_size feature.
19605 (Effective-Target Keywords, Other attributes): Suggest using
19606 dg-add-options stack_size feature to get stack limit in stack_size
19607 effective target documentation.
19609 2017-06-21 Julian Brown <julian@codesourcery.com>
19610 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19612 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19613 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19614 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19616 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19617 attribute type list for neon_multiply.
19618 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19619 attribute type list for neon_multiply.
19620 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19621 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19622 attribute type list for neon_multiply.
19623 * config/arm/types.md (crypto_pmull): Add.
19624 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19625 attribute type list.
19627 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19629 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19632 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19634 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19635 to make sure not to dereference a NULL cost_classes_ptr pointer.
19637 2017-06-20 Carl Love <cel@us.ibm.com>
19639 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19640 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19641 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19642 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19643 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19644 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19645 VMULOSW): New enum "unspec" values.
19646 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19647 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19648 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19649 altivec_vmulosw): New patterns.
19650 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19651 VMULOSW): Add definitions.
19653 2017-06-20 Julia Koval <julia.koval@intel.com>
19655 * config/i386/i386.c: Fix rounding expand for new pattern.
19656 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19658 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19660 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19661 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19663 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19665 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19668 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19670 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19671 architecture, then by implementer ID.
19672 * config/aarch64/aarch64-tune.md: Regenerate.
19674 2017-06-20 Richard Biener <rguenther@suse.de>
19676 PR middle-end/81097
19677 * fold-const.c (split_tree): Fold to type before negating.
19679 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19681 * diagnostic-show-locus.c
19682 (selftest::test_fixit_deletion_affecting_newline): New function.
19683 (selftest::diagnostic_show_locus_c_tests): Call it.
19685 2017-06-20 Andreas Schwab <schwab@suse.de>
19688 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19691 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19693 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19694 __ARM_FEATURE_COPROC according to support.
19696 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19698 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19699 Rewritten to avoid overflow for > 32-bit pointers.
19702 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19703 by removing enum keyword.
19704 (ubsan_type_descriptor): Likewise. Formatting fix.
19707 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19708 splitter): Require TARGET_SSE2 in the condition.
19710 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19713 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19714 for doing vector set of SFmode on ISA 3.0.
19715 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19716 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19718 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19719 SFmode value into a V4SF variable that was extracted from another
19720 V4SF variable without converting the element to double precision
19721 and back to single precision vector format.
19722 (vsx_insert_extract_v4sf_p9_2): Likewise.
19724 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19726 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19727 in UWHI to avoid undefined overflow.
19730 * ubsan.h (enum ubsan_encode_value_phase): New.
19731 (ubsan_encode_value): Change second argument to
19732 enum ubsan_encode_value_phase with default value of
19733 UBSAN_ENCODE_VALUE_GENERIC.
19734 * ubsan.c (ubsan_encode_value): Change second argument to
19735 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19736 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19737 create_tmp_var_raw instead of create_tmp_var and use a
19739 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19740 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19741 ubsan_encode_value callers.
19744 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19745 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19746 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19748 2017-06-19 Richard Biener <rguenther@suse.de>
19750 PR middle-end/81118
19751 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19752 estimates if we changed anything.
19754 2017-06-19 Richard Biener <rguenther@suse.de>
19756 PR tree-optimization/80887
19757 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19758 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19759 simplified lookups, then reset mprts_hook.
19760 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19762 (try_to_simplify): Likewise.
19764 2017-06-19 Martin Liska <mliska@suse.cz>
19767 * gimplify.c (gimplify_switch_expr):
19768 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19770 2017-06-19 Martin Liska <mliska@suse.cz>
19772 * doc/install.texi: Document that PGO runs in 4 stages.
19774 2017-06-19 Martin Liska <mliska@suse.cz>
19777 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19778 to dispatcher function name.
19779 * multiple_target.c (replace_function_decl): New function.
19780 (create_dispatcher_calls): Redirect both edges and references.
19782 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19784 * profile-count.c (profile_count::dump): Dump quality.
19785 (profile_count::differs_from_p): Update for unsigned val.
19786 * profile-count.h (profile_count_quality): New enum.
19787 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19789 2017-06-19 Richard Biener <rguenther@suse.de>
19791 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19792 struct function as arg.
19793 (estimate_numbers_of_iterations): Export overload with loop arg.
19794 (free_numbers_of_iterations_estimates_loop): Use an overload of
19795 free_numbers_of_iterations_estimates instead.
19796 * tree-cfg.c (remove_bb): Adjust.
19797 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19798 * tree-parloops.c (gen_parallel_loop): Likewise.
19799 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19801 (tree_unroll_loops_completely): Likewise.
19802 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19803 Use an overload instead and export.
19804 (estimated_loop_iterations): Adjust.
19805 (max_loop_iterations): Likewise.
19806 (likely_max_loop_iterations): Likewise.
19807 (estimate_numbers_of_iterations): Take struct function as arg
19809 (loop_exits_before_overflow): Adjust.
19810 (free_numbers_of_iterations_estimates_loop): Use an overload.
19811 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19812 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19814 2017-06-19 Richard Biener <rguenther@suse.de>
19817 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19818 RANGE_EXPR conservatively.
19820 2017-06-16 Carl Love <cel@us.ibm.com>
19822 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19823 definitions for vec_float, vec_float2, vec_floato,
19824 vec_floate built-ins.
19825 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19826 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19828 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19829 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19830 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19831 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19832 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19833 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19834 vec_floato): Add builtin defines.
19835 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19836 Update the built-in documentation file for the new built-in
19839 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19841 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19842 (mthumb): Mark as the negative of -marm.
19844 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19846 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19848 (ARM Options, -mtune): Document that this accepts the same
19849 extension options as -mcpu.
19850 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19852 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19854 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19855 permitted extensions.
19857 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19859 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19860 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19861 (armv8-m.main): Add option +nodsp.
19862 * config/arm/arm-cpu-cdata.h: Regenerated.
19864 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19866 * config/arm/t-fuchsia: New file.
19867 * config.gcc (arm*-*-fuchsia*): Use it.
19869 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19871 * config/arm/t-symbian: Rewrite for new option infrastructure.
19873 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19875 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19876 (MULTILIB_REQUIRED): Likewise.
19878 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19880 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19881 (MULTILIB_RESUE): Likewise.
19882 (MULTILIB_MATCHES): Likewise.
19883 (MULTLIB_REQUIRED): Likewise.
19885 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19887 * config/arm/t-rtems: Rewrite for new option framework.
19889 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19891 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19892 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19893 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19894 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19895 * config/arm/t-multilib: ... here.
19896 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19897 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19898 armv7-a and armv8*-a when A-profile libraries have not been built.
19899 * config/arm/t-rmprofile: Rewrite.
19901 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19903 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19904 with a backslash. Remove the backslash after substituting unescaped
19906 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19908 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19910 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19911 the driver to pass through the default setting of -mfloat-abi.
19912 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19914 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19915 (all_feat_combs): New rule.
19916 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19918 * config/arm/t-aprofile: Rewrite.
19920 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19922 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19923 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19924 fpu is not specified by the user/command-line.
19925 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19926 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19927 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19928 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19929 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19930 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19932 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19934 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19935 * config/arm/t-arm-elf: Rewritten.
19937 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19939 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19940 have some floating-point instructions.
19941 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19942 (TARGET_MAYBE_HARD_FLOAT): New macro.
19943 * config/arm/arm-builtins.c (arm_init_builtins): Use
19944 TARGET_MAYBE_HARD_FLOAT.
19945 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19947 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19949 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19950 (configargs.h): Include it.
19951 (arm_print_hint_for_fpu_option): New function.
19952 (arm_parse_fpu_option): New function.
19953 (candidate_extension): New class.
19954 (arm_canon_for_multilib): New function.
19955 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19956 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19957 (ARCH_CANONICAL_SPECS): New macro.
19958 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19962 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19963 are set after handling multilib fragments. Set target_cpu_default2
19966 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19968 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19970 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19971 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19973 (check_arch): Likewise.
19974 * config/arm/arm.c (arm_configure_build_target): Handle
19975 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19976 options in the default.
19978 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19980 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19981 when an option is an alias of another.
19982 * config/arm/parsecpu.awk (optalias): New parser token.
19983 (gen_comm_data): Mark non-alias options as such. Emit entries
19984 for extension aliases.
19985 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19986 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19987 (armv6kz, armv6zk, armv6t2): Likewise.
19988 (armv7): Make vfpv3-d16 an alias.
19989 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19991 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19992 Sort in canonical order.
19993 (armv8-a): Sort in canonical order.
19994 (armv8.1-a, armv8.2-a): Likewise.
19995 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19997 (cortex-a9): Sort in canonical order.
19998 * config/arm/arm.c (selftests.h): Include it.
19999 (arm_test_cpu_arch_data): New function.
20000 (arm_run_self_tests): New function.
20001 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20002 (targetm): Move declaration to the end of the file.
20003 * arm-cpu-cdata.h: Regenerated.
20005 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20007 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20008 call to target_mode_check describing the type of option passed.
20009 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20010 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20011 arm_parse_cpu_option_name to match parameters against list of
20013 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20014 arm_arch_core_flags data structure.
20015 * config/arm/arm-cpu_cdata.h: Regenerated.
20017 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20019 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20021 (arm_print_hint_for_cpu_option): Likewise.
20022 (arm_print_hint_for_arch_option): Likewise.
20023 (arm_parse_cpu_option_name): Likewise.
20024 (arm_parse_arch_option_name): Likewise.
20025 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20026 of entries in the all_fpus list.
20027 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20028 (arm_parse_cpu_option_name): Declare.
20029 (arm_parse_arch_option_name): Declare.
20030 (arm_parse_option_features): Declare.
20031 (arm_intialize_isa): Declare.
20032 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20034 (gen_comm_data): ... here. Make definitions non-static.
20035 * config/arm/arm-cpu-data.h: Regenerated.
20036 * config/arm/arm-cpu-cdata.h: Regenerated.
20038 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20040 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20041 (cpu_arch_extension): New structure.
20042 (cpu_arch_option, arch_option, cpu_option): New structures.
20043 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20044 architecture types.
20045 (gen_data): Generate new format data tables.
20046 * config/arm/arm.c (cpu_tune): New structure.
20047 (cpu_option, processors): Delete.
20048 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20049 (arm_print_hint_for_cpu_option): ... this and ...
20050 (arm_print_hint_for_arch_option): ... this.
20051 (arm_parse_arch_cpu_name): Delete. Replace with ...
20052 (arm_parse_cpu_option_name): ... this and ...
20053 (arm_parse_arch_option_name): ... this.
20054 (arm_unrecognized_feature): Change type of target parameter to
20056 (arm_parse_arch_cpu_features): Delete. Replace with ...
20057 (arm_parse_option_features): ... this.
20058 (arm_configure_build_target): Rework to use new configuration data
20060 (arm_print_tune_info): Rework for new configuration data tables.
20061 * config/arm/arm-cpu-data.h: Regenerated.
20062 * config/arm/arm-cpu.h: Regenerated.
20064 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20066 * Makefile.in (OBJS): Move sbitmap.o from here ...
20067 (OBJS-libcommon): ... to here.
20069 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20071 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20072 (ISA_ALL_CRYPTO): New macro.
20073 (ISA_ALL_SIMD): New macro
20074 (ISA_ALL_FP): New macro.
20075 * config/arm/arm.c (fpu_bitlist): Update initializer.
20076 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20078 (arm9e): Add fpu. Add option for nofp
20079 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20080 (arm926ej-s, arm1026ej-s): Likewise.
20081 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20082 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20083 neon-fp16, neon-vfpv4, nofp and nosimd.
20084 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20085 (cortex-a8): Add fpu. Add option for nofp.
20086 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20087 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20088 (cortex-r4f): Add fpu.
20089 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20090 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20092 (cortex-r8): Likewise.
20093 (cortex-m4): Add fpu. Add option for nofp.
20094 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20095 (cortex-a17.cortex-a7): Likewise.
20096 (cortex-a32): Add fpu. Add options for crypto and nofp.
20097 (cortex-a35, cortex-a53): Likewise.
20098 (cortex-a57): Add fpu. Add option for crypto.
20099 (cortex-a72, cortex-a73): Likewise.
20100 (exynos-m1): Likewise.
20101 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20102 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20103 (cortex-m33): Add fpu. Add option for nofp.
20104 * config/arm/arm-cpu-cdata.h: Regenerated
20105 * config/arm/arm-cpu-data.h: Regenerated.
20107 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20109 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20110 (armv5te, armv5tej): Likewise.
20111 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20112 (armv7): Add options fp and vfpv3-d16.
20113 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20114 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20116 (armv7ve): Likewise.
20117 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20118 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20119 (armv8-a): Add nocrypto option.
20120 (armv8.1-a, armv8.2-a): Likewise.
20121 (armv8-m.main): add options fp, fp.dp and nofp.
20123 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20125 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20127 (armv8-a+crc): Delete.
20128 (armv8.1-a): Add options simd, crypto and nofp.
20129 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20130 (armv8.2-a+fp16): Delete.
20131 (armv8-m.main): Add option dsp.
20132 (armv8-m.main+dsp): Delete.
20133 (cortex-a8): Add fpu. Add nofp option.
20134 (cortex-a9): Add fpu. Add nofp and nosimd options.
20135 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20136 link to main cpu and architecture data structures.
20137 (gen_comm_data): Only put isa attributes from the main architecture
20139 (option): New statement for architecture and CPU entries.
20140 * arm.c (struct cpu_option): New structure.
20141 (struct processors): Add entry for options.
20142 (arm_unrecognized_feature): New function.
20143 (arm_parse_arch_cpu_name): Ignore any characters after the first
20145 (arm_parse_arch_cpu_feature): New function.
20146 (arm_configure_build_target): Separate out any CPU and architecture
20147 features and parse separately. Don't error out if -mfpu=auto is
20148 used with only an architecture string.
20149 (arm_print_asm_arch_directives): New function.
20150 (arm_file_start): Call it.
20151 * config/arm/arm-cpu-cdata.h: Regenerated.
20152 * config/arm/arm-cpu-data.h: Likewise.
20153 * config/arm/arm-tables.opt: Likewise.
20155 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20157 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20158 assembler when it is not -mfpu=auto.
20160 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20162 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20163 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20164 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20165 (ASM_CPU_SPEC): Rewrite.
20166 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20167 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20168 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20169 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20170 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20171 copied string is NUL-terminated. Also strip any characters prefixed
20173 (arm_rewrite_selected_arch): New function.
20174 (arm_rewrite_march): New function.
20176 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20178 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20179 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20180 (march, mcpu, mtune): Convert to string-based options.
20181 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20182 (arm_parse_arch_cpu_name): New function.
20183 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20184 identify selected architecture or CPU.
20185 (arm_option_save): New function.
20186 (TARGET_OPTION_SAVE): Redefine.
20187 (arm_option_restore): Restore string options.
20188 (arm_option_print): Print string options.
20190 2017-06-16 Martin Sebor <msebor@redhat.com>
20192 PR tree-optimization/80933
20193 PR tree-optimization/80934
20194 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20195 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20196 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20197 (gimple_fold_builtin): Call them.
20199 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20201 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20202 as unlikely; update profile.
20204 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20206 * predict.c (force_edge_cold): Handle declaring edges impossible
20209 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20211 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20213 (try_unroll_loop_completely): Fix reporting.
20215 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20217 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20219 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20222 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20223 if given a non-constant argument for an intrinsic which requires a
20226 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20228 * profile.c (compare_freqs): New function.
20229 (branch_prob): Sort edge list.
20230 (find_spanning_tree): Assume that the list is priority sorted.
20232 2017-06-16 Richard Biener <rguenther@suse.de>
20234 PR tree-optimization/81090
20235 * passes.def (pass_record_bounds): Remove.
20236 * tree-pass.h (make_pass_record_bounds): Likewise.
20237 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20238 make_pass_record_bounds): Likewise.
20239 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20240 not free niter estimates at the beginning but at the end.
20241 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20243 2017-06-16 Richard Biener <rguenther@suse.de>
20245 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20246 initializer to workaround ICE in host GCC 4.8.
20248 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20250 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20252 (clone_inlined_nodes): Update.
20254 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20256 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20257 prefetch settings, and enable prefetching by default at -O3.
20259 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20261 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20262 Set flag_prefetch_loop_arrays according to tuning data.
20264 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20266 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20267 New tune structure.
20268 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20269 [Unrelated to main purpose of the patch] Place the pointer field last
20270 to enable type checking errors when tune structure are wrongly merged.
20271 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20272 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20273 (thunderx2t99_prefetch_tune): New tune constants.
20274 (tune_params *_tunings): Update all tunings (no functional change).
20275 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20276 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20277 from tunings structures.
20279 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20282 * ubsan.c (instrument_null): Add T argument, use it instead
20283 of computing it based on IS_LHS.
20284 (instrument_object_size): Likewise.
20285 (pass_ubsan::execute): Adjust instrument_null and
20286 instrument_object_size callers to pass gimple_get_lhs or
20287 gimple_assign_rhs1 result to it. Use instrument_null instead of
20288 calling get_base_address and instrument_mem_ref. Handle
20289 aggregate call arguments for object-size sanitization.
20291 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20293 PR tree-optimization/81089
20294 * tree-vrp.c (is_masked_range_test): Validate operands of
20297 2017-06-15 Martin Sebor <msebor@redhat.com>
20300 * dumpfile.c (dump_register): Avoid calling memset to initialize
20301 a class with a default ctor.
20302 * gcc.c (struct compiler): Remove const qualification.
20303 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20304 * hash-table.h: Ditto.
20305 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20307 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20308 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20310 * params.h (struct param_info): Make struct members non-const.
20311 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20312 with default initialization.
20313 * vec.h (vec_copy_construct, vec_default_construct): New helper
20315 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20316 with vec_copy_construct.
20317 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20318 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20319 * doc/invoke.texi (-Wclass-memaccess): Document.
20321 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20323 * emit-rtl.h (is_leaf): Update comment about local
20324 register allocator.
20326 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20329 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20330 for a variable to have a section before checking if the section has a
20332 Set section to.persistent if persistent attribute is set.
20333 Warn if .persistent attribute is used on an automatic variable.
20335 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20337 PR rtl-optimization/80474
20338 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20340 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20342 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20345 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20347 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20348 (casesi): Emit bounds checking as RTL.
20349 (casesi_internal_mips16_<mode>): Remove bounds checking.
20351 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20353 * config/xtensa/xtensa.c (xtensa_option_override): Append
20354 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20355 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20356 xtensa_doloop_hooks): Define unconditionally.
20357 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20359 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20360 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20361 for it in xtensa_option_override.
20362 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20363 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20365 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20367 * doc/cppopts.texi: Document '-' special value to -MF.
20369 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20371 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20372 (cortex_a53_fconst): Likewise.
20373 (cortex_a53_fpmul): Likewise.
20374 (cortex_a53_f_load_64): Likewise.
20375 (cortex_a53_f_load_many): Likewise.
20376 (cortex_a53_advsimd_alu): Likewise.
20377 (cortex_a53_advsimd_alu_q): Likewise.
20378 (cortex_a53_advsimd_mul): Likewise.
20379 (cortex_a53_advsimd_mul_q): Likewise.
20380 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20381 Add missing fmul, r2f_cvt and fconst cases.
20383 2017-06-14 Richard Biener <rguenther@suse.de>
20385 PR middle-end/81088
20386 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20388 (fold_binary_loc): When associating do not treat pre-existing
20389 TREE_OVERFLOW on literal constants as a reason to allow
20390 TREE_OVERFLOW on associated literal constants.
20392 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20394 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20395 (MASK_FEATURES): New macro.
20396 * config/sparc/sparc.c (sparc_option_override): Remove the special
20397 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20399 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20401 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20402 a division of 0 if non-call exceptions are enabled.
20404 2017-06-14 Andrew Pinski <apinski@cavium.com>
20405 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20408 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20409 Improve vector initialization code gen for only variable case.
20411 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20413 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20415 2017-06-14 Richard Biener <rguenther@suse.de>
20417 PR tree-optimization/81083
20418 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20421 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20423 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20424 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20425 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20426 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20427 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20428 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20430 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20432 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20433 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20435 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20437 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20439 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20441 * config/rs6000/t-rtems: Don't handle SPE.
20443 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20445 * config/rs6000/t-linux: Don't handle SPE.
20447 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20449 * config/rs6000/eabispe.h: Delete file.
20451 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20453 * config/rs6000/t-spe: Delete file.
20455 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20457 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20458 (rs6000_legitimate_offset_address_p): Return false for anything in
20459 V2SImode or V2SFmode.
20461 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20463 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20464 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20465 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20467 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20468 (rs6000_legitimate_offset_address_p): Ditto.
20469 (rs6000_emit_move): Ditto.
20470 (rs6000_init_builtins): Remove V4HI_type_node.
20472 2017-06-13 Martin Liska <mliska@suse.cz>
20475 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20476 (gate_asan): Likewise.
20477 * asan.h (asan_no_sanitize_address_p): Remove the function.
20478 (sanitize_flags_p): New function.
20479 * builtins.def: Fix coding style.
20480 * common.opt: Use renamed enum value.
20481 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20482 * doc/extend.texi: Document no_sanitize attribute.
20483 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20484 to SANITIZE_UNDEFINED_NONDEFAULT.
20485 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20486 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20487 Use sanitize_flags_p.
20488 * gimplify.c (gimplify_function_tree): Likewise.
20489 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20490 * opts.c (parse_no_sanitize_attribute): New function.
20491 (common_handle_option): Use renamed enum value.
20492 * opts.h (parse_no_sanitize_attribute): Declare.
20493 * tree.c (sanitize_flags_p): New function.
20494 * tree.h: Declared here.
20495 * tsan.c: Use sanitize_flags_p.
20496 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20497 (instrument_mem_ref): Likewise.
20498 (instrument_bool_enum_load): Likewise.
20499 (do_ubsan_in_current_function): Remove the function.
20500 (pass_ubsan::execute): Use sanitize_flags_p.
20501 * ubsan.h: Remove do_ubsan_in_current_function
20502 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20503 (dump_function_to_file): Use it here.
20505 2017-06-13 Martin Jambor <mjambor@suse.cz>
20507 PR tree-optimization/80803
20508 PR tree-optimization/81063
20509 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20510 (propagate_subaccesses_across_link): Enqueue subtree whenever
20511 necessary instead of relying on the caller.
20513 2017-06-13 Martin Jambor <mjambor@suse.cz>
20515 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20516 that have a first_link.
20517 (sort_and_splice_var_accesses): Do not check first_link before
20519 (subtree_mark_written_and_enqueue): Likewise.
20520 (propagate_all_subaccesses): Likewise and do not stop at first
20521 parent with a first_link.
20523 2017-06-13 Martin Jambor <mjambor@suse.cz>
20525 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20528 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20530 * match.pd: New pattern.
20532 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20534 * tree-vrp.c (is_masked_range_test): New function.
20535 (register_edge_assert_for): Determine ranges for
20538 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20540 PR tree-optimization/67328
20541 * fold-const.c (maskable_range_p): New function.
20542 (build_range_check): Generate bittests if possible.
20544 2017-06-13 Martin Liska <mliska@suse.cz>
20546 * gimple-pretty-print.c (dump_probability): Add new argument.
20547 (dump_edge_probability): Dump both probability and count.
20548 (dump_gimple_label): Likewise.
20549 (dump_gimple_bb_header): Likewise.
20551 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20554 * config/avr/avr-devices.c: Fix indentation.
20555 * config/avr/gen-avr-mmcu-specs.c: Dito.
20557 2017-06-13 Richard Biener <rguenther@suse.de>
20559 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20560 instead get vector type from stmt_info.
20561 (vectorizable_reduction): Adjust. Remove dead code.
20563 2017-06-13 Richard Biener <rguenther@suse.de>
20565 PR middle-end/81065
20566 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20567 case of C * (x * C2 + C3).
20568 (fold_addr_of_array_ref_difference): Properly fold index difference.
20570 2017-06-12 David S. Miller <davem@davemloft.net>
20573 * config/sparc/sparc.md (return expander): Emit frame blockage if
20574 function uses alloca.
20576 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20578 * combine.c (make_field_assignment): Check len rather than the mode
20579 precision when calling force_to_mode.
20581 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20583 Support multilibs and devices that see flash in RAM address range.
20586 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20587 (avr_mcu_t) <flash_pm_offset>: New field.
20588 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20589 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20590 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20591 (AVR_TINY_PM_OFFSET): Remove macro.
20592 * config/avr/avr.opt (-mshort-calls): New option.
20593 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20594 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20595 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20596 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20597 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20598 instead of avr_arch->have_jmp_call.
20599 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20600 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20601 avr_arch->flash_pm_offset to define.
20602 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20603 new field flash_pm_offset. Add entry for avrxmega3.
20604 (avr_texinfo): Add entry for avrxmega3.
20605 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20606 attiny212, attiny214,
20607 attiny412, attiny414, attiny416, attiny417,
20608 attiny814, attiny816, attiny817,
20609 attiny1614, attiny1616, attiny1617,
20610 attiny3214, attiny3216, attiny3217.
20611 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20612 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20613 (avr_print_operand_address) [AVR_TINY]: Same.
20614 (avr_asm_init_sections) <readonly_data_section>: Only patch
20615 callback if avr_arch->flash_pm_offset = 0.
20616 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20617 for rodata if avr_arch->flash_pm_offset != 0.
20618 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20619 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20620 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20621 (m_options): Append opt_rcall.
20622 (m_dirnames): Append dir_rcall.
20623 * config/avr/t-multilib: Regenerate.
20625 * configure.ac [target=avr]: Check whether avrxmega3 default
20626 linker description file works as needed.
20627 * configure: Regenerate.
20628 * doc/avr-mmcu.texi: Regenerate.
20629 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20630 <__AVR_ARCH__>: Document avrxmega3 and 103.
20631 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20632 <__AVR_SHORT_CALLS__>: Document it.
20633 <__AVR_PM_BASE_ADDRESS__>: Document it.
20634 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20635 (AVR Variable Attributes) <progmem>: Document this is
20636 not needed for avrxmega3.
20637 (AVR Named Address Spaces) <__flash>: Dito.
20639 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20641 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20643 2017-06-12 Doug Rupp <rupp@adacore.com>
20645 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20646 Append vxworks-stdint.h to the tm_file list.
20647 * config/vxworks-stdint.h: New file.
20649 2017-06-12 Martin Liska <mliska@suse.cz>
20651 PR tree-optimization/81041
20652 * tree-profile.c (gimple_gen_ic_func_profiler):
20653 Create an extra BB in profile-generate
20654 (gimple_gen_time_profiler): Likewise.
20656 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20658 PR tree-optimization/81003
20659 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20660 (update_range_test): Use it instead of force_gimple_operand_gsi.
20662 2017-06-12 Richard Biener <rguenther@suse.de>
20664 PR tree-optimization/81053
20665 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20666 with backedge value not defined in loop. Simplify def stmt
20669 2017-06-11 Tom de Vries <tom@codesourcery.com>
20672 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20674 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20675 nvptx_cannot_force_const_mem.
20677 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20679 * opts.c (finish_options): Move test for flag_split_stack after
20680 it has been initialized.
20682 2017-06-11 Jason Merrill <jason@redhat.com>
20684 * tree.h (id_equal): New.
20685 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20686 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20687 instead of strcmp of IDENTIFIER_POINTER.
20689 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20691 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20692 (mark_all_inlined_calls_cdtor): Fix formating.
20693 (inline_transform): Rescale profile before inlining.
20695 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20697 * cgraph.h (cgraph_edge::clone): Update prototype.
20698 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20699 (cgraph_node::create_clone): Update.
20700 (cgraph_node::create_version_clone): Update.
20701 * tree-inline.c (copy_bb): Update.
20702 (expand_call_inline): Update.
20704 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20706 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20707 factored out from ...
20708 (rs6000_emit_prologue): ... here.
20710 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20712 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20713 factored out from ...
20714 (rs6000_emit_prologue): ... here.
20716 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20718 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20720 (handle_missing_profiles): Fix computation of tp_first_run.
20721 (counts_to_freqs): Do not touch freqs when count is 0.
20723 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20725 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20728 2017-06-10 Tom de Vries <tom@codesourcery.com>
20730 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20731 attributes): Document signal effective target.
20733 2017-06-10 Tom de Vries <tom@codesourcery.com>
20735 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20736 Document effective target stack_size.
20738 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20740 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20741 to the edit_context if they can be auto-applied.
20743 2017-06-9 Ian Lance Taylor <iant@golang.org>
20745 * opts.c (finish_options): If -fsplit-stack, disable implicit
20746 -forder-blocks-and-partition.
20747 * doc/invoke.texi (Optimize Options): Document that when using
20748 -fsplit-stack -forder-blocks-and-partition is not implicitly
20751 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20753 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20754 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20755 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20756 * builtins.def (abort, trap, unreachable): Declare cold.
20757 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20758 * tree-core.h (ECF_COLD): New.
20759 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20760 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20762 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20764 * predict.c (unlikely_executed_stmt_p): Cleanup.
20766 2017-06-09 Richard Biener <rguenther@suse.de>
20768 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20769 model if the ref is always written to.
20771 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20773 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20775 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20777 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20780 2017-06-09 Tom de Vries <tom@codesourcery.com>
20783 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20784 "target cannot support label values" when encountering LABEL_REF.
20786 2017-06-09 Martin Liska <mliska@suse.cz>
20788 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20789 (gimple_gen_ic_func_profiler): Emit direct comparison
20790 of __gcov_indirect_call_callee with NULL.
20791 (gimple_gen_time_profiler): Change probability from
20792 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20794 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20796 * profile.c (edge_gcov_counts): Turn to pointer.
20797 (compute_branch_probabilities, compute_branch_probabilities): Update.
20798 (branch_prob): Do not clear edge_gcov_count.
20799 * profile.h (edge_gcov_counts): Turn to pointer.
20800 (edge_gcov_count): Update.
20802 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20804 * gimple.h (gimple_check_failed): Mark cold.
20806 2017-06-09 Richard Biener <rguenther@suse.de>
20808 PR tree-optimization/66623
20809 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20810 refactor check_reduction into two parts, properly computing
20811 whether we have to check reduction validity for outer loop
20814 2017-06-09 Richard Biener <rguenther@suse.de>
20816 PR tree-optimization/79483
20817 * graphite-scop-detection.c (order): New global.
20818 (get_order): Compute bb to order mapping that satisfies code
20819 generation constraints.
20820 (cmp_pbbs): New helper.
20821 (build_scops): Start domwalk at entry block, sort generated
20824 2017-06-09 Richard Biener <rguenther@suse.de>
20826 PR middle-end/81007
20827 * ipa-polymorphic-call.c
20828 (ipa_polymorphic_call_context::restrict_to_inner_class):
20829 Skip FIELD_DECLs with error_mark_node type.
20830 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20833 2017-06-09 Martin Liska <mliska@suse.cz>
20835 * predict.c (struct branch_predictor): New struct.
20836 (test_prediction_value_range): New test.
20837 (predict_c_tests): New function.
20838 * selftest-run-tests.c (selftest::run_tests): Run the function.
20839 * selftest.h: Declare new tests.
20841 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20844 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20845 gen_add3_insn did not fail.
20846 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20847 r0, construct that number in a temporary reg and add that reg to r0.
20848 If asked to put the result in r0 as well, fail.
20850 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20852 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20853 for early expansion of vec_eqv.
20855 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20857 PR middle-end/81005
20858 * ubsan.c (instrument_null): Avoid pointless code temporary.
20859 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20861 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20865 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20868 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20871 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20873 * predict.c (unlikely_executed_edge_p): Move ahead.
20874 (probably_never_executed_edge_p): Use it.
20876 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20878 PR middle-end/79988
20879 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20880 gimple_call_builtin_p call.
20882 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20884 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20885 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20886 rtl_check_failed_type2, rtl_check_failed_code1,
20887 rtl_check_failed_code2, rtl_check_failed_code_mode,
20888 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20889 rtvec_check_failed_bounds, rtl_check_failed_flag,
20890 _fatal_insn_not_found, _fatal_insn): Likewise.
20891 * tree.h (tree_contains_struct_check_failed,
20892 tree_check_failed, tree_not_check_failed,
20893 tree_class_check_failed, tree_range_check_failed,
20894 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20895 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20896 tree_operand_check_failed, omp_clause_check_failed,
20897 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20900 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20902 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20903 flag_branch_probabilities.
20904 * ipa-inline.c (edge_badness): Likewise.
20905 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20906 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20907 * predict.c (maybe_hot_frequency_p): Likewise.
20908 (probably_never_executed): Likewise.
20909 * sched-ebb.c (schedule_ebbs): Likewise.
20910 * sched-rgn.c (find_single_block_region): Likewise.
20911 * tracer.c (tail_duplicate): Likewise.
20913 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20915 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20916 longer requires x_flag_profile_use.
20918 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20920 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20921 instead of flag_reorder_blocks_and_partition.
20922 * dbxout.c (dbxout_function_end): Likewise.
20923 * dwarf2out.c (gen_subprogram_die): Likewise.
20924 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20925 * hw-doloop.c (reorg_loops): Likewise.
20926 * varasm.c (assemble_start_function,
20927 assemble_end_function): Likewise.
20928 (decide_function_section): Do not check for
20929 flag_reorder_blocks_and_partition.
20931 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20933 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20935 (chkp_get_hard_register_fake_addr_expr): Ditto.
20936 (chkp_build_addr_expr): Add check for hard reg case.
20937 (chkp_parse_array_and_component_ref): Ditto.
20938 (chkp_find_bounds_1): Ditto.
20939 (chkp_process_stmt): Don't generate bounds store for
20942 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20944 * predict.c (maybe_hot_bb_p): Do not check profile status.
20945 (maybe_hot_edge_p): Likewise.
20946 (probably_never_executed): Check for zero counts even if profile
20948 (unlikely_executed_edge_p): New function.
20949 (unlikely_executed_stmt_p): New function.
20950 (unlikely_executed_bb_p): New function.
20951 (set_even_probabilities): Use unlikely predicates.
20952 (combine_predictions_for_bb): Likewise.
20953 (predict_paths_for_bb): Likewise.
20954 (predict_paths_leading_to_edge): Likewise.
20955 (determine_unlikely_bbs): New function.
20956 (estimate_bb_frequencies): Use it.
20957 (compute_function_frequency): Use zero counts even if profile is
20959 * profile-count.h: Fix typo.
20961 2017-08-08 Julia Koval <julia.koval@intel.com>
20963 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20964 _mm512_mask_cvtsepi16_storeu_epi8,
20965 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20966 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20967 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20968 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20969 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20970 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20971 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20972 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20973 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20974 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20975 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20976 __builtin_ia32_pmovuswb256mem_mask,
20977 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20978 __builtin_ia32_pmovwb512mem_mask): New builtins.
20980 2017-08-08 Julia Koval <julia.koval@intel.com>
20982 PR target/73350,80862
20983 * config/i386/subst.md (round): Fix round pattern.
20984 * config/i386/i386.c (ix86_erase_embedded_rounding):
20985 Fix erasing rounding for the fixed pattern.
20987 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20989 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20991 2017-06-08 Martin Liska <mliska@suse.cz>
20993 PR gcov-profile/80911
20994 * gcov.c (block_info::block_info): New constructor.
20996 2017-06-07 Carl Love <cel@us.ibm.com>
20998 * config/rs6000/rs6000-c: The return type of the following
20999 built-in functions was implemented as int not long long. Fix sign
21000 of return value for the unsigned version of vec_mulo and vec_mule.
21001 vector unsigned long long vec_bperm (vector unsigned long long,
21002 vector unsigned char)
21003 vector signed long long vec_mule (vector signed int,
21005 vector unsigned long long vec_mule (vector unsigned int,
21006 vector unsigned int)
21007 vector signed long long vec_mulo (vector signed int,
21009 vector unsigned long long vec_mulo (vector unsigned int,
21010 vector unsigned int)
21011 * doc/extend.texi: Fix the documentation for the built-in
21014 2017-06-07 Carl Love <cel@us.ibm.com>
21017 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21020 2017-06-07 Carl Love <cel@us.ibm.com>
21022 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21023 support, Generate doublehv for signed int/float for BE case only.
21025 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21027 * doc/invoke.texi (mcx16): Rewrite.
21029 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21031 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21032 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21033 *mov<mode>_softfloat, and an anonymous splitter): Use
21034 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21036 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21038 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21040 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21041 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21042 (rs6000_debug_reg_global): Adjust.
21043 (rs6000_init_hard_regno_mode_ok): Adjust.
21044 (rs6000_dbx_register_number): Adjust.
21045 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21046 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21047 Remove SPE_ACC and SPEFSCR.
21048 (REG_ALLOC_ORDER): Ditto.
21049 (FRAME_POINTER_REGNUM): Change to 111.
21050 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21051 (REG_CLASS_NAMES): Ditto.
21052 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21053 (REGISTER_NAMES): Ditto.
21054 (ADDITIONAL_REG_NAMES): Ditto.
21055 (rs6000_reg_names): Ditto.
21056 * config/rs6000/rs6000.md: Renumber some register number
21059 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21061 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21063 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21064 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21066 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21067 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21068 Delete the SPE high registers.
21069 (REG_ALLOC_ORDER): Ditto.
21070 (enum reg_class): Remove SPE_HIGH_REGS.
21071 (REG_CLASS_NAMES): Ditto.
21072 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21073 (REGISTER_NAMES): Ditto.
21074 (rs6000_reg_names): Ditto.
21075 * doc/tm.texi.in: Remove SPE as example.
21076 * doc/tm.texi: Regenerate.
21078 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21080 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21081 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21082 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21083 * config/rs6000/rs6000.md (type): Remove "brinc".
21085 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21087 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21088 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21089 * config/rs6000/linuxspe.h: Delete file.
21090 * config/rs6000/rs6000.md: Don't include spe.md.
21091 * config/rs6000/spe.h: Delete file.
21092 * config/rs6000/spe.md: Delete file.
21093 * config/rs6000/t-rs6000: Remove spe.md.
21095 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21097 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21098 (reg_or_none500mem_operand): Delete.
21099 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21100 instead of reg_or_none500mem_operand.
21102 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21104 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21105 handling of SPE flags.
21106 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21108 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21110 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21112 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21113 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21114 paired_divv2sf3): Similar.
21115 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21116 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21117 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21119 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21120 Rename the paired_* instruction patterns.
21121 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21123 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21124 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21125 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21126 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21127 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21128 PAIRED_VECTOR_MODE.
21129 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21130 (spe_func_has_64bit_regs_p): Delete.
21131 (spe_expand_predicate_builtin): Delete.
21132 (spe_expand_evsel_builtin): Delete.
21133 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21134 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21135 (invalid_e500_subreg): Delete.
21136 (rs6000_legitimize_address): Always force_reg op2 as well, for
21137 paired single memory accesses.
21138 (rs6000_member_type_forces_blk): Delete.
21139 (rs6000_spe_function_arg): Delete.
21140 (rs6000_expand_unop_builtin): Delete SPE handling.
21141 (rs6000_expand_binop_builtin): Ditto.
21142 (spe_expand_stv_builtin): Delete.
21143 (bdesc_2arg_spe): Delete.
21144 (spe_expand_builtin): Delete.
21145 (spe_expand_predicate_builtin): Delete.
21146 (spe_expand_evsel_builtin): Delete.
21147 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21148 (spe_init_builtins): Delete.
21149 (spe_func_has_64bit_regs_p): Delete.
21150 (savres_routine_name): Delete "info" parameter. Adjust callers.
21151 (rs6000_emit_stack_reset): Ditto.
21152 (rs6000_dwarf_register_span): Delete.
21153 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21154 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21155 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21157 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21159 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21160 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21161 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21162 mulv2sf3, divv2sf3): Delete expanders.
21164 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21166 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21168 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21170 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21171 * config/rs6000/rs6000.c: Ditto.
21173 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21175 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21176 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21177 comparison_operator.
21179 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21181 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21182 * config/rs6000/rs6000.opt: Ditto.
21183 * config/rs6000/t-rtems: Ditto.
21185 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21187 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21188 TARGET_E500_SINGLE by 0, simplify.
21189 * config/rs6000/rs6000.c: Ditto.
21190 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21191 (spe_build_register_parallel): Delete.
21192 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21193 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21194 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21195 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21196 (E500_CONVERT): Delete.
21197 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21199 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21201 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21202 * config/rs6000/dfp.md: Ditto.
21203 (negdd2, *negdd2_fpr): Merge.
21204 (absdd2, *absdd2_fpr): Merge.
21205 (negtd2, *negtd2_fpr): Merge.
21206 (abstd2, *abstd2_fpr): Merge.
21207 * config/rs6000/e500.h: Delete file.
21208 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21209 TARGET_FPRS by 1 and simplify.
21210 * config/rs6000/rs6000-c.c: Ditto.
21211 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21212 TARGET_DF_SPE by 0.
21213 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21215 * config/rs6000/rs6000.md: Ditto.
21216 (floatdidf2, *floatdidf2_fpr): Merge.
21217 (move_from_CR_gt_bit): Delete.
21218 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21219 (E500_CR_IOR_COMPARE): Delete.
21220 (All patterns that require !TARGET_FPRS): Delete.
21221 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21223 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21225 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21227 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21229 * graphds.c (add_edge): Intitialize edge's attached data.
21230 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21231 pointer parameter. Call pointed function on each edge during
21232 graph traversing. Skip traversing the edge when the function
21234 (graphds_dfs, graphds_scc): Ditto.
21235 (for_each_edge): New parameter. Pass the new parameter to callback
21237 * graphds.h (skip_edge_callback): New function pointer type.
21238 (graphds_dfs, graphds_scc): New function pointer parameter.
21239 (graphds_edge_callback, for_each_edge): New parameter.
21241 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21243 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21244 out code checking if runtime alias check is possible to below ...
21245 Call the new function.
21246 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21247 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21249 2017-06-07 Marek Polacek <polacek@redhat.com>
21252 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21253 TYPE_OVERFLOW_WRAPS check.
21255 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21257 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21258 if versioning is required.
21259 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21260 peeling with the check for versioning.
21262 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21264 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21265 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21266 Set true to new parameter if new ssa variable is defined.
21267 (vect_gen_vector_loop_niters): Refactor. Set range information
21268 for the new vector loop bound variable.
21269 (vect_do_peeling): Ditto.
21271 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21273 * tree-affine.c (ssa.h): Include header file.
21274 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21275 has wrapping overflow behavior.
21277 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21279 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21281 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21283 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21284 (tree_to_aff_combination): ... here.
21286 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21288 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21289 reg_pressure model function.
21290 (ivopts_global_cost_for_size): Delete.
21291 (determine_set_costs, iv_ca_recount_cost): Call new model function
21292 ivopts_estimate_reg_pressure.
21294 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21296 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21297 expensive than udiv. Remove floating point cases from mod.
21299 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21301 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21302 Increase idiv cost.
21304 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21306 * config/aarch64/aarch64.md
21307 (copysignsf3): Fix mask generation.
21309 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21311 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21313 (class dump_manager): Add register_dumps method.
21314 * dumpfile.c: Include langhooks.h.
21315 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21316 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21317 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21318 (dump_manager::dump_register): Start with 512 entries instead of 32.
21319 (dump_manager::register_dumps): New method.
21320 * toplev.c (general_init): Instead of invoking register_dumps
21321 langhook, invoke register_dumps method on the dump manager.
21322 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21325 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21327 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21328 Say that # requires an associated define_split to exist, and that
21329 the define_split must be suitable for use after register allocation.
21331 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21333 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21334 (compute_outgoing_frequencies): Also initialize zero counts.
21335 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21336 around loops; preserve more of profile when nothing changes.
21338 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21340 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21342 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21343 * config/arm/arm-cpu-cdata.h: Regenerate.
21344 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21345 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21346 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21347 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21348 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21350 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21352 * config/arm/t-rmprofile: Likewise.
21353 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21355 2017-06-06 David S. Miller <davem@davemloft.net>
21358 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21359 blockage if function uses alloca.
21361 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21363 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21364 New "uid" fields to hold pretty-print IDs of group and ref.
21365 Memory references are now identified as <group_id>:<ref_id>
21366 instead of using [random] addresses.
21367 (dump_mem_details): Simplify, no functional change.
21368 (dump_mem_ref): Simplify and make output more concise.
21369 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21370 (find_or_create_group): Initialize group uid.
21371 (record_ref): Initialize ref uid. Improve debug output.
21372 (prune_group_by_reuse, should_issue_prefetch_p,)
21373 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21374 (mark_nontemporal_store, determine_loop_nest_reuse):
21375 Improve debug output.
21377 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21379 * dbgcnt.def (prefetch): New debug counter.
21380 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21381 (schedule_prefetches): Stop issueing prefetches if debug counter
21384 2017-06-06 Tom de Vries <tom@codesourcery.com>
21386 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21387 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21389 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21391 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21392 Use aarch64_reg_or_zero predicate for operand 4.
21393 (aarch64_compare_and_swap<mode> define_insn_and_split):
21394 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21395 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21397 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21399 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21400 (arm_compute_save_core_reg_mask): This.
21401 (thumb1_compute_save_reg_mask): Rename into ...
21402 (thumb1_compute_save_core_reg_mask): This.
21403 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21404 (arm_compute_frame_layout): Likewise.
21406 2017-06-06 Richard Biener <rguenther@suse.de>
21408 PR tree-optimization/80974
21409 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21410 keep or clear leaders SSA info.
21412 2017-06-06 Tom de Vries <tom@codesourcery.com>
21414 * config/nvptx/nvptx.c (split_mode_p): New function.
21415 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21417 2017-06-06 Tom de Vries <tom@codesourcery.com>
21419 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21421 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21424 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21427 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21429 * shrink-wrap.c (handle_simple_exit): Update profile.
21430 (try_shrink_wrapping): Upate profile.
21432 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21434 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21435 (tree_guess_outgoing_edge_probabilities): New.
21436 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21437 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21439 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21441 * ipa-split.c (split_function): Initialize return bb profile.
21443 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21445 * profile.c (compute_branch_probabilities): Also initialize
21446 EXIT_BLOCK profile.
21448 2017-06-06 Richard Biener <rguenther@suse.de>
21450 PR tree-optimization/80928
21451 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21452 (vect_analyze_loop_operations): Properly guard analysis for
21454 (vect_transform_loop): Likewise.
21455 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21456 (vect_model_induction_cost): Do not cost for pure SLP.
21457 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21458 of induction in inner loop vectorization.
21459 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21460 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21461 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21463 (vect_analyze_slp_cost_1): Cost induction.
21464 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21465 (vect_get_slp_vect_defs): Likewise.
21466 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21467 (vect_transform_stmt): Handle SLP reductions.
21468 * tree-vectorizer.h (vectorizable_induction): Adjust.
21470 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21472 * config/rs6000/rs6000.c (make_resolver_func): Update
21473 init_lowered_empty_function call.
21475 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21477 * doc/invoke.texi: Document the -fprofile-abs-path option.
21478 * common.opt (fprofile-abs-path): New option.
21479 * gcov-io.h (gcov_write_filename): Declare.
21480 * gcov-io.c (gcov_write_filename): New function.
21481 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21482 * profile.c (output_location): Likewise.
21484 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21486 * shring-wrap.c: Revert accidental commit.
21488 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21490 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21492 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21494 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21496 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21497 profile in callgraph edge.
21498 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21499 (apply_scale): Likewise.
21500 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21501 Also scale profile when inlining function with zero profile.
21502 (initialize_cfun): Update exit block profile even when it is zero.
21503 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21504 when profile is read.
21506 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21508 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21509 (CLONE_*): New constants to define the processors we can generate
21510 code for with the target_clone attribute.
21511 (rs6000_clone_map): New array to identify which clone processors
21512 the current program is running on.
21513 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21514 target_clone attribute.
21515 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21516 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21517 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21518 (cpu_expand_builtin): Add support for target_clone attribute.
21519 (rs6000_valid_attribute_p): Allow "default" attribute.
21520 (get_decl_name): New debug function to simplify printing the
21521 current function name in debugging statements.
21522 (rs6000_clone_priority): New functions to support the target_clone
21523 attribute, and be able to generate code to switch between ISA 2.05
21524 through ISA 3.0 (power6 through power9).
21525 (rs6000_compare_version_priority): Likewise.
21526 (rs6000_get_function_versions_dispatcher): Likewise.
21527 (make_resolver_func): Likewise.
21528 (add_condition_to_bb): Likewise.
21529 (dispatch_function_versions): Likewise.
21530 (rs6000_generate_version_dispatcher_body): Likewise.
21531 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21532 (fusion_gpr_load_p): Fix a spacing issue.
21533 * doc/extend.texi (Common Function Attributes): Document that the
21534 PowerPC supports the target_clone attribute.
21536 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21538 * config/arm/arm.h: explain F symbol found in description of ARM
21539 register allocation in its legend.
21541 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21543 * config/mips/frame-header-opt.c: Include profile-count.h.
21544 * config/riscv/riscv.c: Include profile-count.h
21546 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21548 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21550 (sm_set_flag_if_changed): Add bbs field.
21551 (execute_sm_if_changed_flag_set): Pass BBS.
21552 (execute_sm): Update.
21554 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21556 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21559 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21561 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21562 (peephole2): New peephole2 to emit the above.
21563 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21565 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21567 * config/aarch64/aarch64.c (define_peephole2 above
21568 *sub_<shift>_<mode>): New peephole.
21570 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21572 * config/i386/i386.c (make_resolver_func): Update.
21573 * Makefile.in: Add profile-count.h and profile-count.o
21574 * auto-profile.c (afdo_indirect_call): Update to new API.
21575 (afdo_set_bb_count): Update.
21576 (afdo_propagate_edge): Update.
21577 (afdo_propagate_circuit): Update.
21578 (afdo_calculate_branch_prob): Update.
21579 (afdo_annotate_cfg): Update.
21580 * basic-block.h: Include profile-count.h
21581 (struct edge_def): Turn count to profile_count.
21582 (struct basic_block_def): Likewie.
21583 (REG_BR_PROB_BASE): Move to profile-count.h
21584 (RDIV): Move to profile-count.h
21585 * bb-reorder.c (max_entry_count): Turn to profile_count.
21586 (find_traces): Update.
21587 (rotate_loop):Update.
21588 (connect_traces):Update.
21589 (sanitize_hot_paths):Update.
21590 * bt-load.c (migrate_btr_defs): Update.
21591 * cfg.c (RDIV): Remove.
21592 (init_flow): Use alloc_block.
21593 (alloc_block): Uninitialize count.
21594 (unchecked_make_edge): Uninitialize count.
21595 (check_bb_profile): Update.
21596 (dump_edge_info): Update.
21597 (dump_bb_info): Update.
21598 (update_bb_profile_for_threading): Update.
21599 (scale_bbs_frequencies_int): Update.
21600 (scale_bbs_frequencies_gcov_type): Update.
21601 (scale_bbs_frequencies_profile_count): New.
21602 * cfg.h (update_bb_profile_for_threading): Update.
21603 (scale_bbs_frequencies_profile_count): Declare.
21604 * cfgbuild.c (compute_outgoing_frequencies): Update.
21605 (find_many_sub_basic_blocks): Update.
21606 * cfgcleanup.c (try_forward_edges): Update.
21607 (try_crossjump_to_edge): Update.
21608 * cfgexpand.c (expand_gimple_tailcall): Update.
21609 (construct_exit_block): Update.
21610 * cfghooks.c (verify_flow_info): Update.
21611 (dump_bb_for_graph): Update.
21612 (split_edge): Update.
21613 (make_forwarder_block): Update.
21614 (duplicate_block): Update.
21615 (account_profile_record): Update.
21616 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21617 (get_estimated_loop_iterations): Update.
21618 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21619 (single_likely_exit): Update.
21620 * cfgloopmanip.c (scale_loop_profile): Update.
21622 (set_zero_probability): Update.
21623 (lv_adjust_loop_entry_edge): Update.
21624 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21625 (purge_dead_edges): Update.
21626 (rtl_account_profile_record): Update.
21627 * cgraph.c (cgraph_node::create): Uninitialize count.
21628 (symbol_table::create_edge): Uninitialize count.
21629 (cgraph_update_edges_for_call_stmt_node): Update.
21630 (cgraph_edge::dump_edge_flags): Update.
21631 (cgraph_node::dump): Update.
21632 (cgraph_edge::maybe_hot_p): Update.
21633 * cgraph.h: Include profile-count.h
21634 (create_clone), create_edge, create_indirect_edge): Update.
21635 (cgraph_node): Turn count to profile_count.
21636 (cgraph_edge0: Likewise.
21637 (make_speculative, clone): Update.
21638 (create_edge): Update.
21639 (init_lowered_empty_function): Update.
21640 * cgraphclones.c (cgraph_edge::clone): Update.
21641 (duplicate_thunk_for_node): Update.
21642 (cgraph_node::create_clone): Update.
21643 * cgraphunit.c (cgraph_node::analyze): Update.
21644 (cgraph_node::expand_thunk): Update.
21645 * final.c (dump_basic_block_info): Update.
21646 * gimple-streamer-in.c (input_bb): Update.
21647 * gimple-streamer-out.c (output_bb): Update.
21648 * graphite.c (print_global_statistics): Update.
21649 (print_graphite_scop_statistics): Update.
21650 * hsa-brig.c: Include basic-block.h.
21651 * hsa-dump.c: Include basic-block.h.
21652 * hsa-gen.c (T sum_slice): Update.
21653 (convert_switch_statements):Update.
21654 * hsa-regalloc.c: Include basic-block.h.
21655 * ipa-chkp.c (chkp_produce_thunks): Update.
21656 * ipa-cp.c (struct caller_statistics): Update.
21657 (init_caller_stats): Update.
21658 (gather_caller_stats): Update.
21659 (ipcp_cloning_candidate_p): Update.
21660 (good_cloning_opportunity_p): Update.
21661 (get_info_about_necessary_edges): Update.
21662 (dump_profile_updates): Update.
21663 (update_profiling_info): Update.
21664 (update_specialized_profile): Update.
21665 (perhaps_add_new_callers): Update.
21666 (decide_about_value): Update.
21667 (ipa_cp_c_finalize): Update.
21668 * ipa-devirt.c (struct odr_type_warn_count): Update.
21669 (struct decl_warn_count): Update.
21670 (struct final_warning_record): Update.
21671 (possible_polymorphic_call_targets): Update.
21672 (ipa_devirt): Update.
21673 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21674 * ipa-icf.c (sem_function::merge): Update.
21675 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21676 * ipa-inline.c (compute_uninlined_call_time): Update.
21677 (compute_inlined_call_time): Update.
21678 (want_inline_small_function_p): Update.
21679 (want_inline_self_recursive_call_p): Update.
21680 (edge_badness): Update.
21681 (lookup_recursive_calls): Update.
21682 (recursive_inlining): Update.
21683 (inline_small_functions): Update.
21684 (dump_overall_stats): Update.
21685 (dump_inline_stats): Update.
21686 * ipa-profile.c (ipa_profile_generate_summary): Update.
21687 (ipa_propagate_frequency): Update.
21688 (ipa_profile): Update.
21689 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21690 * ipa-utils.c (ipa_merge_profiles): Update.
21691 * loop-doloop.c (doloop_modify): Update.
21692 * loop-unroll.c (report_unroll): Update.
21693 (unroll_loop_runtime_iterations): Update.
21694 * lto-cgraph.c (lto_output_edge): Update.
21695 (lto_output_node): Update.
21696 (input_node): Update.
21697 (input_edge): Update.
21698 (merge_profile_summaries): Update.
21699 * lto-streamer-in.c (input_cfg): Update.
21700 * lto-streamer-out.c (output_cfg): Update.
21701 * mcf.c (create_fixup_graph): Update.
21702 (adjust_cfg_counts): Update.
21703 (sum_edge_counts): Update.
21704 * modulo-sched.c (sms_schedule): Update.
21705 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21706 * predict.c (maybe_hot_count_p): Update.
21707 (probably_never_executed): Update.
21708 (dump_prediction): Update.
21709 (combine_predictions_for_bb): Update.
21710 (propagate_freq): Update.
21711 (handle_missing_profiles): Update.
21712 (counts_to_freqs): Update.
21713 (rebuild_frequencies): Update.
21714 (force_edge_cold): Update.
21715 * predict.h: Include profile-count.h
21716 (maybe_hot_count_p, counts_to_freqs): UPdate.
21717 * print-rtl-function.c: Do not include cfg.h
21718 * print-rtl.c: Include basic-block.h
21719 * profile-count.c: New file.
21720 * profile-count.h: New file.
21721 * profile.c (is_edge_inconsistent): Update.
21722 (correct_negative_edge_counts): Update.
21723 (is_inconsistent): Update.
21724 (set_bb_counts): Update.
21725 (read_profile_edge_counts): Update.
21726 (compute_frequency_overlap): Update.
21727 (compute_branch_probabilities): Update; Initialize and deinitialize
21729 (branch_prob): Update.
21730 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21731 (edge_gcov_count): New.
21732 (bb_gcov_count): New.
21733 * shrink-wrap.c (try_shrink_wrapping): Update.
21734 * tracer.c (better_p): Update.
21735 * trans-mem.c (expand_transaction): Update.
21736 (ipa_tm_insert_irr_call): Update.
21737 (ipa_tm_insert_gettmclone_call): Update.
21738 * tree-call-cdce.c: Update.
21739 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21740 (gimple_duplicate_sese_tail): Update.
21741 (gimple_account_profile_record): Update.
21742 (execute_fixup_cfg): Update.
21743 * tree-inline.c (copy_bb): Update.
21744 (copy_edges_for_bb): Update.
21745 (initialize_cfun): Update.
21746 (freqs_to_counts): Update.
21747 (copy_cfg_body): Update.
21748 (expand_call_inline): Update.
21749 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21750 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21751 (try_unroll_loop_completely): Update.
21752 (try_peel_loop): Update.
21753 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21754 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21755 * tree-ssa-loop-split.c (connect_loops): Update.
21756 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21757 * tree-ssa-reassoc.c (branch_fixup): Update.
21758 * tree-ssa-tail-merge.c (replace_block_by): Update.
21759 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21760 (compute_path_counts): Update.
21761 (update_profile): Update.
21762 (recompute_probabilities): Update.
21763 (update_joiner_offpath_counts): Update.
21764 (estimated_freqs_path): Update.
21765 (freqs_to_counts_path): Update.
21766 (clear_counts_path): Update.
21767 (ssa_fix_duplicate_block_edges): Update.
21768 (duplicate_thread_path): Update.
21769 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21770 (struct switch_conv_info): Update.
21771 * tree-tailcall.c (decrease_profile): Update.
21772 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21773 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21774 * value-prof.c (check_counter): Update.
21775 (gimple_divmod_fixed_value): Update.
21776 (gimple_mod_pow2): Update.
21777 (gimple_mod_subtract): Update.
21778 (gimple_ic_transform): Update.
21779 (gimple_stringop_fixed_value): Update.
21780 * value-prof.h (gimple_ic): Update.
21782 2017-06-02 Carl Love <cel@us.ibm.com>
21784 * config/rs6000/rs6000-c: Add support for built-in functions
21785 vector double vec_doublee (vector signed int);
21786 vector double vec_doublee (vector unsigned int);
21787 vector double vec_doublee (vector float);
21788 vector double vec_doubleh (vector signed int);
21789 vector double vec_doubleh (vector unsigned int);
21790 vector double vec_doubleh (vector float);
21791 vector double vec_doublel (vector signed int);
21792 vector double vec_doublel (vector unsigned int);
21793 vector double vec_doublel (vector float);
21794 vector double vec_doubleo (vector signed int);
21795 vector double vec_doubleo (vector unsigned int);
21796 vector double vec_doubleo (vector float);.
21797 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21798 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21800 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21801 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21802 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21804 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21805 vec_doublel, vec_doubleh.
21806 * doc/extend.texi: Update the built-in documentation file for the
21807 new built-in functions.
21809 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21812 * ipa-inline-analysis.c (free_growth_caches): Set
21813 edge_removal_hook_holder to NULL after removing it.
21815 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21817 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21818 comparision with zero.
21820 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21821 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21822 for early expansion of vec_min and vec_max builtins.
21823 (builtin_function_type): Add min/max unsigned variants to those
21824 identified as having unsigned arguments.
21826 2017-06-02 Olivier Hainque <hainque@adacore.com>
21828 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21830 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21832 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21833 Use VALL_F16 iterator rather than VALL.
21835 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21837 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21838 Emit CBNZ inside loop when doing a strong exchange and comparing
21839 against zero. Generate the CC flags after the loop.
21841 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21843 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21844 (dl_section_ref): New.
21845 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21846 On AIX, append an expression to subtract the size of the
21847 section length to dl_section_ref.
21849 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21851 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21852 for early expansion of vector absolute builtins.
21854 2017-06-02 Richard Biener <rguenther@suse.de>
21856 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21857 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21859 2017-06-02 Richard Biener <rguenther@suse.de>
21861 PR tree-optimization/80948
21862 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21863 stmt order as well.
21865 2017-06-02 Richard Biener <rguenther@suse.de>
21867 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21869 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21870 for inductions as relevant.
21872 2017-06-02 Richard Biener <rguenther@suse.de>
21874 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21875 (vectorizable_induction): ... this. Remove dead code.
21877 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21879 * builtins. (expand_builtin_alloca): Remove second parameter and
21880 infer its value from the first parameter instead.
21881 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21883 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21885 PR rtl-optimization/80903
21886 * loop-doloop.c (add_test): Unshare sequence.
21888 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21890 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21892 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21894 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21896 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21897 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21898 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21899 (xlogue_layout::get_stub_rtx): Make static.
21900 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21901 (xlogue_layout::compute_stub_managed_regs): Rename to...
21902 (xlogue_layout::count_stub_managed_regs): ...this.
21903 (xlogue_layout::is_stub_managed_reg): New function.
21904 (xlogue_layout::m_stub_names): Rename to...
21905 (xlogue_layout::s_stub_names): ...this, make static.
21906 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21907 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21908 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21909 xlogue_layout::s_stub_names): Instantiate statics.
21910 (stub_managed_regs): Remove.
21911 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21912 (disable_call_ms2sysv_xlogues): Rename to...
21913 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21914 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21916 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21917 change after reload_completed.
21918 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21920 (ix86_expand_prologue): Likewise.
21921 (ix86_expand_epilogue): Likewise.
21922 (ix86_expand_split_stack_prologue): Likewise.
21923 (ix86_compute_frame_layout): Remove frame parameter ...
21924 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21925 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21927 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21928 (ix86_frame): Move from here ...
21929 * config/i386/i386.h (ix86_frame): ... to here.
21930 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21931 complete ix86_frame data structure instead. Remove some_ld_name.
21933 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21935 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21936 symbols that hold a DECL_VALUE_EXPR.
21938 2017-06-01 Martin Jambor <mjambor@suse.cz>
21940 PR tree-optimization/80898
21941 * tree-sra.c (process_subtree_disqualification): Removed.
21942 (disqualify_candidate): Do not acll
21943 process_subtree_disqualification.
21944 (subtree_mark_written_and_enqueue): New function.
21945 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21946 RHS has been disqualified and re-queue LHS if necessary. Apart
21947 from that, ignore disqualified RHS.
21949 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21951 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21952 address fetch for z10 or later.
21954 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21956 * config/arc/arc.md (tst_movb): Add guard when splitting.
21958 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21960 * config/arc/arc.c (arc_can_eliminate): Test against
21961 arc_frame_pointer_needed.
21963 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21965 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21966 to prevent store reordering.
21967 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21968 (type): Add block type.
21969 (stack_tie): Define special instruction to be used in
21972 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21974 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21975 constraint. It is not valid for the pattern.
21976 (noncommutative_binary_comparison): Likewise.
21978 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21980 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21983 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21985 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21986 be used by the reg-alloc.
21988 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21990 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21991 reg-alloc when having mul64 or mul32x16 instructions.
21992 (mulsidi3): Likewise.
21993 (umulsidi3): Likewise.
21994 (mulsi32x16): New pattern.
21995 (mulsi64): Likewise.
21996 (mulsidi64): Likewise.
21997 (umulsidi64): Likewise.
21998 (MUL32x16_REG): Define.
21999 (mul64_600): Use MUL32x16_REG.
22000 (mac64_600): Likewise.
22001 (umul64_600): Likewise.
22002 (umac64_600): Likewise.
22004 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22006 * config/arc/arc.md (mulsi3_700): Make it commutative.
22008 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22010 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22012 (*sign_extendsidi2_insn): Likewise for movstosw.
22014 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22016 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22017 the type of the input discriminant value. Convert the
22018 discriminant value of signedness vary.
22020 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22022 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22023 Add to -Wall section.
22025 2017-06-01 Richard Biener <rguenther@suse.de>
22027 PR middle-end/66313
22028 * fold-const.c (fold_plusminus_mult_expr): If the factored
22029 factor may be zero use a wrapping type for the inner operation.
22030 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22031 and handle moved defs.
22032 (process_assignment): Properly guard the unary op case. Return a
22033 tri-state indicating that moving the stmt before the call may allow
22034 to continue. Pass through to_move.
22035 (find_tail_calls): Handle moving unrelated defs before
22038 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22041 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22042 splitter result in the canonical way.
22044 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22046 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22047 also for 32bit target. Update insn attributes.
22048 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22050 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22052 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22053 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22054 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22055 (_mm_maskz_min_ss): New intrinsics.
22057 2017-05-31 Martin Liska <mliska@suse.cz>
22059 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22060 Change comment style to one we normally use.
22061 (vectorizable_reduction): Likewise.
22062 (vectorizable_induction): Likewise.
22063 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22064 (vectorizable_call): Likewise.
22065 (vectorizable_simd_clone_call): Likewise.
22066 (vectorizable_conversion): Likewise.
22067 (vectorizable_assignment): Likewise.
22068 (vectorizable_shift): Likewise.
22069 (vectorizable_operation): Likewise.
22070 (vectorizable_store): Likewise.
22071 (vectorizable_load): Likewise.
22072 * tree-vectorizer.h: Likewise.
22074 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22076 * passes.c (emergency_dump_function): New.
22077 * tree-pass.h (emergency_dump_function): Declare.
22078 * plugin.c (plugins_internal_error_function): Remove.
22079 * plugin.h (plugins_internal_error_function): Remove declaration.
22080 * toplev.c (internal_error_function): New static function. Use it...
22081 (general_init): ...here.
22083 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22085 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22086 (arc_rtx_costs): Add costs for new patterns.
22087 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22088 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22090 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22092 * tree-ssa-strlen.c (get_next_strinfo): New function.
22093 (get_stridx_plus_constant): Use it.
22094 (zero_length_string): Likewise.
22095 (adjust_related_strinfos): Likewise.
22096 (adjust_last_stmt): Likewise.
22098 2017-05-31 Richard Biener <rguenther@suse.de>
22101 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22102 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22104 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22106 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22107 loop_vinfo argument and use of dependence distance vectors.
22108 Check instead whether the two references differ only in their
22109 initial value and assume that they have the same alignment if the
22110 difference is a multiple of the vector alignment.
22111 (vect_analyze_data_refs_alignment): Update call accordingly.
22113 2017-05-31 Martin Liska <mliska@suse.cz>
22116 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22118 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22120 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22121 (create_intersect_range_checks): Move from ...
22122 * tree-data-ref.c (create_intersect_range_checks_index)
22123 (create_intersect_range_checks): ... to here.
22124 (create_runtime_alias_checks): New function factored from ...
22125 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22126 here. Call above function.
22127 * tree-data-ref.h (create_runtime_alias_checks): New function.
22129 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22131 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22132 segment length for dr_b and compute it in wide_int.
22134 2017-05-31 Richard Biener <rguenther@suse.de>
22136 PR tree-optimization/80906
22137 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22138 and pass through iv_map.
22139 (copy_bb_and_scalar_dependences): Adjust.
22140 (translate_pending_phi_nodes): Likewise.
22141 (copy_loop_close_phi_args): Handle code-generating IVs instead
22144 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22146 * diagnostic-color.c (color_dict): Add "type-diff".
22147 (parse_gcc_colors): Update comment.
22148 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22149 -fdiagnostics-show-template-tree and -fno-elide-type.
22150 (GCC_COLORS): Add type-diff to example.
22152 (-fdiagnostics-show-template-tree): New.
22153 (-fno-elide-type): New.
22154 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22155 the pp_format_decoder callback. Call any m_format_postprocessor's
22157 (pretty_printer::pretty_printer): Initialize
22158 m_format_postprocessor.
22159 (pretty_printer::~pretty_printer): Delete any
22160 m_format_postprocessor.
22161 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22162 (class format_postprocessor): New class.
22163 (struct pretty_printer::format_decoder): Document the new parameters.
22164 (struct pretty_printer::m_format_postprocessor): New field.
22165 * tree-diagnostic.c (default_tree_printer): Update for new
22166 bool and const char ** params.
22167 * tree-diagnostic.h (default_tree_printer): Likewise.
22169 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22171 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22172 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22173 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22174 rs6000_gen_cell_microcode code.
22175 (rs6000_final_prescan_insn): Delete.
22176 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22177 "warn-cell-microcode" entries.
22178 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22179 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22180 throughout. Change cc_reg_not_micro_cr0_operand to
22181 cc_reg_not_cr0_operand throughout.
22182 (*extendhi<mode>2_noload): Delete.
22183 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22184 (mwarn-cell-microcode): Delete.
22185 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22186 -mgen-cell-microcode and -mwarn-cell-microcode.
22188 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22191 * config/i386/constraints.md (Yd): New constraint.
22193 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22194 and (?Yd, r) alternatives. Update insn attributes.
22195 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22196 and (?*Yd, r) alternatives. Update insn attributes.
22197 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22199 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22201 * gimplify.c (gimplify_modify_expr): Don't create a
22202 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22205 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22207 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22209 2017-05-30 Richard Biener <rguenther@suse.de>
22211 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22212 and reduc_def fields.
22213 (STMT_VINFO_REDUC_TYPE): New define.
22214 (STMT_VINFO_REDUC_DEF): Likewise.
22215 (vect_force_simple_reduction): Adjust prototype.
22216 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22217 (vect_is_simple_reduction): Remove check_reduction argument.
22218 (vect_force_simple_reduction): Adjust and set
22219 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22220 (vectorizable_reduction): Do not re-do reduction analysis
22221 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22222 * tree-parloops.c (gather_scalar_reductions): Adjust.
22224 2017-05-30 Richard Biener <rguenther@suse.de>
22226 PR middle-end/80901
22227 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22230 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22232 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22233 Introduce unknown_misalignment parameter and remove vf.
22234 (vect_peeling_hash_get_lowest_cost):
22235 Pass unknown_misalignment parameter.
22236 (vect_enhance_data_refs_alignment):
22237 Fix unsupportable data ref treatment.
22239 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22241 * tree-vect-data-refs.c (vect_get_data_access_cost):
22242 Workaround for SLP handling.
22243 (vect_enhance_data_refs_alignment):
22244 Compute costs for doing no peeling at all, compare to the best
22245 peeling costs so far and avoid peeling if cheaper.
22247 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22249 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22250 Return peeling info and set costs to zero for unlimited cost
22252 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22253 with unknown misalignment. Compute and costs for unknown
22254 misalignment, compare them to the costs for known misalignment
22255 and choose the cheapest for peeling.
22257 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22259 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22260 (vect_get_peeling_costs_all_drs): Create function.
22261 (vect_peeling_hash_get_lowest_cost):
22262 Use vect_get_peeling_costs_all_drs.
22263 (vect_peeling_supportable): Create function.
22264 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22266 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22268 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22269 DR_HAS_NEGATIVE_STEP.
22270 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22271 (vect_enhance_data_refs_alignment): Use.
22272 (vect_duplicate_ssa_name_ptr_info): Use.
22273 * tree-vectorizer.h (dr_misalignment): Use.
22274 (known_alignment_for_access_p): Use.
22276 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22279 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22281 (has_section_name): New function.
22283 2017-05-30 Martin Liska <mliska@suse.cz>
22286 * auto-profile.c (get_function_decl_from_block): Fix
22289 2017-05-30 Richard Biener <rguenther@suse.de>
22291 PR middle-end/80876
22292 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22294 2017-05-30 Martin Liska <mliska@suse.cz>
22296 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22297 * dumpfile.h (struct dump_file_info): Remove ctors.
22299 2017-05-30 Martin Liska <mliska@suse.cz>
22301 * predict.def: Fix GNU coding style.
22303 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22305 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22306 Mark 'to' argument with ATTRIBUTE_UNUSED.
22308 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22310 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22311 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22312 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22315 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22317 * doc/install.texi (Options specification): Restore entry of
22318 --enable-sjlj-exceptions.
22320 2017-05-27 Michael Eager <eager@eagercon.com>
22323 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22325 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22327 * config/microblaze/microblaze.h
22328 (FIXED_REGISTERS): Update in macro.
22329 (CALL_USED_REGISTERS): Update in macro.
22331 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22333 * doc/install.texi: Add links to macOS binary distributions.
22335 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22339 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22341 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22343 2017-05-26 Martin Liska <mliska@suse.cz>
22345 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22347 2017-05-26 Martin Liska <mliska@suse.cz>
22349 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22350 always leading ';; '.
22351 (dump_bb_info): Likewise.
22352 (brief_dump_cfg): Likewise.
22353 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22354 * dumpfile.c: Remove usage of TDF_VERBOSE.
22355 * dumpfile.h (enum dump_kind): Likewise.
22356 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22357 * print-tree.c (debug_verbose): Remove.
22358 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22359 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22360 * tree-diagnostic.c (default_tree_printer): Replace
22361 TDF_DIAGNOSTIC with TDF_SLIM.
22363 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22365 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22366 in parameter loop, rather than loop_vinfo.
22367 (create_intersect_range_checks): Ditto.
22368 (vect_create_cond_for_alias_checks): Update call to above functions.
22370 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22372 PR tree-optimization/80815
22373 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22374 for merging runtime alias checks. Handle negative DR_STEPs.
22376 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22378 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22380 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22381 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22382 out code pruning runtime alias checks.
22383 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22384 factored out from above.
22385 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22387 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22389 (prune_runtime_alias_test_list): New decalaration.
22391 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22393 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22394 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22395 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22396 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22397 (operator==, comp_dr_with_seg_len_pair): Ditto.
22398 (vect_prune_runtime_alias_test_list): Ditto.
22400 2017-05-26 Martin Liska <mliska@suse.cz>
22403 * params.def: Bound partial-inlining-entry-probability param.
22405 2017-05-26 Marek Polacek <polacek@redhat.com>
22408 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22411 2017-05-26 Richard Biener <rguenther@suse.de>
22413 PR tree-optimization/80842
22414 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22417 2017-05-26 Richard Biener <rguenther@suse.de>
22419 PR tree-optimization/80844
22420 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22422 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22424 * doc/md.texi (Machine Constraints): Update x86 family
22425 machine constraints section to match 'config/i386/constraints.md'.
22427 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22429 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22431 2017-05-25 Nathan Sidwell <nathan@acm.org>
22433 * doc/invoke.texi (--enable-languages): Update documentation.
22435 2017-05-25 Martin Liska <mliska@suse.cz>
22437 * dumpfile.c: Add TDF_FOLDING.
22438 * dumpfile.h (enum dump_kind): Likewise.
22439 * genmatch.c (dt_simplify::gen_1): Use it.
22441 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22443 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22445 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22447 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22448 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22450 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22452 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22453 * match.pd (X == C): Rewrite it here.
22454 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22455 with_certain_nonzero_bits2): New predicates.
22456 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22458 2017-05-24 Nathan Sidwell <nathan@acm.org>
22460 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22463 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22466 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22468 * config/powerpcspe: New port. Files are copied from the rs6000
22469 port, with "rs6000" in filenames replaced by "powerpcspe".
22471 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22473 PR rtl-optimization/80754
22474 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22476 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22478 * config/sparc/sparc.md (length): Return the correct value for -mflat
22479 sibcalls to match output_sibcall.
22481 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22485 * config/rs6000/rs6000.c (struct machine_function): Add new field
22487 (rs6000_get_separate_components): Init that field, use it.
22488 (rs6000_components_for_bb): Use the field.
22490 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22492 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22494 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22496 PR middle-end/80823
22497 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22499 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22502 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22503 address against address_operand predicate.
22504 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22506 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22508 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22509 parameters passed indirectly.
22511 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22513 * config/i386/i386.md (*movdi_internal): Remove SSE4
22514 alternative 18 (?r, *v). Update insn attributes.
22515 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22516 Update insn attributes.
22517 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22518 Update insn attributes.
22519 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22520 alternative 1 (r, v). Remove isa attribute.
22521 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22522 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22523 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22525 2017-05-23 Tom de Vries <tom@codesourcery.com>
22527 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22530 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22532 * cgraphunit.c (symbol_table::process_new_functions): Update.
22533 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22534 (inline_generate_summary): Rename to ...
22535 (ipa_fn_summary_generate): ... this one.
22536 (inline_read_summary): Rename to ...
22537 (ipa_fn_summary_read): ... this one.
22538 (inline_write_summary): Rename to ...
22539 (ipa_fn_summary_write): ... this one.
22540 (inline_free_summary): Rename to ...
22541 (ipa_free_fn_summary): ... this one.
22542 (pass_data_local_fn_summary, pass_local_fn_summary,
22543 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22544 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22545 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22546 make_pass_ipa_fn_summary): New.
22547 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22548 inline_write_summary, inline_free_summary): Remove.
22549 (ipa_free_fn_summary) : New.
22550 * ipa-inline.c (ipa_inline): Update.
22551 (pass_ipa_inline): Do not generate summaries.
22552 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22554 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22555 and add pass_ipa_fn_summary.
22556 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22558 (make_pass_inline_parameters): Remove.
22560 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22562 * omp-low.c (struct omp_context): Remove "default_kind" member.
22565 * omp-offload.c (execute_oacc_device_lower): Remove the
22566 parallelism dimensions function attributes for unparallelized
22567 OpenACC kernels constructs.
22569 2017-05-23 Martin Liska <mliska@suse.cz>
22571 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22573 (cgraph_edge::make_speculative): Likewise.
22574 (cgraph_edge::resolve_speculation): Likewise.
22575 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22576 (cgraph_node::dump): Likewise.
22577 * cgraph.h: Likewise.
22578 * cgraphunit.c (analyze_functions): Likewise.
22579 (symbol_table::compile): Likewise.
22580 * ipa-cp.c (print_all_lattices): Likewise.
22581 (determine_versionability): Likewise.
22582 (initialize_node_lattices): Likewise.
22583 (ipcp_verify_propagated_values): Likewise.
22584 (estimate_local_effects): Likewise.
22585 (update_profiling_info): Likewise.
22586 (create_specialized_node): Likewise.
22587 (perhaps_add_new_callers): Likewise.
22588 (decide_about_value): Likewise.
22589 (decide_whether_version_node): Likewise.
22590 (identify_dead_nodes): Likewise.
22591 (ipcp_store_bits_results): Likewise.
22592 * ipa-devirt.c (dump_targets): Likewise.
22593 (ipa_devirt): Likewise.
22594 * ipa-icf.c (sem_item::dump): Likewise.
22595 (sem_function::equals): Likewise.
22596 (sem_variable::equals): Likewise.
22597 (sem_item_optimizer::read_section): Likewise.
22598 (sem_item_optimizer::execute): Likewise.
22599 (congruence_class::dump): Likewise.
22600 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22601 (dump_inline_summary): Likewise.
22602 (estimate_node_size_and_time): Likewise.
22603 (inline_analyze_function): Likewise.
22604 * ipa-inline-transform.c (inline_call): Likewise.
22605 * ipa-inline.c (report_inline_failed_reason): Likewise.
22606 (want_early_inline_function_p): Likewise.
22607 (edge_badness): Likewise.
22608 (update_edge_key): Likewise.
22609 (inline_small_functions): Likewise.
22610 * ipa-profile.c (ipa_profile): Likewise.
22611 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22612 (ipa_make_edge_direct_to_target): Likewise.
22613 (remove_described_reference): Likewise.
22614 (ipa_impossible_devirt_target): Likewise.
22615 (propagate_controlled_uses): Likewise.
22616 (ipa_print_node_params): Likewise.
22617 (ipcp_transform_function): Likewise.
22618 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22619 (propagate_pure_const): Likewise.
22620 * ipa-reference.c (generate_summary): Likewise.
22621 (read_write_all_from_decl): Likewise.
22622 (propagate): Likewise.
22623 (ipa_reference_read_optimization_summary): Likewise.
22624 * ipa-utils.c (ipa_merge_profiles): Likewise.
22625 * ipa.c (walk_polymorphic_call_targets): Likewise.
22626 (symbol_table::remove_unreachable_nodes): Likewise.
22627 (ipa_single_use): Likewise.
22628 * passes.c (execute_todo): Likewise.
22629 * predict.c (drop_profile): Likewise.
22630 * symtab.c (symtab_node::get_dump_name): New function.
22631 (symtab_node::dump_name): Likewise.
22632 (symtab_node::dump_asm_name): Likewise.
22633 (symtab_node::dump_references): Likewise.
22634 (symtab_node::dump_referring): Likewise.
22635 (symtab_node::dump_base): Likewise.
22636 (symtab_node::debug_symtab): Likewise.
22637 * tree-sra.c (convert_callers_for_node): Likewise.
22638 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22639 * value-prof.c (init_node_map): Likewise.
22641 2017-05-23 Martin Liska <mliska@suse.cz>
22643 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22644 and symtab_node::debug_symtab to symbol_table::debug.
22645 * cgraphunit.c (analyze_functions): Use the renamed function.
22646 (symbol_table::compile): Likewise.
22647 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22648 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22649 * passes.c (execute_todo): Likewise.
22650 * symtab.c (symbol_table::dump): New function.
22651 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22653 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22655 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22656 that nonconst implies exec.
22658 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22660 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22661 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22662 (inline_edge_summary_vec): Turn into ...
22663 (ipa_call_summaries): ... this one.
22664 (redirect_to_unreachable, edge_set_predicate,
22665 evaluate_properties_for_edge, inline_summary_alloc,
22666 reset_ipa_call_summary, reset_inline_summary,
22667 inline_summary_t::duplicate): Update.
22668 (inline_edge_duplication_hook): Turn to ...
22669 (ipa_call_summary_t::duplicate): ... this one.
22670 (inline_edge_removal_hook): Turn to ...
22671 (ipa_call_summary_t::remove): ... this one.
22672 (dump_inline_edge_summary): Turn to ...
22673 (dump_ipa_call_summary): ... this one.
22674 (estimate_function_body_sizes): Update.
22675 (inline_update_callee_summaries): Update.
22676 (remap_edge_change_prob): Update.
22677 (remap_edge_summaries): Update.
22678 (inline_merge_summary): Update.
22679 (do_estimate_edge_time): Update.
22680 (inline_generate_summary): Update.
22681 (inline_read_section): Update.
22682 (inline_read_summary): Update.
22683 (inline_free_summary): Update.
22684 * ipa-inline.c (can_inline_edge_p): Update.
22685 (compute_inlined_call_time): Update.
22686 (want_inline_small_function_p): Update.
22687 (edge_badness): Update.
22688 (early_inliner): Update.
22689 * ipa-inline.h (inline_edge_summary): Turn to ...
22690 (ipa_call_summary): ... this one.
22691 (ipa_call_summary_t): New class.
22692 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22693 (ipa_call_summaries): New.
22694 (inline_edge_summary): Remove.
22695 (estimate_edge_growth): Update.
22696 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22697 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22698 * ipa-split.c (execute_split_functions): Update.
22699 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22701 2017-05-23 Tom de Vries <tom@codesourcery.com>
22703 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22704 attributes): Document rdrand effective target.
22706 2017-05-23 Tom de Vries <tom@codesourcery.com>
22708 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22709 attributes): Sort alphabetically.
22711 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22713 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22715 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22718 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22719 V2DF/V2DI splat into two separate patterns, one that handles
22720 registers, and the other that only handles memory. Drop support
22721 for splatting from a GPR on ISA 2.07 and then splitting the
22722 splat into direct move and splat.
22723 (vsx_splat_<mode>_reg): Likewise.
22724 (vsx_splat_<mode>_mem): Likewise.
22726 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22728 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22730 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22732 PR middle-end/80809
22733 * omp-low.c (finish_taskreg_remap): New function.
22734 (finish_taskreg_scan): If unit size of ctx->record_type
22735 is non-constant, unshare the size expression and replace
22736 decls in it with possible outer var refs.
22738 PR middle-end/80809
22739 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22740 GOVD_SHARED rather than GOVD_PRIVATE with it.
22741 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22742 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22744 PR middle-end/80853
22745 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22746 as last argument to build_outer_var_ref for pointer bases of array
22747 section reductions.
22749 2017-05-19 Martin Sebor <msebor@redhat.com>
22751 * print-tree.c (print_node): Print DECL_READ_P flag.
22753 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22755 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22756 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22757 * cgraph.c: Likewise.
22758 * cgraphunit.c: Likewise.
22759 * gengtype.c: Likewise.
22760 * ipa-cp.c: Likewise.
22761 * ipa-devirt.c: Likewise.
22762 * ipa-icf.c: Likewise.
22763 * ipa-predicate.c: Likewise.
22764 * ipa-profile.c: Likewise.
22765 * ipa-prop.c: Likewise.
22766 * ipa-split.c: Likewise.
22768 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22769 edge_predicate_pool, dump_inline_hints,
22770 inline_summary::account_size_time, redirect_to_unreachable,
22771 edge_set_predicate, set_hint_predicate,
22772 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22773 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22774 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22775 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22776 ipa_call_summary_t::remove, initialize_growth_caches,
22777 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22778 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22779 mark_modified, unmodified_parm_1, unmodified_parm,
22780 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22781 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22782 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22783 will_be_nonconstant_predicate, record_modified_bb_info,
22784 get_minimal_bb, record_modified, param_change_prob,
22785 phi_result_unknown_predicate, predicate_for_phi_result,
22786 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22787 estimate_function_body_sizes, compute_inline_parameters,
22788 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22789 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22790 inline_update_callee_summaries, remap_edge_change_prob,
22791 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22792 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22793 inline_analyze_function, inline_summary_t::insert,
22794 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22795 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22796 inline_free_summary): Move to ipa-fnsummary.h
22797 (predicate_t): Remove.
22798 * ipa-fnsummary.c: New file.
22799 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22800 (enum inline_hints_vals, inline_hints, agg_position_info,
22801 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22802 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22803 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22804 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22805 inline_read_summary, inline_write_summary, inline_free_summary,
22806 inline_analyze_function, initialize_inline_failed,
22807 inline_merge_summary, inline_update_overall_summary,
22808 compute_inline_parameters): Move to ipa-fnsummary.h
22809 * ipa-fnsummary.h: New file.
22810 * ipa-inline-transform.h: Include ipa-inline.h.
22811 * ipa-inline.c: LIkewise.
22813 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22815 * ipa-inline.c (edge_badness): Use inlined_time instead of
22816 inline_summaries->get.
22818 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22820 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22822 2017-05-22 Nathan Sidwell <nathan@acm.org>
22824 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22825 (fdump-lang): Document 'raw' option.
22826 * dumpfile.h (TDI_tu): Delete.
22827 * dumpfile.c (dump_files): Remove translation-unit.
22828 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22830 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22832 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22833 command option from $(AWK) call.
22834 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22835 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22836 [FORMAT]: Remove handling of variable.
22837 * config/avr/t-multilib: Regenerate.
22839 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22841 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22843 (dump_inline_summary): Do not print self_time.
22844 (estimate_function_body_sizes): Do not set self_time.
22845 (compute_inline_parameters): Likewise.
22846 (inline_read_section, inline_write_summary): Do not stream self_time.
22847 * ipa-inline.h (inline_summary): Drop self_time.
22849 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22851 * ipa-inline-analysis.c (account_size_time): Rename to ...
22852 (inline_summary::account_size_time): ... this one.
22853 (reset_ipa_call_summary): Turn to ...
22854 (ipa_call_summary::reset): ... this one.
22855 (reset_inline_summary): Turn to ...
22856 (inline_summary::reset): ... this one.
22857 (inline_summary_t::remove): Update.
22858 (inline_summary_t::duplicate): Update.
22859 (ipa_call_summary_t::remove): Update.
22860 (dump_inline_summary): Update.
22861 (estimate_function_body_sizes): Update.
22862 (compute_inline_parameters): Update.
22863 (estimate_node_size_and_time): Update.
22864 (inline_merge_summary): Update.
22865 (inline_update_overall_summary): Update.
22866 (inline_read_section): Update.
22867 (inline_write_summary): Update.
22868 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22869 add account_size_time and reset member functions.
22870 (ipa_call_summary): Add reset function.
22871 * ipa-predicate.h (predicate::operator &): Constify.
22873 2017-05-22 Richard Biener <rguenther@suse.de>
22875 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22877 2017-05-19 Jason Merrill <jason@redhat.com>
22879 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22881 2017-05-19 Marek Polacek <polacek@redhat.com>
22884 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22885 TYPE_OVERFLOW_WRAPS checks.
22887 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22889 * tree-core.h (enum omp_clause_default_kind): Add
22890 "OMP_CLAUSE_DEFAULT_PRESENT".
22891 * tree-pretty-print.c (dump_omp_clause): Handle it.
22892 * gimplify.c (enum gimplify_omp_var_data): Add
22893 "GOVD_MAP_FORCE_PRESENT".
22894 (gimplify_adjust_omp_clauses_1): Map it to
22895 "GOMP_MAP_FORCE_PRESENT".
22896 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22898 * gimplify.c (oacc_default_clause): Clarify.
22900 2017-05-19 Nathan Sidwell <nathan@acm.org>
22902 LANG_HOOK_REGISTER_DUMPS
22903 * toplev.c (general_init): Call register dump lang hook.
22904 * doc/invoke.texi: Document -fdump-lang option family.
22905 * dumpfile.c (dump_files): Remove class dump here.
22906 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22907 * dumpfile.h (tree_dump_index): Remove TDI_class.
22908 * langhooks-def.h (lhd_register_dumps): Declare.
22909 (LANG_HOOKS_REGISTER_DUMPS): Define.
22910 (LANG_HOOKS_INITIALIZER): Add it.
22911 * langhooks.c (lhd_register_dumps): Define.
22912 * langhooks.h (struct lang_hooks): Add register_dumps.
22914 2017-05-19 Nathan Sidwell <nathan@acm.org>
22916 * context.h (context::set_passes): New.
22917 * context.c (context::context): Do not create pass manager.
22918 * toplev.c (general_init): Create pass manager here.
22920 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22922 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22923 use this splitter if two add or or instructions would also work for
22924 the constant we want to generate.
22926 2017-05-19 Richard Biener <rguenther@suse.de>
22929 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22930 predicate evaluation.
22932 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22934 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22936 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22939 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22941 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22942 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22943 (inline_edge_summary_vec): Turn into ...
22944 (ipa_call_summaries): ... this one.
22945 (redirect_to_unreachable, edge_set_predicate,
22946 evaluate_properties_for_edge, inline_summary_alloc,
22947 reset_ipa_call_summary, reset_inline_summary,
22948 inline_summary_t::duplicate): Update.
22949 (inline_edge_duplication_hook): Turn to ...
22950 (ipa_call_summary_t::duplicate): ... this one.
22951 (inline_edge_removal_hook): Turn to ...
22952 (ipa_call_summary_t::remove): ... this one.
22953 (dump_inline_edge_summary): Turn to ...
22954 (dump_ipa_call_summary): ... this one.
22955 (estimate_function_body_sizes): Update.
22956 (inline_update_callee_summaries): Update.
22957 (remap_edge_change_prob): Update.
22958 (remap_edge_summaries): Update.
22959 (inline_merge_summary): Update.
22960 (do_estimate_edge_time): Update.
22961 (inline_generate_summary): Update.
22962 (inline_read_section): Update.
22963 (inline_read_summary): Update.
22964 (inline_free_summary): Update.
22965 * ipa-inline.c (can_inline_edge_p): Update.
22966 (compute_inlined_call_time): Update.
22967 (want_inline_small_function_p): Update.
22968 (edge_badness): Update.
22969 (early_inliner): Update.
22970 * ipa-inline.h (inline_edge_summary): Turn to ...
22971 (ipa_call_summary): ... this one.
22972 (ipa_call_summary_t): New class.
22973 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22974 (ipa_call_summaries): New.
22975 (inline_edge_summary): Remove.
22976 (estimate_edge_growth): Update.
22977 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22978 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22979 * ipa-split.c (execute_split_functions): Update.
22980 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22982 2017-05-19 Richard Biener <rguenther@suse.de>
22984 PR middle-end/80764
22985 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22987 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22989 * config/rs6000/rs6000.c (struct machine_function): Add field
22990 fpr_is_wrapped_separately.
22991 (rs6000_get_separate_components): Use 64 components. Handle the
22992 new FPR components.
22993 (rs6000_components_for_bb): Handle the FPR components.
22994 (rs6000_emit_prologue_components): Handle the FPR components.
22995 (rs6000_emit_epilogue_components): Handle the FPR components.
22996 (rs6000_set_handled_components): Handle the FPR components.
22997 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22998 that are already separately shrink-wrapped.
22999 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23000 that are already separately shrink-wrapped.
23002 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23005 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23008 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23009 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23010 cases where the register allocator uses a move and an offsettable
23011 memory operation to/from a FPR register on ISA 2.06/2.07.
23012 (define_peephole2 for Altivec d-form store): Likewise.
23014 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23017 * config/i386/mmx.md (*mov<mode>_internal): Enable
23018 alternatives 11, 12, 13 and 14 also for 32bit targets.
23019 Remove alternatives 15, 16, 17 and 18.
23020 * config/i386/sse.md (vec_concatv2di): Change
23021 alternative (!x, *y) to (x, ?!*Yn).
23023 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23025 * dumpfile.h (enum dump_kind): Remove stray comma.
23027 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23029 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23030 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23031 predicate::num_conditions
23032 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23033 (CHANGED): turn into predicate::changed.
23034 (agg_position_info): Move to ipa-predicate.h
23035 (add_condition, predicate::add_clause, predicate::operator &=,
23036 predicate::or_with, predicate::evaluate, predicate::probability,
23037 dump_condition, dump_clause, predicate::dump,
23038 predicate::remap_after_duplication, predicate::remap_after_inlining,
23039 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23040 (evaluate_conditions_for_known_args): Update.
23041 (set_cond_stmt_execution_predicate): Update.
23042 * ipa-inline.h: Include ipa-predicate.h
23043 (condition, inline_param_summary, conditions, agg_position_info,
23044 predicate): Move to ipa-predicate.h
23045 * ipa-predicate.c: New file.
23046 * ipa-predicate.h: New file.
23048 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23050 * final.c (leaf_function_p): Check we are not in a sequence.
23052 2017-05-18 Martin Liska <mliska@suse.cz>
23054 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23055 * dumpfile.c (dump_register): Use new enum dump_kind.
23056 (get_dump_file_name): Likewise.
23057 (dump_enable_all): Likewise.
23058 (dump_switch_p_1): Likewise.
23059 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23060 * dumpfile.h (enum dump_kind): New enum type.
23061 (struct dump_file_info): Create constructor and
23062 format fields and comments.
23063 * passes.c (pass_manager::register_one_dump_file):
23065 * statistics.c (statistics_early_init): Likewise.
23066 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23067 TDF_TREE with TDF_SLIM.
23068 (gather_memory_references_ref): Likewise.
23070 2017-05-18 Martin Liska <mliska@suse.cz>
23072 * vec.h (struct vnull): Use it.
23074 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23076 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23077 (true_predicate, false_predicate, true_predicate_p,
23078 false_predicate_p): Remove.
23079 (single_cond_predicate, not_inlined_predicate): Turn to member function
23081 (add_condition): Update.
23082 (add_clause): Turn to...
23083 (predicate::add_clause): ... this one; update; allow passing NULL
23085 (and_predicates): Turn to ...
23086 (predicate::operator &=): ... this one.
23087 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23088 (or_predicates): Turn to ...
23089 (predicate::or_with): ... this one.
23090 (evaluate_predicate): Turn to ...
23091 (predicate::evaluate): ... this one.
23092 (predicate_probability): Turn to ...
23093 (predicate::probability): ... this one.
23094 (dump_condition): Update.
23095 (dump_predicate): Turn to ...
23096 (predicate::dump): ... this one.
23097 (account_size_time): Update.
23098 (edge_set_predicate): Update.
23099 (set_hint_predicate): UPdate.
23100 (evaluate_conditions_for_known_args): Update.
23101 (evaluate_properties_for_edge): Update.
23102 (remap_predicate_after_duplication): Turn to...
23103 (predicate::remap_after_duplication): ... this one.
23104 (remap_hint_predicate_after_duplication): Update.
23105 (inline_summary_t::duplicate): UPdate.
23106 (dump_inline_edge_summary): Update.
23107 (dump_inline_summary): Update.
23108 (set_cond_stmt_execution_predicate): Update.
23109 (set_switch_stmt_execution_predicate): Update.
23110 (compute_bb_predicates): Update.
23111 (will_be_nonconstant_expr_predicate): Update.
23112 (will_be_nonconstant_predicate): Update.
23113 (phi_result_unknown_predicate): Update.
23114 (predicate_for_phi_result): Update.
23115 (array_index_predicate): Update.
23116 (estimate_function_body_sizes): Update.
23117 (estimate_node_size_and_time): Update.
23118 (estimate_ipcp_clone_size_and_time): Update.
23119 (remap_predicate): Rename to ...
23120 (predicate::remap_after_inlining): ... this one.
23121 (remap_hint_predicate): Update.
23122 (inline_merge_summary): Update.
23123 (inline_update_overall_summary): Update.
23124 (estimate_size_after_inlining): Update.
23125 (read_predicate): Rename to ...
23126 (predicate::stream_in): ... this one.
23127 (read_inline_edge_summary): Update.
23128 (write_predicate): Rename to ...
23129 (predicate::stream_out): ... this one.
23130 (write_inline_edge_summary): Update.
23131 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23132 (clause_t): Turn to uint32_t
23133 (predicate): Turn to class; implement constructor and operators
23135 (size_time_entry): Update.
23136 (inline_summary): Update.
23137 (inline_edge_summary): Update.
23139 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23141 * fold-const.c (fold_binary_loc): Move transformation...
23142 * match.pd (C - X CMP X): ... here.
23144 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23146 * config/sparc/sparc.c (sparc_option_override): Set function
23147 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23148 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23150 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23152 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23154 2017-05-18 Marek Polacek <polacek@redhat.com>
23157 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23158 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23159 gimple_assign_load_p.
23161 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23163 PR middle-end/80692
23164 * real.c (do_compare): Give decimal_do_compare preference over
23165 comparing just the signs.
23167 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23169 * doc/md.texi (Canonicalization of Instructions): Describe the
23170 canonical form of instructions that inherently set a condition
23173 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23175 PR middle-end/80775
23176 * tree-cfg.c: Move deletion of unreachable case statements to after
23177 the merging of consecutive case labels.
23179 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23181 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23182 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23183 restoring of callee-saved registers.
23185 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23187 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23188 * config/visium/visium.c (single_set_and_flags): Likewise.
23189 * config/visium/visium.md (Substitutions): Likewise.
23191 2017-05-17 Martin Liska <mliska@suse.cz>
23193 * cfg.c: Introduce dump_flags_t type and
23194 use it instead of int type.
23196 * cfghooks.c: Likewise.
23197 * cfghooks.h (struct cfg_hooks): Likewise.
23198 * cfgrtl.c: Likewise.
23199 * cfgrtl.h: Likewise.
23200 * cgraph.c (cgraph_node::get_body): Likewise.
23201 * coretypes.h: Likewise.
23202 * domwalk.c: Likewise.
23203 * domwalk.h: Likewise.
23204 * dumpfile.c (struct dump_option_value_info): Likewise.
23205 (dump_enable_all): Likewise.
23206 (dump_switch_p_1): Likewise.
23207 (opt_info_switch_p): Likewise.
23208 * dumpfile.h (enum tree_dump_index): Likewise.
23209 (struct dump_file_info): Likewise.
23210 * genemit.c: Likewise.
23211 * generic-match-head.c: Likewise.
23212 * gengtype.c (open_base_files): Likewise.
23213 * gimple-pretty-print.c: Likewise.
23214 * gimple-pretty-print.h: Likewise.
23215 * graph.c (print_graph_cfg): Likewise.
23216 * graphite-scop-detection.c (dot_all_sese): Likewise.
23217 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23218 * loop-unroll.c (report_unroll): Likewise.
23219 * passes.c (pass_manager::register_one_dump_file): Likewise.
23220 * print-tree.c: Likewise.
23221 * statistics.c: Likewise.
23222 * tree-cfg.c: Likewise.
23223 * tree-cfg.h: Likewise.
23224 * tree-dfa.c: Likewise.
23225 * tree-dfa.h: Likewise.
23226 * tree-dump.c (dump_function): Likewise.
23227 * tree-dump.h (struct dump_info): Likewise.
23228 * tree-pretty-print.c: Likewise.
23229 * tree-pretty-print.h: Likewise.
23230 * tree-ssa-live.c: Likewise.
23231 * tree-ssa-live.h: Likewise.
23232 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23233 * tree-vect-loop.c: Likewise.
23234 * tree-vect-slp.c: Likewise.
23236 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23237 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23239 PR tree-optimization/80457
23240 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23241 of all arguments to a statement as scalar_to_vec operations.
23242 (vectorizable_call): Adjust call to vect_model_simple_cost for
23244 (vectorizable_conversion): Likewise.
23245 (vectorizable_assignment): Likewise.
23246 (vectorizable_shift): Likewise.
23247 (vectorizable_operation): Likewise.
23248 (vectorizable_comparison): Likewise.
23249 (vect_is_simple_cond): Record the def types for operands.
23250 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23251 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23252 for statement argument count.
23254 2017-05-16 Carl Love <cel@us.ibm.com>
23256 * config/rs6000/rs6000-c: Add support for built-in functions
23257 vector unsigned long long vec_bperm (vector unsigned long long,
23258 vector unsigned char)
23259 vector signed long long vec_mule (vector signed int,
23261 vector unsigned long long vec_mule (vector unsigned int,
23262 vector unsigned int)
23263 vector signed long long vec_mulo (vector signed int,
23265 vector unsigned long long vec_mulo (vector unsigned int,
23266 vector unsigned int)
23267 vector signed char vec_sldw (vector signed char,
23268 vector signed char,
23270 vector unsigned char vec_sldw (vector unsigned char,
23271 vector unsigned char,
23273 vector signed short vec_sldw (vector signed short,
23274 vector signed short,
23276 vector unsigned short vec_sldw (vector unsigned short,
23277 vector unsigned short,
23279 vector signed int vec_sldw (vector signed int,
23282 vector unsigned int vec_sldw (vector unsigned int,
23283 vector unsigned int,
23285 vector signed long long vec_sldw (vector signed long long,
23286 vector signed long long,
23288 vector unsigned long long vec_sldw (vector unsigned long long,
23289 vector unsigned long long,
23291 * config/rs6000/rs6000-c: Add support for built-in functions
23292 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23293 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23294 * doc/extend.texi: Update the built-in documentation for the
23295 new built-in functions.
23297 2017-05-16 Marek Polacek <polacek@redhat.com>
23301 * tree.c (save_expr): Don't fold the expression.
23303 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23305 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23306 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23307 and (?*y,m). Update insn attributes.
23309 2017-05-16 Martin Liska <mliska@suse.cz>
23311 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23312 flags argument of print_gimple_stmt, print_gimple_expr,
23313 print_generic_stmt and print_generic_expr.
23314 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23315 * coretypes.h: Likewise.
23316 * except.c (dump_eh_tree): Likewise.
23317 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23318 * gimple-pretty-print.h: Likewise.
23319 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23320 (backprop::push_to_worklist): Likewise.
23321 (backprop::pop_from_worklist): Likewise.
23322 (backprop::process_use): Likewise.
23323 (backprop::intersect_uses): Likewise.
23324 (note_replacement): Likewise.
23325 * gimple-ssa-store-merging.c
23326 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23327 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23328 (pass_store_merging::execute): Likewise.
23329 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23330 (ssa_base_cand_dump_callback): Likewise.
23331 (dump_incr_vec): Likewise.
23332 (replace_refs): Likewise.
23333 (replace_mult_candidate): Likewise.
23334 (create_add_on_incoming_edge): Likewise.
23335 (create_phi_basis): Likewise.
23336 (insert_initializers): Likewise.
23337 (all_phi_incrs_profitable): Likewise.
23338 (introduce_cast_before_cand): Likewise.
23339 (replace_one_candidate): Likewise.
23340 * gimplify.c (gimplify_expr): Likewise.
23341 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23342 (set_rename): Likewise.
23343 (rename_uses): Likewise.
23344 (copy_loop_phi_nodes): Likewise.
23345 (add_close_phis_to_merge_points): Likewise.
23346 (copy_loop_close_phi_args): Likewise.
23347 (copy_cond_phi_args): Likewise.
23348 (graphite_copy_stmts_from_block): Likewise.
23349 (translate_pending_phi_nodes): Likewise.
23350 * graphite-poly.c (print_pdr): Likewise.
23351 (dump_gbb_cases): Likewise.
23352 (dump_gbb_conditions): Likewise.
23353 (print_scop_params): Likewise.
23354 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23355 (build_cross_bb_scalars_use): Likewise.
23356 (gather_bbs::before_dom_children): Likewise.
23357 * hsa-dump.c (dump_hsa_immed): Likewise.
23358 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23359 (get_replacement_map): Likewise.
23360 * ipa-inline-analysis.c (dump_condition): Likewise.
23361 (estimate_function_body_sizes): Likewise.
23362 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23363 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23364 * ipa-prop.c (ipa_dump_param): Likewise.
23365 (ipa_print_node_jump_functions_for_edge): Likewise.
23366 (ipa_modify_call_arguments): Likewise.
23367 (ipa_modify_expr): Likewise.
23368 (ipa_dump_param_adjustments): Likewise.
23369 (ipa_dump_agg_replacement_values): Likewise.
23370 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23371 * ipa-pure-const.c (check_stmt): Likewise.
23372 (pass_nothrow::execute): Likewise.
23373 * ipa-split.c (execute_split_functions): Likewise.
23374 * omp-offload.c (dump_oacc_loop_part): Likewise.
23375 (dump_oacc_loop): Likewise.
23376 * trans-mem.c (tm_log_emit): Likewise.
23377 (tm_memopt_accumulate_memops): Likewise.
23378 (dump_tm_memopt_set): Likewise.
23379 (dump_tm_memopt_transform): Likewise.
23380 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23381 (print_loop): Likewise.
23382 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23383 (chkp_gather_checks_info): Likewise.
23384 (chkp_get_check_result): Likewise.
23385 (chkp_remove_check_if_pass): Likewise.
23386 (chkp_use_outer_bounds_if_possible): Likewise.
23387 (chkp_reduce_bounds_lifetime): Likewise.
23388 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23389 (chkp_mark_completed_bounds): Likewise.
23390 (chkp_register_incomplete_bounds): Likewise.
23391 (chkp_mark_invalid_bounds): Likewise.
23392 (chkp_maybe_copy_and_register_bounds): Likewise.
23393 (chkp_build_returned_bound): Likewise.
23394 (chkp_get_bound_for_parm): Likewise.
23395 (chkp_build_bndldx): Likewise.
23396 (chkp_get_bounds_by_definition): Likewise.
23397 (chkp_generate_extern_var_bounds): Likewise.
23398 (chkp_get_bounds_for_decl_addr): Likewise.
23399 * tree-chrec.c (chrec_apply): Likewise.
23400 * tree-data-ref.c (dump_data_reference): Likewise.
23401 (dump_subscript): Likewise.
23402 (dump_data_dependence_relation): Likewise.
23403 (analyze_overlapping_iterations): Likewise.
23404 * tree-inline.c (expand_call_inline): Likewise.
23405 (tree_function_versioning): Likewise.
23406 * tree-into-ssa.c (dump_defs_stack): Likewise.
23407 (dump_currdefs): Likewise.
23408 (dump_names_replaced_by): Likewise.
23409 (dump_update_ssa): Likewise.
23410 (update_ssa): Likewise.
23411 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23412 * tree-parloops.c (build_new_reduction): Likewise.
23413 (try_create_reduction_list): Likewise.
23414 (ref_conflicts_with_region): Likewise.
23415 (oacc_entry_exit_ok_1): Likewise.
23416 (oacc_entry_exit_single_gang): Likewise.
23417 * tree-pretty-print.h: Likewise.
23418 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23419 (get_scalar_evolution): Likewise.
23420 (add_to_evolution): Likewise.
23421 (get_loop_exit_condition): Likewise.
23422 (analyze_evolution_in_loop): Likewise.
23423 (analyze_initial_condition): Likewise.
23424 (analyze_scalar_evolution): Likewise.
23425 (instantiate_scev): Likewise.
23426 (number_of_latch_executions): Likewise.
23427 (gather_chrec_stats): Likewise.
23428 (final_value_replacement_loop): Likewise.
23429 (scev_const_prop): Likewise.
23430 * tree-sra.c (dump_access): Likewise.
23431 (disqualify_candidate): Likewise.
23432 (create_access): Likewise.
23433 (reject): Likewise.
23434 (maybe_add_sra_candidate): Likewise.
23435 (create_access_replacement): Likewise.
23436 (analyze_access_subtree): Likewise.
23437 (analyze_all_variable_accesses): Likewise.
23438 (sra_modify_assign): Likewise.
23439 (initialize_constant_pool_replacements): Likewise.
23440 (find_param_candidates): Likewise.
23441 (decide_one_param_reduction): Likewise.
23442 (replace_removed_params_ssa_names): Likewise.
23443 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23444 * tree-ssa-copy.c (dump_copy_of): Likewise.
23445 (copy_prop_visit_cond_stmt): Likewise.
23446 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23447 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23448 (record_equivalences_from_stmt): Likewise.
23449 * tree-ssa-dse.c (compute_trims): Likewise.
23450 (delete_dead_call): Likewise.
23451 (delete_dead_assignment): Likewise.
23452 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23453 (forward_propagate_into_cond): Likewise.
23454 (pass_forwprop::execute): Likewise.
23455 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23456 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23458 (move_computations_worker): Likewise.
23459 (execute_sm): Likewise.
23460 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23461 (remove_exits_and_undefined_stmts): Likewise.
23462 (remove_redundant_iv_tests): Likewise.
23463 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23464 (adjust_iv_update_pos): Likewise.
23465 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23466 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23467 (value_replacement): Likewise.
23468 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23469 * tree-ssa-pre.c (print_pre_expr): Likewise.
23470 (get_representative_for): Likewise.
23471 (create_expression_by_pieces): Likewise.
23472 (insert_into_preds_of_block): Likewise.
23473 (eliminate_insert): Likewise.
23474 (eliminate_dom_walker::before_dom_children): Likewise.
23475 (eliminate): Likewise.
23476 (remove_dead_inserted_code): Likewise.
23477 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23478 * tree-ssa-reassoc.c (get_rank): Likewise.
23479 (eliminate_duplicate_pair): Likewise.
23480 (eliminate_plus_minus_pair): Likewise.
23481 (eliminate_not_pairs): Likewise.
23482 (undistribute_ops_list): Likewise.
23483 (eliminate_redundant_comparison): Likewise.
23484 (update_range_test): Likewise.
23485 (optimize_range_tests_var_bound): Likewise.
23486 (optimize_vec_cond_expr): Likewise.
23487 (rewrite_expr_tree): Likewise.
23488 (rewrite_expr_tree_parallel): Likewise.
23489 (linearize_expr): Likewise.
23490 (break_up_subtract): Likewise.
23491 (linearize_expr_tree): Likewise.
23492 (attempt_builtin_powi): Likewise.
23493 (attempt_builtin_copysign): Likewise.
23494 (transform_stmt_to_copy): Likewise.
23495 (transform_stmt_to_multiply): Likewise.
23496 (dump_ops_vector): Likewise.
23497 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23498 (print_scc): Likewise.
23499 (set_ssa_val_to): Likewise.
23500 (visit_reference_op_store): Likewise.
23501 (visit_use): Likewise.
23502 (sccvn_dom_walker::before_dom_children): Likewise.
23503 (run_scc_vn): Likewise.
23504 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23506 (expr_hash_elt::print): Likewise.
23507 (const_and_copies::pop_to_marker): Likewise.
23508 (const_and_copies::record_const_or_copy_raw): Likewise.
23509 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23510 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23511 (dump_predicates): Likewise.
23512 (find_uninit_use): Likewise.
23513 (warn_uninitialized_phi): Likewise.
23514 (pass_late_warn_uninitialized::execute): Likewise.
23515 * tree-ssa.c (verify_vssa): Likewise.
23516 (verify_ssa): Likewise.
23517 (maybe_optimize_var): Likewise.
23518 * tree-vrp.c (dump_value_range): Likewise.
23519 (dump_all_value_ranges): Likewise.
23520 (dump_asserts_for): Likewise.
23521 (register_edge_assert_for_2): Likewise.
23522 (vrp_visit_cond_stmt): Likewise.
23523 (vrp_visit_switch_stmt): Likewise.
23524 (vrp_visit_stmt): Likewise.
23525 (vrp_visit_phi_node): Likewise.
23526 (simplify_cond_using_ranges_1): Likewise.
23527 (fold_predicate_in): Likewise.
23528 (evrp_dom_walker::before_dom_children): Likewise.
23529 (evrp_dom_walker::push_value_range): Likewise.
23530 (evrp_dom_walker::pop_value_range): Likewise.
23531 (execute_early_vrp): Likewise.
23533 2017-05-16 Richard Biener <rguenther@suse.de>
23535 * dwarf2out.c (loc_list_from_tree_1): Do not create
23536 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23538 2017-05-16 Richard Biener <rguenther@suse.de>
23540 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23542 (note_variable_value_in_expr): If we resolved the decl ref
23543 do not push to the stack.
23545 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23547 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23548 operations in fast-math mode.
23549 (vaddq_f16): Likewise.
23550 (vmul_f16): Likewise.
23551 (vmulq_f16): Likewise.
23552 (vsub_f16): Likewise.
23553 (vsubq_f16): Likewise.
23554 * config/arm/neon.md (add<mode>3): New.
23556 (fma:<VH:mode>3): New. Also remove outdated comment.
23559 2017-05-16 Martin Liska <mliska@suse.cz>
23563 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23564 (odr_types_equivalent_p): Likewise.
23566 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23568 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23570 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23573 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23574 non-interunit SSE move alternatives with '?'.
23575 (zero-extendsidi peephole2): New peephole to skip intermediate
23576 general register in SSE zero-extend sequence.
23578 2017-05-15 Jeff Law <law@redhat.com>
23580 * reorg.c (relax_delay_slots): Create a new variable to hold
23581 the temporary target rather than clobbering TARGET_LABEL.
23583 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23584 missing argument to extract_bit_field call.
23585 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23587 2017-05-15 Martin Liska <mliska@suse.cz>
23590 * gcc.c (process_command): Do not allow empty argument of -o option.
23592 2017-05-15 Renlin Li <renlin.li@arm.com>
23594 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23595 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23596 * config/aarch64/constraints.md (Usf): Add long call check.
23597 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23598 (call_value): Likewise.
23599 (sibcall): Likewise.
23600 (sibcall_value): Likewise.
23602 (call_value_insn): New.
23603 (sibcall_insn): Update rtx pattern.
23604 (sibcall_value_insn): Likewise.
23605 (call_internal): Remove.
23606 (call_value_internal): Likewise.
23607 (sibcall_internal): Likewise.
23608 (sibcall_value_internal): Likewise.
23609 (call_reg): Likewise.
23610 (call_symbol): Likewise.
23611 (call_value_reg): Likewise.
23612 (call_value_symbol): Likewise.
23614 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23617 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23619 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23621 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23622 compatible with CCGOCmode and with CCZmode.
23624 2017-05-14 Martin Sebor <msebor@redhat.com>
23626 PR middle-end/77671
23627 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23628 (gimple_fold_builtin_snprintf): Same.
23629 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23630 (gimple_fold_builtin_snprintf): Same.
23631 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23632 of character types.
23633 (is_call_safe): New function.
23634 (try_substitute_return_value): Call it.
23635 (try_simplify_call): New function.
23636 (pass_sprintf_length::handle_gimple_call): Call it.
23638 2017-05-14 Martin Sebor <msebor@redhat.com>
23640 PR middle-end/80669
23641 * builtins.c (expand_builtin_stpncpy): Simplify.
23643 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23645 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23646 * config/i386/i386.h
23647 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23648 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23649 (struct machine_function): Add new members call_ms2sysv,
23650 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23651 (struct machine_frame_state): New fields sp_realigned and
23652 sp_realigned_offset.
23653 * config/i386/i386.c
23654 (enum xlogue_stub): New enum.
23655 (enum xlogue_stub_sets): New enum.
23656 (class xlogue_layout): New class.
23657 (struct ix86_frame): New fields stack_realign_allocate_offset,
23658 stack_realign_offset and outlined_save_offset. Modify comments to
23659 detail stack layout when using out-of-line stubs.
23660 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23661 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23662 -mcall-ms2sysv-xlogues.
23663 (stub_managed_regs): New static variable.
23664 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23665 registers managed by out-of-line stub.
23666 (disable_call_ms2sysv_xlogues): New function.
23667 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23668 m->call_ms2sysv when appropriate and compute frame layout for
23670 (sp_valid_at, fp_valid_at): New inline functions.
23671 (choose_basereg): New function.
23672 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23674 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23675 Use align parameter of choose_baseaddr to generated aligned SSE movs
23677 (pro_epilogue_adjust_stack): Modify to track
23678 machine_frame_state::sp_realigned.
23679 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23680 (ix86_nsaved_sseregs): Likewise.
23681 (ix86_emit_save_regs): Likewise.
23682 (ix86_emit_save_regs_using_mov): Likewise.
23683 (ix86_emit_save_sse_regs_using_mov): Likewise.
23684 (get_scratch_register_on_entry): Likewise.
23685 (gen_frame_set): New function.
23686 (gen_frame_load): Likewise.
23687 (gen_frame_store): Likewise.
23688 (emit_outlined_ms2sysv_save): Likewise.
23689 (emit_outlined_ms2sysv_restore): Likewise.
23690 (ix86_expand_prologue): Modify stack re-alignment code and call
23691 emit_outlined_ms2sysv_save when appropriate.
23692 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23693 parameter rtx_insn *insn, which allows the function to be used to only
23694 generate the notes.
23695 (ix86_expand_epilogue): Modify validity checks of frame and stack
23696 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23697 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23698 * config/i386/predicates.md
23699 (save_multiple): New predicate.
23700 (restore_multiple): Likewise.
23701 * config/i386/sse.md
23702 (save_multiple<mode>): New pattern.
23703 (save_multiple_realign<mode>): Likewise.
23704 (restore_multiple<mode>): Likewise.
23705 (restore_multiple_and_return<mode>): Likewise.
23706 (restore_multiple_leave_return<mode>): Likewise.
23707 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23709 2017-05-14 Julia Koval <julia.koval@intel.com>
23711 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23712 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23713 (__builtin_ia32_xsetbv): New builtins.
23714 * config/i386/i386.c (ix86_expand_special_args_builtin):
23716 (ix86_expand_builtin): Special expand for new intrinsics.
23717 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23718 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23719 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23721 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23723 * cfganal.c (inverted_post_order_compute): Change argument type
23725 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23726 * df-core.c (rest_of_handle_df_initialize): Adjust.
23727 (rest_of_handle_df_finish): Likewise.
23728 (df_analyze_1): Likewise.
23729 (df_analyze): Likewise.
23730 (loop_inverted_post_order_compute): Change argument to be a vec *.
23731 (df_analyze_loop): Adjust.
23732 (df_get_n_blocks): Likewise.
23733 (df_get_postorder): Likewise.
23734 * df.h (struct df_d): Change field to be a vec.
23735 * lcm.c (compute_laterin): Adjust.
23736 (compute_available): Likewise.
23737 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23738 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23739 * tree-ssa-pre.c (compute_antic): Likewise.
23741 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23743 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23744 (depth_first_search::depth_first_search): Change structure init
23745 function to this constructor.
23746 (depth_first_search::add_bb): Rename function to this member.
23747 (depth_first_search::execute): Likewise.
23748 (flow_dfs_compute_reverse_finish): Adjust.
23750 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23752 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23753 (longest_simple_path): Likewise.
23754 * shrink-wrap.c (spread_components): Likewise.
23755 (disqualify_problematic_components): Likewise.
23756 (emit_common_heads_for_components): Likewise.
23757 (emit_common_tails_for_components): Likewise.
23758 (insert_prologue_epilogue_for_components): Likewise.
23760 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23762 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23765 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23767 * df-core.c (df_set_blocks): Start using auto_bitmap.
23768 (df_compact_blocks): Likewise.
23769 * df-problems.c (df_rd_confluence_n): Likewise.
23770 * df-scan.c (df_insn_rescan_all): Likewise.
23771 (df_process_deferred_rescans): Likewise.
23772 (df_update_entry_block_defs): Likewise.
23773 (df_update_exit_block_uses): Likewise.
23774 (df_entry_block_bitmap_verify): Likewise.
23775 (df_exit_block_bitmap_verify): Likewise.
23776 (df_scan_verify): Likewise.
23777 * lra-constraints.c (lra_constraints): Likewise.
23778 (undo_optional_reloads): Likewise.
23779 (lra_undo_inheritance): Likewise.
23780 * lra-remat.c (calculate_gen_cands): Likewise.
23781 (do_remat): Likewise.
23782 * lra-spills.c (assign_spill_hard_regs): Likewise.
23783 (spill_pseudos): Likewise.
23784 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23785 (bitmap_set_subtract_values): Likewise.
23787 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23789 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23790 management with auto_bitmap.
23791 (fix_inter_tick): Likewise.
23792 (fix_recovery_deps): Likewise.
23793 * ira.c (add_store_equivs): Likewise.
23794 (find_moveable_pseudos): Likewise.
23795 (split_live_ranges_for_shrink_wrap): Likewise.
23796 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23797 (rtx_reuse_manager::seen_def_p): Likewise.
23798 (rtx_reuse_manager::set_seen_def): Likewise.
23799 * print-rtl.h (class rtx_reuse_manager): Likewise.
23801 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23803 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23805 (migrate_btr_def): Likewise.
23806 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23807 * df-core.c (loop_post_order_compute): Likewise.
23808 (loop_inverted_post_order_compute): Likewise.
23809 * hsa-common.h: Likewise.
23810 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23811 * init-regs.c (initialize_uninitialized_regs): Likewise.
23812 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23813 (inline_small_functions): Likewise.
23814 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23815 * ira.c (combine_and_move_insns): Likewise.
23816 (build_insn_chain): Likewise.
23817 * loop-invariant.c (find_invariants): Likewise.
23818 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23819 * predict.c (tree_predict_by_opcode): Likewise.
23820 (predict_paths_leading_to): Likewise.
23821 (predict_paths_leading_to_edge): Likewise.
23822 (estimate_loops_at_level): Likewise.
23823 (estimate_loops): Likewise.
23824 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23825 (spread_components): Likewise.
23826 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23827 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23828 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23829 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23830 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23831 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23832 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23833 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23834 (mark_threaded_blocks): Likewise.
23835 (thread_through_all_blocks): Likewise.
23836 * tree-ssa.c (verify_ssa): Likewise.
23837 (execute_update_addresses_taken): Likewise.
23838 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23840 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23842 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23844 (post_order_compute): Likewise.
23845 (inverted_post_order_compute): Likewise.
23846 (pre_and_rev_post_order_compute_fn): Likewise.
23848 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23850 * genrecog.c (int_set::int_set): Explicitly construct our
23851 auto_vec base class.
23852 * vec.h (auto_vec::auto_vec): New constructor.
23854 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23856 * bitmap.h (class auto_bitmap): New constructor taking
23857 bitmap_obstack * argument.
23859 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23861 * bitmap.h (class auto_bitmap): Change type of m_bits to
23862 bitmap_head, and adjust ctor / dtor and member operators.
23864 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23866 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23867 when returned register mode doesn't match original mode.
23869 2017-05-12 Jeff Law <law@redhat.com>
23870 Jakub Jelinek <jakub@redhat.com>
23872 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23873 we look for cc setter after the compare-elim changes.
23874 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23875 within the vector to match what compare-elim now expects.
23876 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23877 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23879 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23880 after the compare-elim changes.
23881 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23882 the vector to match what compare-elim now expects.
23883 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23884 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23885 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23886 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23887 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23889 * config/visium/visium.c (single_set_and_flags): Fix where
23890 we look for cc setter after the compare-elim changes.
23891 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23892 with the vector to match what compare-elim now expects.
23893 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23894 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23895 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23896 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23897 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23898 (neg<mode>2_insn_set_overflow): Likewise.
23900 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23902 PR middle-end/79794
23903 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23904 maybe_expand_insn call, set ops[0].target. If still set after call,
23905 set alt_rtl. Add extra arg to recursive calls.
23906 (extract_bit_field): Add alt_rtl argument. Pass to
23908 * expmed.h (extract_bit_field): Fix prototype.
23909 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23910 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23911 to extract_bit_field_calls.
23912 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23913 Pass alt_rtl to extract_bit_field calls.
23914 * calls.c (store_unaligned_arguments_into_psuedos)
23915 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23916 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23918 * optabs.h (struct expand_operand): Add target bitfield.
23920 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23922 * compare-elim.c (try_eliminate_compare): Canonicalize
23923 operation with embedded compare to
23924 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23925 (set (reg) (operation)].
23927 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23929 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23932 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23933 cost of adding a carry flag for ADC instruction.
23934 [case MINUS]: Ignore the cost of subtracting a carry flag
23935 for SBB instruction.
23937 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23939 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23941 * config/rs6000/bmiintrin.h: New file.
23942 * config/rs6000/bmi2intrin.h: New file.
23943 * config/rs6000/x86intrin.h: New file.
23945 2017-05-12 Jeff Law <law@redhat.com>
23947 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23950 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23952 PR middle-end/80707
23953 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23955 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23957 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23958 early expansion of vector divide builtins.
23959 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23960 builtins identified as having unsigned arguments.
23962 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23964 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23965 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23966 expansion of vector logical operations (and, andc, or, xor,
23969 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23971 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23972 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23974 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23976 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23977 early GIMPLE expansion of vector multiplies.
23979 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23981 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23982 TARGET_HAVE_MOVT conditional.
23983 (movt splitter): Likewise.
23985 2017-05-12 Richard Biener <rguenther@suse.de>
23987 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23988 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23989 Fold all stmts not inplace.
23991 2017-05-12 Richard Biener <rguenther@suse.de>
23993 PR tree-optimization/80713
23994 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23995 inserted_exprs bit for not removed stmts.
23997 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23999 PR middle-end/69921
24000 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24001 parallelized" attribute for parallelized OpenACC kernels.
24002 * omp-offload.c (execute_oacc_device_lower): Use it.
24004 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24005 Set "oacc kernels" attribute.
24006 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24007 parameter. Adjust all users.
24008 (oacc_fn_attrib_kernels_p): Remove function.
24009 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24010 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24011 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24012 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24013 assert "oacc kernels" attribute is set.
24015 2017-05-11 Carl Love <cel@us.ibm.com>
24017 * config/rs6000/rs6000-c: Add support for built-in functions
24018 vector unsigned char vec_popcnt (vector signed char)
24019 vector unsigned char vec_popcnt (vector unsigned char)
24020 vector unsigned short vec_popcnt (vector signed short)
24021 vector unsigned short vec_popcnt (vector unsigned short)
24022 vector unsigned int vec_popcnt (vector signed int)
24023 vector unsigned int vec_popcnt (vector unsigned int)
24024 vector unsigned long long vec_popcnt (vector signed long long)
24025 vector unsigned long long vec_popcnt (vector unsigned long long)
24026 vector signed long long vec_slo (vector signed long long,
24027 vector signed char)
24028 vector signed long long vec_slo (vector signed long long,
24029 vector unsigned char)
24030 vector unsigned long long vec_slo (vector unsigned long long,
24031 vector signed char)
24032 vector unsigned long long vec_slo (vector unsigned long long,
24033 vector unsigned char)
24034 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24035 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24036 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24037 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24038 * doc/extend.texi: Update the built-in documentation file for the
24039 new built-in functions.
24041 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24043 * attribs.h (sorted_attr_string): Move machine independent
24044 functions for target clone support from the i386 port to common
24045 code. Rename ix86_function_versions to common_function_versions.
24046 Rename make_name to make_unique_name.
24047 (common_function_versions): Likewise.
24048 (make_unique_name): Likewise.
24049 (make_dispatcher_decl): Likewise.
24050 (is_function_default_version): Likewise.
24051 * attribs.c (attr_strcmp): Likewise.
24052 (sorted_attr_string): Likewise.
24053 (common_function_versions): Likewise.
24054 (make_unique_name): Likewise.
24055 (make_dispatcher_decl): Likewise.
24056 (is_function_default_version): Likewise.
24057 * config/i386/i386.c (attr_strcmp): Likewise.
24058 (sorted_attr_string): Likewise.
24059 (ix86_function_versions): Likewise.
24060 (make_name): Likewise.
24061 (make_dispatcher_decl): Likewise.
24062 (is_function_default_version): Likewise.
24063 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24065 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24068 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24069 Account for direct move costs for vec_construct of integer
24072 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24075 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24076 (UNSPEC_STX_ATOMIC): Ditto.
24077 (loaddi_via_sse): New insn.
24078 (storedi_via_sse): Ditto.
24079 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24080 Update corresponding peephole2 patterns.
24081 (atomic_storedi_fpu): Ditto.
24083 2017-05-11 Julia Koval <julia.koval@intel.com>
24085 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24086 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24088 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24089 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24090 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24092 2017-05-11 Nathan Sidwell <nathan@acm.org>
24094 * graphite-poly.c: Include dumpfile.h.
24096 * dumpfle.h (dump_function): Declare here ...
24097 * tree-dump.h (dump_function): ... not here.
24098 * dumpfile.c: #include tree-cfg.h.
24099 (dump_function): Move here from ...
24100 * tree-dump.c (dump_function): ... here.
24101 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24102 * graphite-poly.c: Don't include tree-dump.h.
24103 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24104 * print-tree.c: Likewise.
24105 * stor-layout.c: Likewise.
24106 * tree-nested.c: Likewise.
24108 * dumpfile.c (dump_start): Use TDF_FLAGS.
24109 (dump_enable_all): Fix TDF_KIND check thinko.
24111 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24113 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24114 array entries to represent two legal parameterizations of the
24115 overloaded __builtin_cmpb function, as represented by the
24116 P6_OV_BUILTIN_CMPB constant.
24117 (altivec_resolve_overloaded_builtin): Add special case handling
24118 for the __builtin_cmpb function, as represented by the
24119 P6_OV_BUILTIN_CMPB constant.
24120 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24121 (BU_P6_64BIT_2): New macro.
24122 (BU_P6_OVERLOAD_2): New macro
24123 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24124 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24125 (CMPB): Add overload support to represent both 32-bit and 64-bit
24126 compare-bytes function.
24127 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24128 support for TARGET_CMPB.
24129 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24130 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24131 documentation of the __builtin_cmpb overloaded built-in function.
24133 2017-05-11 Richard Biener <rguenther@suse.de>
24135 PR tree-optimization/80705
24136 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24137 bases are not vectorizable.
24139 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24141 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24142 when counting register pressure.
24144 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24146 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24147 (iv_ca_more_deps): Renamed to ...
24148 (iv_ca_compare_deps): ... this.
24149 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24151 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24153 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24155 (determine_group_iv_costs): ... here.
24156 (find_inv_vars_cb): Record inv var if it's not recorded before.
24158 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24160 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24161 (get_shiftadd_cost): Ditto.
24163 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24165 * tree-ssa-address.c: Include header file.
24166 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24168 (add_to_parts): Refactor.
24169 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24170 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24173 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24175 PR tree-optimization/53090
24176 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24178 (extract_cond_operands): Detect condition with IV on both sides
24179 and return COMP_IV_EXPR_2.
24180 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24181 (rewrite_use_compare): Simplify by removing call to function
24182 extract_cond_operands.
24184 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24186 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24187 (extract_cond_operands): Detect condition comparing against non-
24188 invariant bound and return appropriate enum value.
24189 (find_interesting_uses_cond): Update use of extract_cond_operands.
24190 Handle its return value accordingly.
24191 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24193 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24195 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24196 nonlinear iv_use computation in loop invariant sensitive way.
24198 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24200 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24201 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24203 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24205 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24206 (dump_cand): Support iv_cand.inv_exprs.
24207 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24209 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24212 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24214 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24216 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24217 as local function. Include necessary header files.
24218 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24220 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24222 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24224 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24226 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24227 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24228 RSHIFT_EXPR and BIT_NOT_EXPR.
24230 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24232 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24233 (adjust_setup_cost): New parameter supporting round up adjustment.
24234 (struct address_cost_data): Delete.
24235 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24236 (split_address_cost, ptr_difference_cost): Delete.
24237 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24238 (struct ainc_cost_data): New struct.
24239 (get_address_cost_ainc): New function.
24240 (get_address_cost, get_computation_cost): Reimplement.
24241 (determine_group_iv_cost_address): Record inv_expr for all uses of
24243 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24244 (iv_ca_has_deps): Reimplemented to ...
24245 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24247 (iv_ca_extend): Call iv_ca_more_deps.
24249 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24251 * tree-ssa-address.c (struct mem_address): Move to header file.
24252 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24253 * tree-ssa-address.h (struct mem_address): Move from C file.
24254 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24256 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24258 * tree-affine.h (aff_combination_type): New interface.
24259 (aff_combination_zero_p): Remove static.
24260 (aff_combination_const_p): New interface.
24261 (aff_combination_singleton_var_p): New interfaces.
24263 2017-05-11 Richard Biener <rguenther@suse.de>
24265 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24266 Skip unreachable blocks and destinations.
24267 (eliminate): Move stmt removal and fixup ...
24268 (fini_eliminate): ... here. Skip inserted exprs.
24269 (pass_pre::execute): Move fini_pre after fini_eliminate.
24270 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24271 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24272 PRE to get rid of dead code that has invalid SSA form and
24273 split critical edges again.
24275 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24277 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24279 2017-05-11 Richard Biener <rguenther@suse.de>
24281 * passes.c (execute_function_todo): Verify loops if they are
24282 said to be up-to-date.
24283 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24284 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24286 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24289 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24290 handle calling assemble_external ourself.
24293 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24294 modes with zero size. Enhance comment.
24296 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24298 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24299 built-ins for vec_xl and vec_xst with short and char pointer
24302 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24304 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24305 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24306 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24307 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24308 (_mm_maskz_min_round_ss): New intrinsics.
24309 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24310 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24311 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24312 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24313 (__builtin_ia32_minss_mask_round): New builtins.
24314 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24315 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24316 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24318 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24319 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24321 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24324 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24326 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24327 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24328 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24329 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24330 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24331 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24332 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24333 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24334 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24335 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24336 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24337 (__builtin_ia32_mulss_mask_round): New builtins.
24338 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24339 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24340 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24342 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24343 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24345 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24348 2017-05-10 Julia Koval <julia.koval@intel.com>
24350 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24351 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24352 (_mm256_setr_m128i): New intrinsics.
24354 2017-05-10 Julia Koval <julia.koval@intel.com>
24356 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24357 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24358 (_mm_maskz_rcp14_ss): New intrinsics.
24359 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24360 (__builtin_ia32_rcp14ss_mask): New builtins.
24361 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24363 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24365 PR tree-optimization/51513
24366 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24367 (assert_unreachable_fallthru_edge_p): Use it.
24368 (group_case_labels_stmt): Likewise.
24369 * tree-cfg.h: Prototype it.
24370 * stmt.c: Include cfghooks.h and tree-cfg.h.
24371 (emit_case_dispatch_table) <gap_label>: New local variable.
24372 Use it to fill dispatch table gaps.
24373 Test for default_label before updating probabilities.
24374 (expand_case) <default_label>: Remove unneeded initialization.
24375 Test for unreachable default case statement and remove its edge.
24376 Set default_label accordingly.
24377 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24379 2017-05-10 Carl Love <cel@us.ibm.com>
24381 * config/rs6000/rs6000-c: Add support for built-in functions
24382 vector signed char vec_neg (vector signed char)
24383 vector signed short int vec_neg (vector short int)
24384 vector signed int vec_neg (vector signed int)
24385 vector signed long long vec_neg (vector signed long long)
24386 vector float vec_neg (vector float)
24387 vector double vec_neg (vector double)
24388 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24390 * config/rs6000/altivec.h: Add define for vec_neg
24391 * doc/extend.texi: Update the built-in documentation for the
24392 new built-in functions.
24394 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24396 PR tree-optimization/77644
24397 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24399 2017-05-10 Nathan Sidwell <nathan@acm.org>
24401 * dumpfile.h (TDI_lang_all): New.
24402 (TDF_KIND): New. Renumber others
24403 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24405 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24407 (get_dump_file_name): Adjust suffix generation.
24408 (dump_enable_all): Use TDF_KIND.
24409 * doc/invoke.texi (-fdump-lang-all): Document.
24411 * dumpfile.h: Tabify.
24413 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24416 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24417 Move member access before delete.
24419 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24421 * tree-inline.c (expand_call_inline): Split block at stmt
24424 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24427 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24428 are now unused after splitting mov{sf,sd}_hardfloat.
24429 (f32_lr2): Likewise.
24430 (f32_lm): Likewise.
24431 (f32_lm2): Likewise.
24432 (f32_li): Likewise.
24433 (f32_li2): Likewise.
24434 (f32_lv): Likewise.
24435 (f32_sr): Likewise.
24436 (f32_sr2): Likewise.
24437 (f32_sm): Likewise.
24438 (f32_sm2): Likewise.
24439 (f32_si): Likewise.
24440 (f32_si2): Likewise.
24441 (f32_sv): Likewise.
24442 (f32_dm): Likewise.
24443 (f32_vsx): Likewise.
24444 (f32_av): Likewise.
24445 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24446 For movsf, order stores so the VSX stores occur before the GPR
24447 store which encourages the register allocator to use a traditional
24448 FPR instead of a GPR. For movsd, order the stores so that the GPR
24449 store comes before the VSX stores to allow the power6 to work.
24450 This is due to the power6 not having a 32-bit integer store
24451 instruction from a FPR.
24452 (movsf_hardfloat): Likewise.
24453 (movsd_hardfloat): Likewise.
24455 2017-05-09 Martin Sebor <msebor@redhat.com>
24457 PR translation/80280
24458 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24461 PR translation/80280
24462 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24463 data member added in r247778.
24464 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24466 2017-05-09 Nathan Sidwell <nathan@acm.org>
24468 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24470 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24473 2017-05-09 Marek Polacek <polacek@redhat.com>
24475 * doc/invoke.texi: Fix typo.
24477 2017-05-09 Richard Biener <rguenther@suse.de>
24479 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24480 (vrp_val_is_min): Likewise.
24481 (set_value_range_to_value): Likewise.
24482 (set_value_range_to_nonnegative): Likewise.
24483 (gimple_assign_nonzero_p): Likewise.
24484 (gimple_stmt_nonzero_p): Likewise.
24485 (vrp_int_const_binop): Likewise. Remove unreachable case.
24486 (adjust_range_with_scev): Adjust comments.
24487 (compare_range_with_value): Likewise.
24488 (extract_range_from_phi_node): Likewise.
24489 (test_for_singularity): Likewise.
24491 2017-05-09 Richard Biener <rguenther@suse.de>
24493 * tree-vrp.c (get_single_symbol): Add assert that we don't
24494 get overflowed constants as invariant part.
24495 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24496 checks. Use wi::cmp instead of recursing for integer constants.
24497 (compare_values): Just ignore whether we assumed undefined
24498 overflow instead of failing the compare.
24499 (extract_range_for_var_from_comparison_expr): Add comment before the
24500 TREE_NO_WARNING sets.
24501 (test_for_singularity): Likewise.
24502 (extract_range_from_comparison): Do not disable optimization
24503 when we assumed undefined overflow.
24504 (extract_range_basic): Remove init of unused var.
24506 2017-05-09 Richard Biener <rguenther@suse.de>
24508 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24509 (extract_range_from_multiplicative_op_1): Adjust.
24510 (extract_range_from_binary_expr_1): Use int_const_binop.
24512 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24515 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24516 rs6000_store_data_bypass_p in seven define_bypass directives and
24517 in several comments.
24518 * config/rs6000/rs6000-protos.h: Add prototype for
24519 rs6000_store_data_bypass_p function.
24520 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24521 function implements slightly different (rs6000-specific) semantics
24522 than store_data_bypass_p, returning false rather than aborting
24523 with assertion error when arguments do not satisfy the
24524 requirements of store data bypass.
24525 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24526 rs6000_store_data_bypass_p.
24528 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24530 * config/xtensa/xtensa-protos.h
24531 (xtensa_initial_elimination_offset): New declaration.
24532 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24533 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24534 macro definition, add case for FRAME_POINTER_REGNUM when
24535 FRAME_GROWS_DOWNWARD.
24536 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24537 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24538 xtensa_initial_elimination_offset.
24540 2017-05-08 Nathan Sidwell <nathan@acm.org>
24542 * doc/invoke.texi: Alphabetize -fdump options.
24544 2017-05-08 Martin Sebor <msebor@redhat.com>
24546 PR translation/80280
24547 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24549 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24551 * target.def (compute_frame_layout): New optional target hook.
24552 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24553 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24554 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24556 * reload1.c (verify_initial_elim_offsets): Likewise.
24557 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24558 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24559 (arm_get_frame_offsets): Split up into this ...
24560 (arm_compute_frame_layout): ... and this function.
24562 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24564 * config/aarch64/constraints.md (Usa): New constraint.
24565 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24567 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24569 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24570 with_multilib_list after it has been checked.
24572 2017-05-08 Richard Biener <rguenther@suse.de>
24574 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24575 (bitmap_set_subtract_values): Likewise.
24577 2017-05-08 Richard Biener <rguenther@suse.de>
24579 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24580 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24582 (gimple_stmt_nonzero_warnv_p): Rename to ...
24583 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24585 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24586 (extract_range_basic): Adjust, do not disable propagation on
24587 strict overflow sensitive simplification.
24588 (vrp_visit_cond_stmt): Likewise.
24590 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24592 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24593 body size unconditionally.
24595 2017-05-07 Jeff Law <law@redhat.com>
24598 2017-05-06 Jeff Law <law@redhat.com>
24599 PR tree-optimization/78496
24600 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24603 PR tree-optimization/78496
24604 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24605 (simplify_stmt_using_ranges): Call it.
24606 (vrp_dom_walker::before_dom_children): Extract equivalences
24607 from an ASSERT_EXPR with an equality comparison against a
24610 2017-05-06 Jeff Law <law@redhat.com>
24612 PR tree-optimization/78496
24613 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24616 PR tree-optimization/78496
24617 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24618 (simplify_stmt_using_ranges): Call it.
24619 (vrp_dom_walker::before_dom_children): Extract equivalences
24620 from an ASSERT_EXPR with an equality comparison against a
24623 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24625 * lra-constraints.c (lra_copy_reg_equiv): New function.
24626 (split_reg): Use it to copy equivalence information from the
24627 original register to the spill register.
24629 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24631 PR rtl-optimization/75964
24632 * simplify-rtx.c (simplify_const_relational_operation): Remove
24633 invalid handling of comparisons of integer ABS.
24635 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24637 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24638 initialize to zero.
24639 (init_regs): Remove declaration.
24640 (function_arg_advance_32): Initialize error_p as boolean variable.
24642 2017-05-05 Nathan Sidwell <nathan@acm.org>
24644 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24645 lines. Use for (;;).
24647 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24649 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24650 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24651 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24652 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24653 VF=2 that require versioning.
24655 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24657 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24660 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24662 * diagnostic.h (diagnostic_override_option_index): Convert from
24663 macro to inline function.
24665 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24667 * diagnostic.c (last_module_changed_p): New function.
24668 (set_last_module): New function.
24669 (diagnostic_report_current_module): Convert macro usage to
24670 the above functions.
24671 * diagnostic.h (diagnostic_context::last_module): Strengthen
24672 from const line_map * to const line_map_ordinary *.
24673 (diagnostic_last_module_changed): Delete macro.
24674 (diagnostic_set_last_module): Delete macro.
24676 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24678 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24679 with diagnostic_report_diagnostic.
24680 (diagnostic_n_impl_richloc): Likewise.
24681 * diagnostic.h (report_diagnostic): Delete macro.
24682 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24683 with diagnostic_report_diagnostic.
24684 * substring-locations.c (format_warning_va): Likewise.
24686 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24688 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24689 save/restor of format_spec. Move option-printing code to...
24690 (print_option_information): ...this new function, and
24691 reimplement by simply printing to the pretty_printer,
24692 rather than appending to the format string.
24694 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24696 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24697 handling logic into...
24698 (update_effective_level_from_pragmas): ...this new function.
24700 2017-05-04 Andrew Waterman <andrew@sifive.com>
24702 * config/riscv/riscv.opt (mstrict-align): New option.
24703 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24704 (SLOW_UNALIGNED_ACCESS): Define.
24705 (riscv_slow_unaligned_access): Declare.
24706 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24708 (riscv_slow_unaligned_access): New variable.
24709 (rocket_tune_info): Set slow_unaligned_access to true.
24710 (optimize_size_tune_info): Set slow_unaligned_access to false.
24711 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24712 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24713 (riscv_option_override): Set riscv_slow_unaligned_access.
24714 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24716 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24718 * config/riscv/riscv.md: Unify indentation.
24720 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24725 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24727 (extendsi<mode>2): Add support for doing sign extension via
24728 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24729 don't have ISA 3.0 instructions.
24730 (extendsi<mode>2 splitter): Likewise.
24731 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24732 generate the normal insns since SImode can now go in vector
24733 registers. Disallow the special UNSPECs needed for previous
24734 machines to hide SImode being used. Add new insns
24735 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24736 (fix_trunc<mode>si2_stfiwx): Likewise.
24737 (fix_trunc<mode>si2_internal): Likewise.
24738 (fixuns_trunc<mode>si2): Likewise.
24739 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24740 (fctiw<u>z_<mode>_smallint): Likewise.
24741 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24742 of floating point to 32-bit integer from doing a direct move to
24743 the GPR registers to do a store.
24744 (fctiwz_<mode>): Break long line.
24746 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24748 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24749 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24750 (addr_list, addr_offset_valid_p): New.
24751 (split_address_groups): Check offset validity with above function.
24752 (gt-tree-ssa-loop-ivopts.h): Include header file.
24754 2017-05-05 Nathan Sidwell <nathan@acm.org>
24756 * config.gcc (arm*-*-*): Add missing 'fi'.
24758 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24760 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24761 included in -fopt-info does not matter.
24762 * doc/optinfo.texi (-fopt-info): Fix description of default
24763 behavour. Explicitly say order of options included in -fopt-info
24766 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24768 * config.gcc: Allow combinations of aprofile and rmprofile values for
24769 --with-multilib-list.
24770 * config/arm/t-multilib: New file.
24771 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24772 variables. Remove setting of ISA and floating-point ABI in
24773 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24774 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24775 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24776 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24778 * config/arm/t-rmprofile: Likewise except for the matches changes.
24779 * doc/install.texi (--with-multilib-list): Document the combination of
24780 aprofile and rmprofile values and warn about pitfalls in doing that.
24782 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24784 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24785 (movdi_aarch64): Likewise.
24787 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24789 PR tree-optimization/80632
24790 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24792 (build_arrays): Initialize it for virtual phis.
24793 (fix_phi_nodes): Use it for virtual phis.
24795 PR tree-optimization/80558
24796 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24797 [x, y] op z into [x op, y op z] for op & or | if conditions
24800 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24801 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24804 * config/arm/arm.md (use_literal_pool): Remove.
24805 (64-bit immediate split): No longer takes cost into consideration
24806 if arm_disable_literal_pool is enabled.
24807 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24808 used when arm_disable_literal_pool is enabled.
24809 (arm_max_const_double_inline_cost): Remove use of
24810 arm_disable_literal_pool.
24811 (push_minipool_fix): Add assert.
24812 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24813 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24814 (no_literal_pool_sf_immediate): New.
24816 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24818 PR tree-optimization/80613
24819 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24820 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24822 2017-05-05 Richard Biener <rguenther@suse.de>
24824 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24826 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24828 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24829 of this flag from insn conditions due to removal from r247495.
24831 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24833 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24835 (arm_early_store_addr_dep_ptr): Likewise.
24836 * config/arm/aarch-common-protos.h
24837 (arm_early_load_addr_dep_ptr): Add prototype.
24838 (arm_early_store_addr_dep_ptr): Likewise.
24839 * config/arm/cortex-a53.md: Add new bypasses.
24841 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24843 * tree.c (next_type_uid): Change type to unsigned.
24844 (type_hash_canon): Decrement back next_type_uid if
24845 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24846 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24849 2017-05-04 Martin Sebor <msebor@redhat.com>
24851 * builtins.c: Fix a trivial typo in a comment.
24853 PR middle-end/79234
24854 * builtins.c (check_sizes): Adjust to handle reading past the end.
24855 Avoid printing excessive upper bound of ranges. Use %E to print
24856 tree nodes instead of converting them to %wu.
24857 (expand_builtin_memchr): New function.
24858 (compute_dest_size): Rename...
24859 (compute_objsize): ...to this.
24860 (expand_builtin_memcpy): Adjust.
24861 (expand_builtin_mempcpy): Adjust.
24862 (expand_builtin_strcat): Adjust.
24863 (expand_builtin_strcpy): Adjust.
24864 (check_strncat_sizes): Adjust.
24865 (expand_builtin_strncat): Adjust.
24866 (expand_builtin_strncpy): Adjust and simplify.
24867 (expand_builtin_memset): Adjust.
24868 (expand_builtin_bzero): Adjust.
24869 (expand_builtin_memcmp): Adjust.
24870 (expand_builtin): Handle memcmp.
24871 (maybe_emit_chk_warning): Check strncat just once.
24873 2017-05-04 Martin Sebor <msebor@redhat.com>
24875 PR preprocessor/79214
24876 PR middle-end/79222
24877 PR middle-end/79223
24878 * builtins.c (check_sizes): Add inlining context and issue
24879 warnings even when -Wno-system-headers is set.
24880 (check_strncat_sizes): Same.
24881 (expand_builtin_strncat): Same.
24882 (expand_builtin_memmove): New function.
24883 (expand_builtin_stpncpy): Same.
24884 (expand_builtin): Handle memmove and stpncpy.
24886 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24888 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24889 which is not used any more.
24891 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24893 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24895 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24897 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24898 (cortexa53_tunings): Likewise.
24899 (cortexa57_tunings): Likewise.
24900 (cortexa72_tunings): Likewise.
24901 (cortexa73_tunings): Likewise.
24903 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24905 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24906 Set loop alignment to 8.
24908 2017-05-04 Martin Sebor <msebor@redhat.com>
24910 PR translation/80280
24911 * builtins.c (expand_builtin_object_size): Add missing quoting to
24912 %D and like directives.
24913 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24914 (hsa_type_for_tree_type): Same.
24915 (verify_function_arguments): Same.
24916 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24917 * varasm.c (get_section): Same.
24920 2017-05-04 Martin Sebor <msebor@redhat.com>
24922 PR translation/80280
24923 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24925 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24927 * config/aarch64/aarch64.c (generic_addrcost_table):
24928 Change HI/TI mode setting.
24930 2017-05-04 Martin Jambor <mjambor@suse.cz>
24932 PR tree-optimization/80622
24933 * tree-sra.c (comes_initialized_p): New function.
24934 (build_accesses_from_assign): Only set write lazily when
24935 comes_initialized_p is false.
24936 (analyze_access_subtree): Use comes_initialized_p.
24937 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24938 instead of testing for PARM_DECL.
24940 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24942 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24943 constraint on operand 0 to allow more general addressing modes.
24944 Adjust output template.
24945 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24947 * config/aarch64/aarch64-protos.h
24948 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24949 * config/aarch64/constraints.md (Dp): New address constraint.
24950 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24953 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24955 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24956 update use of estimate_ipcp_clone_size_and_time.
24957 (estimate_local_effects): Update use of
24958 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24959 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24960 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24961 Return nonspecialized time.
24963 2017-05-04 Richard Biener <rguenther@suse.de>
24965 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24966 for the last VUSE which def dominates the PHI. Directly call
24968 (get_continuation_for_phi_1): Remove.
24970 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24972 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24973 to explain the use of truncating division. Cap the number of
24974 iterations to the maximum given by nb_iterations_upper_bound,
24977 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24979 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24980 * configure: Regenerate.
24981 * config.in: Regenerate.
24982 * config/i386/driver-mingw32.c: new file.
24983 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24984 * config.host: Link driver-mingw32.o on MinGW host.
24985 * doc/install.texi: Document new --enable-mingw-wildcard configure
24988 2017-05-04 Marek Polacek <polacek@redhat.com>
24990 PR tree-optimization/80612
24991 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24993 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24994 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24996 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24997 (movt splitter): Likewise.
24998 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24999 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25000 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25001 block for Thumb-1 with MOVT.
25002 (thumb2_legitimate_address_p): Move code block ...
25003 (can_avoid_literal_pool_for_label_p): ... into this new function.
25004 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25006 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25007 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25008 "M-profile targets with the MOVT instruction".
25010 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25012 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25013 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25014 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25016 2017-05-04 Martin Liska <mliska@suse.cz>
25018 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25019 variable cond_code.
25021 2017-05-04 Richard Biener <rguenther@suse.de>
25023 * tree.c (array_at_struct_end_p): Handle arrays at struct
25024 end with flexarrays more conservatively. Refactor and treat
25025 arrays of arrays or aggregates more strict. Fix
25026 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25027 * tree.c (array_at_struct_end_p): Adjust prototype.
25028 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25029 * gimple-fold.c (get_range_strlen): Likewise.
25030 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25032 2017-05-04 Richard Biener <rguenther@suse.de>
25034 PR tree-optimization/31130
25035 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25037 (supports_overflow_infinity): Likewise.
25038 (is_negative_overflow_infinity): Likewise.
25039 (is_positive_overflow_infinity): Likewise.
25040 (is_overflow_infinity): Likewise.
25041 (stmt_overflow_infinity): Likewise.
25042 (overflow_infinity_range_p): Likewise.
25043 (usable_range_p): Remove as always returning true.
25044 (make_overflow_infinity): Remove.
25045 (negative_overflow_infinity): Likewise.
25046 (positive_overflow_infinity): Likewise.
25047 (avoid_overflow_infinity): Likewise.
25048 (set_value_range): Adjust accordingly.
25049 (set_value_range_to_nonnegative): Likewise, remove now unused
25050 overflow_infinity arg.
25051 (vrp_operand_equal_p): Adjust.
25052 (update_value_range): Likewise.
25053 (range_int_cst_singleton_p): Likewise.
25054 (operand_less_p): Likewise.
25055 (compare_values_warnv): Likewise.
25056 (extract_range_for_var_from_comparison_expr): Likewise.
25057 (vrp_int_const_binop): Likewise.
25058 (zero_nonzero_bits_from_vr): Likewise.
25059 (extract_range_from_multiplicative_op_1): Likewise.
25060 (extract_range_from_binary_expr_1): Likewise.
25061 (extract_range_from_unary_expr): Likewise.
25062 (extract_range_from_comparison): Likewise.
25063 (extract_range_basic): Likewise.
25064 (adjust_range_with_scev): Likewise.
25065 (compare_ranges): Likewise.
25066 (compare_range_with_value): Likewise.
25067 (dump_value_range): Likewise.
25068 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25070 (simplify_cond_using_ranges): Adjust.
25072 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25074 * brig-builtins.def: Added a builtin for class_f64.
25075 * builtin-types.def: Added a builtin type needed by class_f64.
25077 2017-05-03 Jason Merrill <jason@redhat.com>
25079 * timevar.def: Add TV_CONSTEXPR.
25081 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25083 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25085 2017-05-03 Martin Jambor <mjambor@suse.cz>
25087 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25088 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25089 * ipa-cp.c (ipcp_propagate_stage): Do not call
25090 ipa_update_after_lto_read.
25091 * ipa-inline.c (ipa_inline): Likewise.
25093 2017-05-03 Martin Jambor <mjambor@suse.cz>
25095 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25096 tag. Added a default constructor and a destructor.
25097 (ipa_edge_args_sum_t): New class;
25098 (ipa_edge_args_sum): Declare.
25099 (ipa_edge_args_vector): Remove declaration.
25100 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25101 (ipa_free_edge_args_substructures): Remove declaration.
25102 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25103 (ipa_edge_args_info_available_for_edge_p): Likewise.
25104 * ipa-prop.c (ipa_edge_args_vector): Removed.
25105 (edge_removal_hook_holder): Likewise.
25106 (edge_duplication_hook_holder): Likewise.
25107 (ipa_edge_args_sum): New variable.
25108 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25109 ipa_edge_args_vector.
25110 (ipa_free_edge_args_substructures): Likewise.
25111 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25112 ipa_edge_args_vector.
25113 (ipa_edge_removal_hook): Turned into method
25114 ipa_edge_args_sum_t::remove.
25115 (ipa_edge_duplication_hook): Turned into method
25116 ipa_edge_args_sum_t::duplicate.
25117 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25118 registering edge hooks.
25119 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25120 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25121 ipa_edge_args_sum instead of ipa_edge_args_vector.
25122 * ipa-profile.c (ipa_profile): Likewise.
25124 2017-05-03 Martin Jambor <mjambor@suse.cz>
25126 * symbol-summary.h (function_summary): New method exists.
25127 (function_summary::symtab_removal): Deallocate through release.
25128 (call_summary): New class.
25129 (gt_ggc_mx): New overload.
25130 (gt_pch_nx): Likewise.
25131 (gt_pch_nx): Likewise.
25133 2017-05-03 Jeff Law <law@redhat.com>
25135 PR tree-optimization/78496
25136 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25137 from simplify_cond_using_ranges. Split off code to walk
25138 backwards through casts into ...
25139 (simplify_cond_using_ranges_2): New function.
25140 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25141 (execute_vrp): After identifying jump threads, call
25142 simplify_cond_using_ranges_2.
25144 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25147 * ipa-inline.h (inline_summary): Add ctor.
25148 (create_ggc): Do not use ggc_cleared_alloc.
25150 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25151 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25153 * gcc.c (handle_braces): Support escaping in switch matching
25155 * doc/invoke.texi (Spec Files): Document it.
25156 Remove superfluous @code markup in items.
25158 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25160 * diagnostic-show-locus.c (struct column_range): New struct.
25161 (get_affected_columns): New function.
25162 (get_printed_columns): New function.
25163 (struct correction): New struct.
25164 (correction::ensure_capacity): New function.
25165 (correction::ensure_terminated): New function.
25166 (struct line_corrections): New struct.
25167 (line_corrections::~line_corrections): New dtor.
25168 (line_corrections::add_hint): New function.
25169 (layout::print_trailing_fixits): Reimplement in terms of the new
25171 (selftest::test_overlapped_fixit_printing): New function.
25172 (selftest::diagnostic_show_locus_c_tests): Call it.
25174 2017-05-03 Nathan Sidwell <nathan@acm.org>
25176 Canonicalize canonical type hashing
25177 * tree.h (type_hash_canon_hash): Declare.
25178 * tree.c (type_hash_list, attribute_hash_list): Move into
25179 type_hash_canon_hash.
25180 (build_type_attribute_qual_variant): Break out hash code calc into
25181 type_hash_canon_hash.
25182 (type_hash_canon_hash): New. Generic type hash computation.
25183 (build_range_type_1, build_array_type_1, build_function_type,
25184 build_method_type_directly, build_offset_type, build_complex_type,
25185 make_vector_type): Call it.
25187 2017-05-03 Richard Biener <rguenther@suse.de>
25189 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25190 When all DRs have unknown misaligned do not always peel
25191 when there is a store but apply the same costing model as if
25192 there were only loads.
25194 2017-05-03 Richard Biener <rguenther@suse.de>
25197 PR tree-optimization/80492
25198 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25199 compare_base_decls returning dont-know properly.
25201 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25203 * config/arm/iterators.md (CCSI): New mode iterator.
25204 (arch): New mode attribute.
25205 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25206 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25207 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25208 code iterator for success result mode.
25209 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25210 the corresponding new insn generators.
25212 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25215 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25216 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25218 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25220 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25221 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25222 (DDR_A): Wrap DDR argument in brackets.
25223 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25224 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25225 (DDR_REVERSED_P): Likewise.
25227 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25229 PR tree-optimization/79472
25230 * tree-switch-conversion.c (struct switch_conv_info): Add
25231 contiguous_range and default_case_nonstandard fields.
25232 (collect_switch_conv_info): Compute contiguous_range and
25233 default_case_nonstandard fields, don't clear final_bb if
25234 contiguous_range and only the default case doesn't have the required
25236 (check_all_empty_except_final): Set default_case_nonstandard instead
25237 of failing if contiguous_range and the default case doesn't have empty
25239 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25240 and only the default case doesn't have the required constants. Skip
25242 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25243 if default_case_nonstandard.
25244 (build_constructors): Build constant 1 just once. Assert that default
25245 values aren't inserted in between cases if contiguous_range. Skip
25247 (build_arrays): Skip virtual phis.
25248 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25249 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25250 Handle virtual phis.
25251 (gen_inbound_check): Handle default_case_nonstandard case.
25252 (process_switch): Adjust check_final_bb caller. Call
25253 gather_default_values with the first non-default case instead of
25254 default case if default_case_nonstandard.
25256 2017-05-02 Nathan Sidwell <nathan@acm.org>
25258 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25259 check. Fix formatting.
25261 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25263 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25264 errors when comparing specialized and unspecialized times.
25266 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25268 * diagnostic-show-locus.c
25269 (layout::should_print_annotation_line_p): Make private.
25270 (layout::print_annotation_line): Make private.
25271 (layout::annotation_line_showed_range_p): Make private.
25272 (layout::show_ruler): Make private.
25273 (layout::print_source_line): Make private. Pass in line and
25274 line_width, rather than calling location_get_source_line. Drop
25276 (layout::print_leading_fixits): New method.
25277 (layout::print_any_fixits): Rename to...
25278 (layout::print_trailing_fixits): ...this, and make private.
25279 Don't print newline fixits.
25280 (diagnostic_show_locus): Move logic for printing one row into...
25281 (layout::print_line): ...this new function. Move the
25282 location_get_source_line call and error-handling from
25283 print_source_line to here. Call print_leading_fixits, and rename
25284 print_any_fixits to print_trailing_fixits.
25285 (selftest::test_fixit_insert_containing_newline): Update now that
25286 newlines are partially supported.
25287 (selftest::test_fixit_insert_containing_newline_2): New test.
25288 (selftest::test_fixit_replace_containing_newline): Update comments.
25289 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25290 * edit-context.c (class added_line): New class.
25291 (class edited_line): Describe newline handling in comment.
25292 (edited_line::actually_edited_p): New method.
25293 (edited_line::print_content): Delete redundant decl.
25294 (edited_line::m_predecessors): New field.
25295 (edited_file::print_content): Call edited_line::print_content.
25296 (edited_file::print_diff): Update to support newlines.
25297 (edited_file::print_diff_hunk): Likewise.
25298 (edited_file::print_run_of_changed_lines): New function.
25299 (edited_file::print_diff_line): Convert to...
25300 (print_diff_line): ...this.
25301 (edited_file::get_effective_line_count): New function.
25302 (edited_line::edited_line): Initialize new field m_predecessors.
25303 (edited_line::~edited_line): Clean up m_predecessors.
25304 (edited_line::apply_fixit): Handle newlines.
25305 (edited_line::get_effective_line_count): New function.
25306 (edited_line::print_content): New function.
25307 (edited_line::print_diff_lines): New function.
25308 (selftest::test_applying_fixits_insert_containing_newline): New
25310 (selftest::test_applying_fixits_replace_containing_newline): New
25312 (selftest::insert_line): New function.
25313 (selftest::test_applying_fixits_multiple_lines): Add example of
25315 (selftest::edit_context_c_tests): Call the new tests.
25317 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25319 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25320 parameter cand. Update dump information.
25321 (get_computation_cost): Update uses.
25323 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25325 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25326 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25327 (get_computation_at, rewrite_use_address): Update use of
25328 get_computation_aff.
25330 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25332 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25333 (get_computation): Delete.
25334 (get_computation_cost): Implement like get_computation_cost_at.
25335 Use get_computation_at.
25336 (get_computation_cost_at): Delete.
25337 (rewrite_use_nonlinear_expr): Use get_computation_at.
25338 (rewrite_use_compare, remove_unused_ivs): Ditto.
25340 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25342 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25344 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25346 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25347 (ivopts_global_cost_for_size): Rename parameter and update uses.
25348 (iv_ca_recount_cost): Update uses.
25349 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25350 candidates seperately in n_invs and n_cands.
25351 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25353 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25355 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25356 (find_inv_vars_cb): New.
25357 (find_depends): Renamed to ...
25358 (find_inv_vars): ... this.
25359 (add_candidate_1, force_var_cost): Call find_inv_vars.
25360 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25362 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25364 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25365 inv_vars. Add inv_exprs.
25366 (struct iv_cand): Rename depends_on to inv_vars.
25367 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25368 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25369 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25370 (dump_cand): Dump inv_vars.
25371 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25372 (record_invariant, find_depends, add_candidate_1): Ditto.
25373 (set_group_iv_cost, force_var_cost): Ditto.
25374 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25375 (get_computation_cost_at, get_computation_cost): Ditto.
25376 (determine_group_iv_cost_generic): Ditto.
25377 (determine_group_iv_cost_address): Ditto.
25378 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25379 (determine_group_iv_costs): Ditto.
25380 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25381 (iv_ca_set_remove_invariants): Renamed to ...
25382 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25383 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25384 (iv_ca_set_add_invariants): Renamed to ...
25385 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25386 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25387 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25388 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25389 (create_new_ivs): Remove useless dump.
25391 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25393 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25395 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25396 (iv_ca_set_no_cp, create_new_iv): Ditto.
25398 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25400 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25402 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25404 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25405 function tree_check2.
25407 2017-05-02 Martin Liska <mliska@suse.cz>
25409 * doc/gcov.texi: Add missing preposition.
25410 * gcov.c (function_info::function_info): Properly fill up
25411 all member variables.
25413 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25415 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25417 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25419 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25421 2017-05-02 Martin Liska <mliska@suse.cz>
25424 * lto-streamer-in.c (lto_read_tree_1): Remove
25425 LTO_STREAMER_DEBUG.
25426 * lto-streamer.c (struct tree_hash_entry): Likewise.
25427 (struct tree_entry_hasher): Likewise.
25428 (tree_entry_hasher::hash): Likewise.
25429 (tree_entry_hasher::equal): Likewise.
25430 (lto_streamer_init): Likewise.
25431 (lto_orig_address_map): Likewise.
25432 (lto_orig_address_get): Likewise.
25433 (lto_orig_address_remove): Likewise.
25434 * lto-streamer.h: Likewise.
25435 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25436 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25438 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25440 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25441 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25442 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25443 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25444 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25445 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25446 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25447 (mm_maskz_sub_ss): New intrinsics.
25448 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25449 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25450 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25451 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25452 (__builtin_ia32_subss_mask_round): New builtins.
25453 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25454 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25455 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25457 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25458 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25460 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25463 2017-05-02 Martin Jambor <mjambor@suse.cz>
25465 PR tree-optimization/78687
25466 * tree-sra.c (access): New field parent.
25467 (process_subtree_disqualification): New function.
25468 (disqualify_candidate): Call it.
25469 (build_accesses_from_assign): Reset write flag if creating an
25471 (build_access_subtree): Fill in parent field and also prpagate
25472 down grp_write flag.
25473 (create_artificial_child_access): New parameter set_grp_write, set
25474 grp_write to its value.
25475 (propagate_subaccesses_across_link): Also propagate grp_write flag
25477 (propagate_all_subaccesses): Push the closest parent back to work
25478 queue if add_access_to_work_queue returned true.
25480 2017-05-02 Richard Biener <rguenther@suse.de>
25482 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25483 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25484 -fstrict-overflow documentation.
25485 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25486 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25487 flag_strict_overflow.
25488 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25489 * lto-opts.c (lto_write_options): Do not stream it.
25490 * lto-wrapper.c (merge_and_complain): Do not handle it.
25491 * opts.c (default_options_table): Do not set -fstrict-overflow.
25492 (finish_options): Likewise do not clear it when sanitizing.
25493 * simplify-rtx.c (simplify_const_relational_operation): Do not
25494 test flag_strict_overflow.
25496 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25498 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25499 using enabled attribute.
25500 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25501 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25502 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25503 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25504 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25505 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25506 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25507 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25508 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25509 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25511 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25513 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25515 2017-05-02 Richard Biener <rguenther@suse.de>
25517 PR tree-optimization/80591
25519 2017-04-10 Richard Biener <rguenther@suse.de>
25521 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25524 2017-05-02 Richard Biener <rguenther@suse.de>
25526 PR tree-optimization/80549
25527 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25528 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25529 headers if they do not have a preheader.
25531 2017-05-02 Martin Liska <mliska@suse.cz>
25534 * common.opt: Fix typo.
25535 * doc/invoke.texi: Likewise.
25537 2017-05-01 Jan Beulich <jbeulich@suse.com>
25539 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25540 swapping, add (x,x,m,x,n) alternative.
25542 2017-05-01 Nathan Sidwell <nathan@acm.org>
25544 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25545 unnecessary unadjusted_alignment check.
25547 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25550 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25552 * gimplify.c (gimplify_cilk_detach): New function.
25553 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25554 * tree-core.h: Document EXPR_CILK_SPAWN.
25555 * tree.h (EXPR_CILK_SPAWN): Define.
25557 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25559 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25560 to use new fixit_hint representation, using the "replace" logic.
25561 (get_line_span_for_fixit_hint): Likewise.
25562 (layout::print_any_fixits): Likewise.
25563 (selftest::test_one_liner_many_fixits): Rename to...
25564 (selftest::test_one_liner_many_fixits_1): ...this, and update
25565 comment and expected output to reflect that the multiple fix-it
25566 hints are now consolidated into one insertion.
25567 (selftest::test_one_liner_many_fixits_2): New test.
25568 (selftest::test_diagnostic_show_locus_one_liner): Update for
25570 (selftest::test_fixit_consolidation): Update for fix-it API
25572 * diagnostic.c (print_parseable_fixits): Likewise.
25573 * edit-context.c (edited_line::m_line_events): Convert from
25574 auto_vec <line_event *> to auto_vec <line_event>.
25575 (class line_event): Convert from abstract base class to a concrete
25576 class, taking over the role of replace_event.
25577 (class insert_event): Delete.
25578 (class replace_event): Rename to class line_event. Convert to
25580 (edit_context::add_fixits): Reimplement.
25581 (edit_context::apply_insert): Delete.
25582 (edit_context::apply_replace): Rename to...
25583 (edit_context::apply_fixit): ...this. Convert to half-open range.
25584 (edited_file::apply_insert): Delete.
25585 (edited_file::apply_replace): Rename to...
25586 (edited_file::apply_fixit): ...this.
25587 (edited_line::~edited_line): Drop deletion of events.
25588 (edited_line::apply_insert): Delete.
25589 (edited_line::apply_replace): Rename to...
25590 (edited_line::apply_fixit): ...this. Convert to half-open range.
25591 Update for change to type of m_line_events.
25592 * edit-context.h (edit_context::apply_insert): Delete.
25593 (edit_context::apply_replace): Rename to...
25594 (edit_context::apply_fixit): ...this.
25596 2017-05-01 Martin Sebor <msebor@redhat.com>
25598 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25601 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25604 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25605 __get_cpuid_max returns 0.
25606 (__get_cpuid_count): Ditto.
25608 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25610 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25611 replacement expression is another instance of one of its arguments.
25613 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25616 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25617 check for stack push/pop autoinc.
25618 * config/i386/i386.c (ix86_agi_dependent): Return false
25619 if the only reason why modified_in_p returned true is that
25620 addr is SP based and set_insn is a push or pop.
25622 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25624 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25627 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25630 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25631 (account_size_time): Use two predicates - exec_pred and
25633 (evaluate_conditions_for_known_args): Compute both clause and
25635 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25636 (inline_summary_t::duplicate): Update.
25637 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25639 (compute_inline_parameters): Likewise.
25640 (estimate_edge_size_and_time): Update caluclation of time.
25641 (estimate_node_size_and_time): Compute both time and nonspecialized
25643 (estimate_ipcp_clone_size_and_time): Update.
25644 (inline_merge_summary): Update.
25645 (do_estimate_edge_time): Update.
25646 (do_estimate_edge_size): Update.
25647 (do_estimate_edge_hints): Update.
25648 (inline_read_section, inline_write_summary): Stream both new predicates.
25649 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25651 (compute_inlined_call_time): Cleanup.
25652 (big_speedup_p): Update.
25653 (edge_badness): Update.
25654 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25655 (size_time_entry): Replace predicate by exec_predicate and
25656 nonconst_predicate.
25657 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25658 (estimate_edge_time): Return also nonspec_time.
25659 (reset_edge_growth_cache): Update.
25661 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25663 PR rtl-optimization/80491
25664 * ifcvt.c (noce_process_if_block): When looking for x setter
25665 with missing else_bb, don't check only the insn right before
25666 cond_earliest, but look for the last insn that x is modified in
25667 within the same bb.
25669 PR rtl-optimization/80491
25670 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25672 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25674 PR tree-optimization/80487
25675 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25677 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25679 PR tree-optimization/79697
25680 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25681 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25682 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25684 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25685 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25687 2017-04-28 Martin Sebor <msebor@redhat.com>
25689 PR tree-optimization/80523
25690 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25691 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25693 (maybe_warn, format_directive, parse_directive): Use new functions.
25694 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25696 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25698 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25700 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25702 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25703 target_header_dir): Set correctly.
25704 * configure: Regenerated.
25705 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25706 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25707 instead of SYSTEM_HEADER_DIR.
25709 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25711 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25712 (estimate_local_effects): Likewise.
25713 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25714 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25715 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25716 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25717 do_estimate_edge_time, estimate_edge_time): Likewise.
25718 * ipa-inline-analysis.c (estimate_node_size_and_time,
25719 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25720 (estimate_time_after_inlining): Remove.
25722 2017-04-28 Martin Liska <mliska@suse.cz>
25724 * doc/gcov.texi: Enhance documentation of gcov.
25726 2017-04-28 Martin Liska <mliska@suse.cz>
25728 * doc/gcov.texi: Sort options in alphabetic order.
25729 * doc/gcov-dump.texi: Likewise.
25730 * doc/gcov-tool.texi: Likewise.
25731 * gcov.c (print_usage): Likewise.
25732 * gcov-dump.c (print_usage): Likewise.
25733 * gcov-tool.c (print_merge_usage_message): Likewise.
25734 (print_rewrite_usage_message): Likewise.
25735 (print_overlap_usage_message): Likewise.
25737 2017-04-28 Martin Liska <mliska@suse.cz>
25739 PR gcov-profile/53915
25740 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25742 2017-04-28 Martin Liska <mliska@suse.cz>
25744 PR gcov-profile/79891
25745 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25746 is marked by compiler as living on a line.
25747 (get_cycles_count): Remove usage of the union.
25748 (output_intermediate_file): Likewise.
25749 (find_source): Fix GNU coding style.
25750 (accumulate_line_counts): Remove old non-all block mode.
25751 (output_lines): Remove usage of the union.
25752 * profile.c (output_location): Include all BBs, even if
25753 belonging to a same line (and file) as a previous BB.
25755 2017-04-28 Martin Liska <mliska@suse.cz>
25757 * gcov.c (process_args): Handle new argument 'w'.
25758 (read_graph_file): Assign ID to BBs.
25759 (output_branch_count): Display BB # if verbose flag is set.
25760 (output_lines): Likewise for arcs.
25761 (print_usage): Add '--verbose' option help.
25762 * doc/gcov.texi: Document --verbose (-w) option.
25764 2017-04-28 Martin Liska <mliska@suse.cz>
25766 * gcov.c (struct block_location_info): New struct.
25767 (process_file): Fill up the new structure.
25768 (read_graph_file): Replace usage of encoding by the newly added
25770 (add_line_counts): Likewise.
25771 (accumulate_line_counts): Remove usage of the union.
25772 (function_info::function_info): New function.
25773 (function_info::~function_info): Likewise.
25774 (process_file): Call delete instead of release_function.
25775 (release_function): Release the function.
25776 (release_structures): Call delete instead of release_function.
25777 (solve_flow_graph): Replace usage of num_blocks.
25778 (find_exception_blocks): Likewise.
25779 (output_lines): Fix GNU coding style.
25781 2017-04-28 Martin Liska <mliska@suse.cz>
25784 * coverage.c (coverage_remove_note_file): New function.
25785 * coverage.h: Declare the function.
25786 * toplev.c (finalize): Clean if an error has been seen.
25788 2017-04-28 Martin Liska <mliska@suse.cz>
25790 PR gcov-profile/80031
25791 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25792 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25793 * gcov.c (read_graph_file): Read just number of blocks.
25794 * profile.c (branch_prob): Do not stream 0 flags per a basic
25797 2017-04-28 Martin Liska <mliska@suse.cz>
25799 * gcov-dump.c (tag_*): Add new argument to declarations.
25800 (dump_gcov_file): Likewise.
25801 (tag_blocks): Add and use new argument depth.
25802 (tag_arcs): Likewise.
25803 (tag_lines): Likewise.
25804 (tag_counters): Likewise.
25805 (tag_summary): Likewise.
25806 (dump_working_sets): Use depth to do a proper indentation.
25808 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25811 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25812 * symtab.c (symtab_node::debug_symtab): Move definition here.
25814 2017-04-28 Richard Biener <rguenther@suse.de>
25816 * lto-streamer.h (LTO_major_version): Bump to 7.
25818 2017-04-28 Richard Biener <rguenther@suse.de>
25820 * tree-vrp.c (assert_info): New struct.
25821 (add_assert_info): New helper.
25822 (register_edge_assert_for_2): Refactor to add asserts to a vector
25824 (register_edge_assert_for_1): Likewise.
25825 (register_edge_assert_for): Likewise.
25826 (finish_register_edge_assert_for): New helper actually registering
25827 asserts where live on edge.
25828 (find_conditional_asserts): Adjust.
25829 (find_switch_asserts): Likewise.
25830 (evrp_dom_walker::try_find_new_range): Generalize.
25831 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25833 2017-04-27 Marek Polacek <polacek@redhat.com>
25836 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25837 arg10 and arg11 to itype.
25839 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25841 * doc/extend.texi (Object Size Checking): Improve grammar.
25843 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25846 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25847 that the logic for permitting reciprocal estimates matches that
25850 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25853 * tree.c (type_cache_hasher::equal): Only compare
25854 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25855 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25856 non-aggregate element types.
25857 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25858 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25860 2017-04-27 Richard Biener <rguenther@suse.de>
25862 PR middle-end/80533
25863 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25864 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25865 keeping a reference to a trailing array.
25867 2017-04-27 Richard Biener <rguenther@suse.de>
25869 PR middle-end/80539
25870 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25871 being in loop-closed SSA form conservatively.
25872 (chrec_fold_multiply_poly_poly): Likewise.
25874 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25876 PR middle-end/79665
25877 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25878 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25880 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25883 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25884 (aarch64_function_arg_alignment): Return unsigned int again, but still
25885 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25886 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25887 Don't emit -Wpsabi note.
25888 (aarch64_function_arg_boundary): Likewise.
25889 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25892 2017-04-26 Nathan Sidwell <nathan@acm.org>
25894 * tree.h (crc32_unsigned_n): Declare.
25895 (crc32_unsigned, crc32_unsigned): Make inline.
25896 * tree.c (crc32_unsigned_bits): Replace with ...
25897 (crc32_unsigned_n): ... this.
25898 (crc32_unsigned, crc32_byte): Remove.
25899 (crc32_string): Remove unnecessary braces.
25901 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25903 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25904 * ipa-inline-analysis.c (MAX_TIME): Remove.
25905 (account_size_time): Use sreal for time.
25906 (dump_inline_summary): Update.
25907 (estimate_function_body_sizes): Update.
25908 (estimate_edge_size_and_time): Update.
25909 (estimate_calls_size_and_time): Update.
25910 (estimate_node_size_and_time): Update.
25911 (inline_merge_summary): Update.
25912 (inline_update_overall_summary): Update.
25913 (estimate_time_after_inlining): Update.
25914 (inline_read_section): Update.
25915 (inline_write_summary): Update.
25916 * ipa-inline.c (compute_uninlined_call_time): Update.
25917 (compute_inlined_call_time): Update.
25918 (recursive_inlining): Update.
25919 (inline_small_functions): Update.
25920 (dump_overall_stats): Update.
25921 * ipa-inline.h: Include sreal.h.
25922 (size_time_entry): Turn time to sreal.
25923 (inline_summary): Turn self_time nad time to sreal.
25925 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25927 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25929 (sreal::stream_out, sreal::stream_in): New.
25930 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25932 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25934 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25937 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25940 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25942 (dimode_scalar_chain::compute_convert_gain): Ditto.
25943 (dimode_scalar_chain::make_vector_copies): Ditto.
25944 (dimode_scalar_chain::convert_reg): Ditto.
25945 (dimode_scalar_chain::convert_insn): Ditto.
25946 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25947 (VI248_AVX512BW_1): New mode iterator.
25948 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25949 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25952 2017-04-25 Martin Sebor <msebor@redhat.com>
25954 PR tree-optimization/80497
25955 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25956 constants are representable in HOST_WIDE_INT.
25957 (parse_directive): Ditto.
25959 2017-04-25 Martin Sebor <msebor@redhat.com>
25962 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25963 (new_zero_array): Adjust signature.
25964 (dom_info::dom_init): Used unsigned rather that size_t.
25965 (dom_info::dom_info): Same.
25967 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25968 Jakub Jelinek <jakub@redhat.com>
25971 * config/arm/arm.c: Include gimple.h.
25972 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25973 returns negative, increment ncrn only if it returned positive.
25974 (arm_needs_doubleword_align): Return int instead of bool,
25975 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25976 members, but if there is any such non-FIELD_DECL
25977 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25978 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25979 returns negative, increment nregs only if it returned positive.
25980 (arm_setup_incoming_varargs): Likewise.
25981 (arm_function_arg_boundary): Emit -Wpsabi note if
25982 arm_needs_doubleword_align returns negative, return
25983 DOUBLEWORD_ALIGNMENT only if it returned positive.
25985 2017-04-25 Marek Polacek <polacek@redhat.com>
25988 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25989 first argument to type.
25991 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25994 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25995 type checks to test for compatibility instead of equality.
25997 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25998 Jakub Jelinek <jakub@redhat.com>
26001 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26003 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26004 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26005 the alignment computation, but return their maximum in warn_alignment.
26006 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26007 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26009 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26010 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26013 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26015 * config/arc/simdext.md (dmpyh): Fix typo.
26017 2017-04-25 Richard Biener <rguenther@suse.de>
26019 PR tree-optimization/80492
26020 * alias.c (compare_base_decls): Handle registers with asm
26021 specification conservatively.
26022 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26023 compare_base_decls returning dont-know properly.
26025 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26027 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26028 (legitimate_offset_address_p): New function.
26029 (arc_legitimate_address_p): Use above function.
26031 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26033 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26035 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26037 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26038 ACCH registers whenever they are available.
26040 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26042 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26043 double regs fix when not used.
26045 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26047 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26049 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26050 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26052 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26054 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26055 when emitting short ADD instructions.
26057 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26059 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26061 (cmpsi_cc_c_insn): Likewise.
26062 (cbranchsi4_scratch): Compute proper instruction length using
26063 compact_hreg_operand.
26064 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26066 2017-04-25 Richard Biener <rguenther@suse.de>
26068 PR middle-end/80509
26069 * passes.c (pass_manager::pass_manager): Initialize
26070 m_name_to_pass_map.
26072 2017-04-25 Richard Biener <rguenther@suse.de>
26074 PR tree-optimization/79201
26075 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26077 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26080 * config/s390/vector.md: Split MEM->GPR vector moves for
26081 non-s_operand addresses.
26083 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26086 * config/s390/predicates.md (reload_const_wide_int_operand): New
26088 * config/s390/s390.md ("movti"): Remove d/P alternative.
26089 ("movti_bigconst"): New pattern definition.
26091 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26094 * s390-protos.h (s390_expand_cs_hqi): Removed.
26095 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26096 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26097 modes as well as CCZ1mode and CCZmode.
26098 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26099 signature of s390_emit_compare_and_swap.
26100 (s390_expand_cs_hqi): Likewise, make static.
26101 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26102 compare-and-swap, in some cases.
26103 (s390_expand_cs): Wrapper function.
26104 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26106 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26107 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26108 patterns for small and large integers. Forbid symref memory operands.
26109 Move expander to s390.c. Require cc register.
26110 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26111 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26112 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26113 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26114 symref memory operands. Remove CC mode and call s390_match_ccmode
26116 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26118 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26120 * config/s390/s390.md (define_peephole2): New peephole to help
26121 combining the load-and-test pattern with volatile memory.
26123 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26125 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26126 with CCZmode for TARGET_Z196.
26128 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26130 PR rtl-optimization/80501
26131 * combine.c (make_compound_operation_int): Set subreg_code to SET
26132 even for AND with mask of the sign bit of mode.
26134 PR rtl-optimization/80500
26135 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26136 sum's initial value.
26138 2017-04-25 Julian Brown <julian@codesourcery.com>
26139 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26141 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26143 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26145 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26147 2017-04-25 Julian Brown <julian@codesourcery.com>
26148 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26150 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26151 (thunderx2t99_sha): New Reservation.
26153 2017-04-25 Julian Brown <julian@codesourcery.com>
26154 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26156 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26157 type for 1-element load.
26159 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26161 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26163 2017-04-24 Martin Jambor <mjambor@suse.cz>
26165 PR tree-optimization/80293
26166 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26167 char arrays not totally scalarizable if it is false.
26168 (analyze_all_variable_accesses): Pass correct value in the new
26169 parameter. Add a statistics counter.
26171 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26173 PR middle-end/79931
26174 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26176 2017-04-24 Richard Biener <rguenther@suse.de>
26178 PR tree-optimization/80494
26179 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26180 out for complex types.
26182 2017-04-24 Richard Biener <rguenther@suse.de>
26184 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26185 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26186 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26187 (DFS): Adjust and never fail.
26188 (sccvn_dom_walker::fail): Remove.
26189 (sccvn_dom_walker::before_dom_children): Adjust.
26190 (run_scc_vn): Likewise and never fail.
26191 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26192 (pass_fre::execute): Likewise.
26194 2017-04-24 Richard Biener <rguenther@suse.de>
26196 PR tree-optimization/79725
26197 * tree-ssa-sink.c (statement_sink_location): Return whether
26198 failure reason was zero uses. Move that check later.
26199 (sink_code_in_bb): Deal with zero uses by removing the stmt
26202 2017-04-24 Richard Biener <rguenther@suse.de>
26205 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26206 pointer-based references.
26208 2017-04-24 Richard Biener <rguenther@suse.de>
26211 * pass_manager.h (pass_manager::operator new): Remove.
26212 (pass_manager::operator delete): Likewise.
26213 * passes.c (pass_manager::operator new): Remove.
26214 (pass_manager::operator delete): Likewise.
26215 (pass_manager::pass_manager): Zero individual pass members.
26217 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26220 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26221 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26222 Check "XEXP (src, 1)" operand here.
26223 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26224 Check "XEXP (src, 1)" operand here.
26225 (dimode_scalar_chain::make_vector_copies): Detect count register
26226 of a shift instruction. Zero extend count register from QImode
26227 to DImode to satisfy vector shift pattern count operand predicate.
26228 Substitute vector shift count operand with a DImode copy.
26229 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26232 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26234 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26235 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26236 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26237 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26238 (UNSPEC_NOREX_MEM): Remove definition.
26240 2017-04-21 Richard Biener <rguenther@suse.de>
26242 PR tree-optimization/79547
26243 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26244 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26245 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26246 without any constraints.
26248 2017-04-21 Richard Biener <rguenther@suse.de>
26250 PR tree-optimization/78847
26251 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26253 2017-04-21 Richard Biener <rguenther@suse.de>
26255 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26256 (build_distinct_type_copy): Likewise.
26257 (build_variant_type_copy): Likewise.
26258 * tree.c (build_qualified_type): Pass down mem-stat info.
26259 (build_distinct_type_copy): Likewise.
26260 (build_variant_type_copy): Likewise.
26262 2017-04-21 Richard Biener <rguenther@suse.de>
26264 PR tree-optimization/80237
26265 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26267 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26268 for a simplified result.
26270 2016-04-21 Richard Biener <rguenther@suse.de>
26272 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26273 sth as strict as a simple_iv but a chrec without symbols and an
26274 operand defined in the loop we are peeling (and not some subloop).
26275 (propagate_constants_for_unrolling): Propagate all constants.
26277 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26280 * config/i386/i386.c (print_reg): Remove assert for disalowed
26281 regno values, call output_operand_lossage instead.
26283 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26286 * config/i386/constraints.md (Yc): New register constraint.
26287 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26288 Use Yc constraint for alternative 2 of operand 0. Remove
26289 preferred_for_speed attribute.
26291 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26293 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26294 lastprivate clauses in SIMT case.
26296 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26298 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26300 2017-04-20 Richard Biener <rguenther@suse.de>
26302 PR tree-optimization/57796
26303 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26304 as N scalar stores.
26305 (vect_model_load_cost): Cost gathers as N scalar loads.
26307 2017-04-20 Richard Biener <rguenther@suse.de>
26309 * ggc-page.c (ggc_allocated_p): Rename to ...
26310 (safe_lookup_page_table_entry): ... this and return the lookup
26312 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26314 2017-04-20 Richard Biener <rguenther@suse.de>
26316 PR tree-optimization/80453
26317 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26318 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26319 from the conditions.
26320 (vn_phi_eq): Pass them down.
26321 (vn_phi_lookup): Record them.
26322 (vn_phi_insert): Likewise.
26324 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26326 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26327 uninitialized variable warning to avoid buffer overrun.
26329 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26332 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26333 is suppressed for '{ 0 }' in C.
26335 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26337 * BASE-VER: Set to 8.0.0.
26339 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26341 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26342 priority .init_array and .fini_array section with SECTION_NOTYPE
26345 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26347 PR middle-end/80423
26348 * tree.h (build_array_type): Add typeless_storage default argument.
26349 * tree.c (type_cache_hasher::equal): Also compare
26350 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26351 (build_array_type): Add typeless_storage argument, set
26352 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26354 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26355 (build_array_type): Likewise. Add typeless_storage argument.
26357 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26358 Jakub Jelinek <jakub@redhat.com>
26360 PR tree-optimization/80426
26361 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26362 operation on symbolic operands, also compute the overflow for the
26363 invariant part when the operation degenerates into a negation.
26365 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26368 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26369 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26372 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26374 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26377 * config/avr/avr.c (tree.h): Include it.
26378 (cgraph.h): Include it.
26379 (avr_encode_section_info): Don't warn for uninitialized progmem
26380 variable if it's just an alias.
26382 2017-04-19 Richard Biener <rguenther@suse.de>
26385 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26386 when needed by AutoPGO.
26388 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26391 * doc/lto.texi: Remove an extra 'that'.
26393 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26395 PR rtl-optimization/80429
26396 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26397 are only used in debug insns.
26399 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26400 Vladimir Makarov <vmakarov@redhat.com>
26402 * config/sparc/predicates.md (input_operand): Add comment. Return
26403 true for any memory operand when LRA is in progress.
26404 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26406 2017-04-18 Jeff Law <law@redhat.com>
26409 * mips.md ({return,simple_return}_internal): Do not overwrite
26412 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26414 PR tree-optimization/80443
26415 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26416 instead of adding 1, subtract -1 and similarly instead of subtracting
26419 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26421 PR rtl-optimization/80357
26422 * haifa-sched.c (tmp_bitmap): New variable.
26423 (model_recompute): Handle duplicate use records.
26424 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26425 (free_global_sched_pressure_data): Free it.
26427 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26430 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26431 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26432 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26433 instead of SYSTEM_HEADER_DIR.
26435 2017-04-18 Jeff Law <law@redhat.com>
26437 PR middle-end/80422
26438 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26439 predecessors after walking up the insn chain.
26441 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26444 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26445 sizetype type into debug info.
26447 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26450 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26451 unneeded test for TARGET_UPPER_REGS_SF.
26452 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26454 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26457 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26458 instead of gsi_after_labels.
26460 2017-04-18 Jeff Law <law@redhat.com>
26462 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26466 2017-04-13 Jeff Law <law@redhat.com>
26467 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26468 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26470 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26473 * config/avr/avr.c (intl.h): Include it.
26474 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26476 2017-04-18 Martin Liska <mliska@suse.cz>
26478 PR gcov-profile/78783
26479 * gcov-tool.c (gcov_output_files): Validate that destination
26480 file is either removed by the tool or by a user.
26482 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26483 Guy Benyei <guybe@mellanox.com>
26485 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26486 block, and do not negate it, the stored id is already negative.
26488 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26490 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26492 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26495 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26496 masks of options that should be turned off if the VSX vector
26497 options are turned off.
26498 (OTHER_P8_VECTOR_MASKS): Likewise.
26499 (OTHER_VSX_VECTOR_MASKS): Likewise.
26500 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26501 rs6000_disable_incompatible_switches to validate no type switches
26503 (rs6000_incompatible_switch): New function to disallow turning on
26504 other vector options if -mno-vsx, -mno-power8-vector, or
26505 -mno-power9-vector are specified.
26507 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26509 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26511 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26513 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26514 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26515 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26516 (ARG_POINTER_CFA_OFFSET): Likewise.
26518 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26520 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26521 conditions to take advantage of various optimizations.
26523 2017-04-13 Jeff Law <law@redhat.com>
26525 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26526 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26527 (zero_extendsidi2_dext): Likewise.
26529 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26532 * fold-const.c (fold_ternary_loc): Revert
26533 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26536 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26538 PR rtl-optimization/80343
26539 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26540 new scratch pseudo.
26542 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26545 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26546 to ubsan_encode_value.
26548 2017-04-13 Jeff Law <law@redhat.com>
26550 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26551 appearing in DEBUG_INSNs.
26553 2017-04-13 Martin Liska <mliska@suse.cz>
26555 PR gcov-profile/80413
26556 * gcov-io.c (gcov_write_string): Copy to buffer just when
26557 allocated size is greater than zero.
26559 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26562 * dwarf2out.c (decls_for_scope): Ignore declarations of
26563 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26565 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26568 * ipa-visibility.c (non_local_p): Fix typos.
26569 (localize_node): When localizing symbol in same comdat group,
26570 dissolve the group only when we know external symbols are going
26572 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26574 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26576 PR tree-optimization/79390
26577 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26578 order does not result in usable sequence, retry with reversed operand
26584 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26585 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26586 op0 instead of fold_convert_loc (loc, type, arg0).
26588 2017-04-12 Jeff Law <law@redhat.com>
26590 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26591 has a delay slot in the generated code.
26593 * config/cris/cris.md (cris_preferred_reload_class): Return
26594 GENNONACR_REGS rather than GENERAL_REGS.
26596 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26599 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26600 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26601 signedness of the result type.
26603 2017-04-12 Richard Biener <rguenther@suse.de>
26604 Jeff Law <law@redhat.com>
26606 PR tree-optimization/80359
26607 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26608 trim stores to TARGET_MEM_REFs.
26610 2017-04-12 Richard Biener <rguenther@suse.de>
26612 PR tree-optimization/79390
26613 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26614 threading case even more.
26616 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26619 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26620 for quad_address_p for TImode, instead of just not indexed_address.
26622 2017-04-12 Richard Biener <rguenther@suse.de>
26623 Bernd Edlinger <bernd.edlinger@hotmail.de>
26625 PR middle-end/79671
26626 * alias.c (component_uses_parent_alias_set_from): Handle
26627 TYPE_TYPELESS_STORAGE.
26628 (get_alias_set): Likewise.
26629 * tree-core.h (tree_type_common): Add typeless_storage flag.
26630 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26631 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26632 for types containing members with TYPE_TYPELESS_STORAGE.
26633 (place_field): Likewise.
26634 (layout_type): Likewise for ARRAY_TYPE.
26635 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26636 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26637 TYPE_TYPELESS_STORAGE.
26638 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26640 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26643 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26644 first argument to type.
26646 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26650 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26651 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26652 (rs6000_expand_binop_builtin): Likewise.
26653 (rs6000_expand_ternop_builtin): Likewise; also add missing
26654 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26655 vshasigma built-ins.
26656 * doc/extend.texi: Document that vec_xxpermdi's third argument
26657 must be a constant.
26659 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26661 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26662 Use shift_const cost parameter when calculating gain of STV shifts.
26664 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26666 PR rtl-optimization/70478
26667 * lra-constraints.c (process_alt_operands): Check memory for
26668 disfavoring memory insn operand.
26670 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26672 PR middle-end/80100
26673 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26674 left shift in unsigned HOST_WIDE_INT type.
26676 PR rtl-optimization/80385
26677 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26678 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26681 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26682 if they have any depend clauses.
26684 2017-04-11 Martin Liska <mliska@suse.cz>
26687 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26688 * ipa-split.c (split_function): Create a local comdat symbol
26689 if caller is in a comdat group.
26691 2017-04-11 Martin Liska <mliska@suse.cz>
26694 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26697 2017-04-11 Martin Sebor <msebor@redhat.com>
26699 PR middle-end/80364
26700 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26701 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26703 (directive::set_width, directive::set_precision, format_character):
26705 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26708 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26711 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26712 conflict, set target->arch_name instead of target->cpu_name.
26714 2017-04-11 Richard Biener <rguenther@suse.de>
26716 PR tree-optimization/80374
26717 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26718 build_zero_cst, remove fold_convertible_p check again.
26720 2017-04-11 Martin Liska <mliska@suse.cz>
26723 * ubsan.c (instrument_object_size): Do not instrument register
26726 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26729 * config/i386/i386-builtin-types.def
26730 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26731 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26732 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26733 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26734 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26735 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26736 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26737 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26738 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26739 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26740 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26741 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26742 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26743 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26744 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26745 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26746 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26747 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26748 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26749 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26750 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26751 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26752 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26753 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26754 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26755 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26756 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26757 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26758 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26759 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26760 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26761 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26762 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26763 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26764 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26765 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26766 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26767 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26768 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26769 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26770 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26771 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26772 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26773 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26774 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26775 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26777 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26778 flag to second_arg_count, handle 4 argument function type _COUNT
26779 aliases, handle second_arg_count on second argument rather than last.
26781 2017-04-10 Jeff Law <law@redhat.com>
26783 PR tree-optimization/80374
26784 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26785 record anything if we can not convert integer_zero_node to the
26788 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26791 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26792 Enhance special handling given to the TARGET_P9_MINMAX option in
26793 relation to certain other options.
26795 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26797 PR tree-optimization/80153
26798 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26799 remove POINTER_PLUS_EXPR's base part directly, rather than through
26802 2017-04-10 Richard Biener <rguenther@suse.de>
26803 Bin Cheng <bin.cheng@arm.com>
26805 PR tree-optimization/80153
26806 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26807 the first element of pointer type aff_tree. Build result expr in
26809 (add_elt_to_tree): Convert to type unconditionally. Remove other
26810 fold_convert calls.
26811 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26812 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26814 2017-04-10 Richard Biener <rguenther@suse.de>
26816 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26819 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26821 PR rtl-optimization/70478
26822 * lra-constraints.c (curr_small_class_check): New.
26823 (update_and_check_small_class_inputs): New.
26824 (process_alt_operands): Update curr_small_class_check. Disfavor
26825 alternative insn memory operands. Check available regs for small
26828 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26831 * config/mips/mips.opt (-mvirt): Update description.
26832 * doc/invoke.texi (-mvirt): Likewise.
26834 2017-04-10 Richard Biener <rguenther@suse.de>
26836 PR middle-end/80362
26837 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26838 looking for NEGATE_EXPR in -A / -B to A / B folding.
26840 2017-04-10 Martin Liska <mliska@suse.cz>
26842 PR gcov-profile/80224
26843 * gcov.c (print_usage): Fix usage string.
26844 (get_gcov_intermediate_filename): Remove.
26845 (output_gcov_file): Use both for normal and intermediate format.
26846 (generate_results): Do not initialize special file for
26847 intermediate format.
26849 2017-04-10 Richard Biener <rguenther@suse.de>
26851 PR tree-optimization/80304
26852 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26855 2017-04-10 Nathan Sidwell <nathan@acm.org>
26858 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26859 (rs6000_init_builtins): Use it.
26861 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26863 * config/arm/arm.md (<mrc>): Add mode to SET source.
26864 (<mrrc>): Likewise.
26866 2017-04-10 Richard Biener <rguenther@suse.de>
26868 PR middle-end/80344
26869 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26871 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26874 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26875 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26876 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26877 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26878 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26879 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26880 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26881 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26882 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26883 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26884 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26885 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26886 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26887 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26888 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26889 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26890 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26891 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26892 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26893 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26894 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26895 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26896 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26898 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26900 PR rtl-optimization/70478
26901 * lra-constraints.c: Reverse the last patch.
26903 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26905 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26906 Add comment for WCHAR_T.
26908 2017-04-08 Martin Liska <mliska@suse.cz>
26911 2017-04-07 Martin Liska <mliska@suse.cz>
26914 * ipa-split.c (split_function): Add function part to a same comdat
26917 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26920 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26922 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26924 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26925 (vec_pairing): ...to this.
26926 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26927 (rs6000_sched_init): Adjust for name change.
26928 (struct rs6000_sched_context): Likewise.
26929 (rs6000_init_sched_context): Likewise.
26930 (rs6000_set_sched_context): Likewise.
26932 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26938 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26940 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26941 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26942 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26944 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26946 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26948 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26950 PR rtl-optimization/70703
26951 * ira-color.c (update_conflict_hard_regno_costs): Use
26952 int64_t instead of HOST_WIDE_INT.
26954 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26956 PR rtl-optimization/70478
26957 * lra-constraints.c (process_alt_operands): Disfavor alternative
26958 insn memory operands.
26960 2017-04-07 Jeff Law <law@redhat.com>
26962 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26963 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26965 2017-04-07 Martin Liska <mliska@suse.cz>
26968 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26969 Show error message instead of an ICE.
26971 2017-04-07 Martin Liska <mliska@suse.cz>
26974 * ipa-split.c (split_function): Add function part to a same comdat
26977 2017-04-07 Richard Biener <rguenther@suse.de>
26979 PR middle-end/80341
26980 * tree.c (get_unwidened): Also handle ! for_type case for
26982 * convert.c (do_narrow): Split out from ...
26983 (convert_to_integer_1): ... here. Do not pass final truncation
26984 type to get_unwidened for TRUNC_DIV_EXPR.
26986 2017-04-07 Richard Biener <rguenther@suse.de>
26988 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26990 (aff_combination_const): Adjust.
26991 (aff_combination_scale): Likewise.
26992 (aff_combination_add_elt): Likewise.
26993 (aff_combination_add_cst): Likewise.
26994 (aff_combination_convert): Likewise.
26995 (add_elt_to_tree): Likewise. Remove unused argument.
26996 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26998 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27000 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27002 * config/arm/arm.c (arm_default_short_enums): Use
27003 ARM_DEFAULT_SHORT_ENUMS.
27004 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27006 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27009 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27010 members with redundant out-of-class redeclaration.
27012 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27015 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27016 * config/i386/i386.md (*zero_extendsidi2):
27017 Add (?*x,*x) and (?*v,*v) alternatives.
27019 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27022 * config/i386/i386.c (ix86_expand_builtin)
27023 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27024 mode from insn data. Convert operands to insn operand mode.
27025 Copy operands that don't satisfy insn predicate to a register.
27027 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27029 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27032 2017-04-06 Richard Biener <rguenther@suse.de>
27034 PR tree-optimization/80334
27035 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27036 preserve alignment of accesses.
27038 2017-04-06 Richard Biener <rguenther@suse.de>
27040 PR tree-optimization/80262
27041 * tree-sra.c (build_ref_for_offset): Preserve address-space
27043 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27044 Drop useless address-space information on MEM_REF offsets.
27046 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27048 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27050 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27052 PR rtl-optimization/70703
27053 * ira-color.c (update_conflict_hard_regno_costs): Use
27054 HOST_WIDE_INT instead of long.
27056 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27059 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27060 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27062 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27063 for x86_64 target. Handle -m3dnowa option.
27065 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27067 PR rtl-optimization/70703
27068 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27069 (update_conflict_hard_regno_costs): Use long instead of unsigned
27070 arithmetic for cost calculation.
27072 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27073 Bernd Edlinger <bernd.edlinger@hotmail.de>
27076 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27079 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27082 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27083 ptr_mode with Pmode throughout.
27084 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27085 into probe_stack_range and use DImode.
27087 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27090 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27091 call_eh_return is true.
27093 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27095 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27096 Initialize last_match_fntype_index.
27098 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27101 * tree-nvr.c: Include internal-fn.h.
27102 (pass_return_slot::execute): Ignore internal calls without
27105 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27106 Richard Biener <rguenther@suse.de>
27109 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27110 captures used multiple times, except for the last use.
27111 * generic-match-head.c: Include gimplify.h.
27113 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27115 PR tree-optimization/79390
27116 * target.h (struct noce_if_info): Declare.
27117 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27118 * target.def (noce_conversion_profitable_p): New target hook.
27119 * ifcvt.h (struct noce_if_info): New type, moved from ...
27120 * ifcvt.c (struct noce_if_info): ... here.
27121 (noce_conversion_profitable_p): Renamed to ...
27122 (default_noce_conversion_profitable_p): ... this. No longer
27124 (noce_try_store_flag_constants, noce_try_addcc,
27125 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27126 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27127 instead of noce_conversion_profitable_p.
27128 * config/i386/i386.c: Include ifcvt.h.
27129 (ix86_option_override_internal): Don't override
27130 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27131 (ix86_noce_conversion_profitable_p): New function.
27132 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27133 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27134 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27135 * doc/tm.texi: Regenerated.
27137 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27139 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27142 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27145 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27146 instructions for small multiply cores.
27148 2017-04-04 Jeff Law <law@redhat.com>
27150 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27152 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27153 that are not set by the loop over the elements.
27155 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27158 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27159 int mode, convert_modes it to mode as unsigned, otherwise use
27160 lowpart_subreg to mode rather than SImode.
27161 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27162 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27163 Use DImode instead of SImode for the shift count operand.
27164 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27167 2017-04-04 Richard Biener <rguenther@suse.de>
27169 PR middle-end/80281
27170 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27171 arithmetic done for the negate or the plus. Simplify.
27172 (A - (-B) -> A + B): Likewise.
27173 * fold-const.c (split_tree): Make sure to not negate pointers.
27175 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27177 PR rtl-optimization/60818
27178 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27179 a compare of comparisons with the thing compared if this results
27180 in a different machine mode.
27182 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27184 * alias.c (base_alias_check): Fix typo in comment.
27185 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27186 * cgraphunit.c (symbol_table::compile): Likewise.
27187 * collect2.c (maybe_run_lto_and_relink): Likewise.
27188 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27189 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27190 * config/avr/avr.c (avr_map_op_t): Likewise.
27191 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27192 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27193 * config/epiphany/epiphany.md (movcc): Likewise.
27194 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27195 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27197 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27198 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27199 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27200 * config/sh/sh.c (sh_rtx_costs): Likewise.
27201 * fold-const.c (fold_truth_andor): Likewise.
27202 * genautomata.c (collapse_flag): Likewise.
27203 * gengtype.h (struct type::u::s): Likewise.
27204 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27205 * input.c (FORMAT_AMOUNT): Likewise.
27206 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27207 (known_aggs_to_agg_replacement_list): Likewise.
27208 * ipa-inline-analysis.c: Likewise.
27209 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27210 * ipa-polymorphic-call.c
27211 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27212 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27213 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27215 * modulo-sched.c (apply_reg_moves): Likewise.
27216 * omp-expand.c (build_omp_regions_1): Likewise.
27217 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27218 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27219 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27220 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27221 * value-prof.c: Likewise.
27222 * var-tracking.c (val_reset): Likewise.
27224 2017-04-03 Richard Biener <rguenther@suse.de>
27226 PR tree-optimization/80275
27227 * fold-const.c (split_address_to_core_and_offset): Handle
27230 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27232 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27233 descriptors is at least equal to that of functions.
27235 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27237 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27239 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27242 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27243 (mov<IMOD4:mode>): New expander.
27244 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27246 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27248 PR rtl-optimization/79405
27249 * fwprop.c (propagations_left): New variable.
27250 (forward_propagate_into): Decrement it.
27251 (fwprop_init): Initialize it.
27252 (fw_prop): If the variable has reached zero, stop propagating.
27253 (fwprop_addr): Ditto.
27255 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27258 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27259 a FUNCTION_DECL, pass it as decl instead of origin to
27262 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27264 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27267 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27270 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27271 TARGET_VSX_SMALL_INTEGER.
27273 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27275 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27276 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27278 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27280 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27281 extraction from odd-numbered MSA register.
27283 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27285 PR middle-end/80173
27286 * expmed.c (store_bit_field_1): Don't attempt to create
27287 a word subreg out of hard registers wider than word if they
27288 have HARD_REGNO_NREGS of 1 for their mode.
27290 PR middle-end/80163
27291 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27292 conversions to integer types wider than word and pointer.
27295 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27296 (rtx_equal_for_cselib_p): Pass 0 to it.
27297 * cselib.c (cselib_hasher::equal): Likewise.
27298 (rtx_equal_for_cselib_1): Add depth argument. If depth
27299 is 128, don't look up VALUE locs and punt. Increment
27300 depth in recursive calls when walking VALUE locs.
27302 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27304 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27305 (make_gcov_file_name): Use the canonical path name for generating
27307 (read_line): Fix handling of files with ascii null bytes.
27309 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27311 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27312 to initialise a vector register instead
27313 of using a const_int.
27315 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27317 PR translation/80189
27318 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27319 diagnostic messages.
27321 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27324 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27325 (dfp_diex_<mode>): Update mode of operand 1.
27326 * doc/extend.texi (dxex, dxexq): Document change to return type.
27327 (diex, diexq): Document change to argument type.
27329 2017-03-30 Martin Jambor <mjambor@suse.cz>
27332 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27333 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27334 it reflects the signature changes performed at the callee side.
27335 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27336 to cgraph_build_function_type_skip_args.
27337 (build_function_decl_skip_args): Adjust call to the above function.
27339 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27342 * config/i386/sse.md
27343 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27344 register as dest whenever it is a MEM not rtx_equal_p to the
27345 corresponding dup operand, and when forcing into reg move the
27346 reg into the memory afterwards.
27347 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27348 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27349 for the force_reg mode.
27350 (avx512vl_vextractf128<mode>): Use register as dest either
27351 always when a MEM, or when it is a MEM not rtx_equal_p to the
27352 corresponding dup operand, or even not when it is a CONST_VECTOR
27353 depending on the mode and lo vs. hi.
27354 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27356 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27357 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27358 Likewise. Require that operands[2] is even.
27359 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27360 Remove extraneous parens. Require that operands[2] is a multiple
27362 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27363 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27364 disallow memory then.
27366 2017-03-30 Richard Biener <rguenther@suse.de>
27368 PR tree-optimization/77498
27369 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27370 to non-constants over backedges.
27372 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27374 PR rtl-optimization/80233
27375 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27376 as last_combined_insn. Do not test for BARRIER_P separately.
27378 2017-03-29 Andreas Schwab <schwab@suse.de>
27381 * calls.c (prepare_call_address): Convert funexp to Pmode before
27382 copying to temp reg.
27384 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27386 PR tree-optimization/80158
27387 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27388 Handle possible future case of more than one alternate
27390 (replace_rhs_if_not_dup): Likewise.
27391 (replace_one_candidate): Likewise.
27393 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27395 PR rtl-optimization/80193
27396 * ira.c (ira): Do not check allocation for LRA.
27398 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27400 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27401 (nvptx_output_simt_exit): Declare.
27402 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27403 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27404 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27405 (nvptx_declare_function_name): ...here. Emit declaration of local
27406 memory space buffer for omp_simt_enter insn.
27407 (nvptx_output_unisimt_switch): New.
27408 (nvptx_output_softstack_switch): New.
27409 (nvptx_output_simt_enter): New.
27410 (nvptx_output_simt_exit): New.
27411 * config/nvptx/nvptx.h (struct machine_function): New fields
27412 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27413 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27414 (UNSPECV_SIMT_EXIT): Ditto.
27415 (omp_simt_enter_insn): New insn.
27416 (omp_simt_enter): New expansion.
27417 (omp_simt_exit): New insn.
27418 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27420 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27421 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27422 (expand_GOMP_SIMT_EXIT): New.
27423 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27424 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27425 (GOMP_SIMT_EXIT): Ditto.
27426 * target-insns.def (omp_simt_enter): New insn.
27427 (omp_simt_exit): Ditto.
27428 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27430 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27431 (lower_rec_input_clauses): Likewise.
27432 (lower_lastprivate_clauses): Handle SIMT privatization.
27434 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27435 (ompdevlow_adjust_simt_enter): New.
27436 (find_simtpriv_var_op): New.
27437 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27438 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27440 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27441 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27442 (copy_decl_for_dup_finish): Ditto.
27444 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27446 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27449 * config/i386/i386.c (ix86_option_override_internal): Always
27450 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27452 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27454 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27456 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27458 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27459 mark new edge's irreducible flag accordign to it.
27460 (vect_do_peeling): Check loop preheader edge's irreducible flag
27461 and pass it to function slpeel_add_loop_guard.
27463 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27465 PR tree-optimization/80218
27466 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27467 Update block frequencies and counts.
27469 2017-03-28 Richard Biener <rguenther@suse.de>
27471 PR tree-optimization/78644
27472 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27473 of a simplification result we may not use it at all.
27475 2017-03-28 Richard Biener <rguenther@suse.de>
27478 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27479 without arguments, generate default definition of a SSA name.
27481 2017-03-28 Richard Biener <rguenther@suse.de>
27483 PR middle-end/80222
27484 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27485 TYPE_REF_CAN_ALIAS_ALL references.
27486 * fold-const.c (fold_indirect_ref_1): Likewise.
27488 2017-03-28 Martin Liska <mliska@suse.cz>
27491 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27492 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27494 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27495 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27497 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27498 (EXTRA_SPECS): Define.
27499 (SUBTARGET_EXTRA_SPECS): Likewise.
27500 (SUBTARGET_CPP_SPEC): Likewise.
27501 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27502 SUBTARGET_EXTRA_SPECS.
27503 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27505 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27507 * config/arc/simdext.md (vst64_insn): Update pattern.
27508 (vld32wh_insn): Likewise.
27509 (vld32wl_insn): Likewise.
27510 (vld64_insn): Likewise.
27511 (vld32_insn): Likewise.
27513 2017-03-28 Marek Polacek <polacek@redhat.com>
27516 * fold-const.c (fold_comparison): Use protected_set_expr_location
27517 instead of SET_EXPR_LOCATION.
27519 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27521 * tree.c (add_expr): Avoid name lookup warning.
27523 2017-03-27 Jeff Law <law@redhat.com>
27525 PR tree-optimization/80216
27526 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27527 function name. Limit recursion depth.
27528 (record_temporary_equivalences): Corresponding changes.
27530 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27532 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27535 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27538 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27540 * cfgcleanup.c (reg_note_cfa_p): New array.
27541 (insns_have_identical_cfa_notes): New function.
27542 (old_insns_match_p): Don't cross-jump in between /f
27543 and non-/f instructions. If both i1 and i2 are frame related,
27544 verify all CFA notes, their order and content.
27546 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27549 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27550 HImode and SImode with zero extend to DImode to one insn.
27551 (bswap<mode>2_extenddi): Likewise.
27552 (bswapsi2_extenddi): Likewise.
27553 (bswaphi2_extendsi): Likewise.
27554 (bswaphi2): Combine bswap HImode and SImode into one insn.
27555 Separate memory insns from swapping register.
27556 (bswapsi2): Likewise.
27557 (bswap<mode>2): Likewise.
27558 (bswaphi2_internal): Delete, no longer used.
27559 (bswapsi2_internal): Likewise.
27560 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27561 store, and gpr<-gpr swap insns.
27562 (bswap<mode>2_store): Likewise.
27563 (bswaphi2_reg): Register only splitter, combine with the splitter.
27564 (bswaphi2 splitter): Likewise.
27565 (bswapsi2_reg): Likewise.
27566 (bswapsi2 splitter): Likewise.
27567 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27568 the insns into load, store, and register/register insns.
27569 (bswapdi2_ldbrx): Likewise.
27570 (bswapdi2_load): Likewise.
27571 (bswapdi2_store): Likewise.
27572 (bswapdi2_reg): Likewise.
27574 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27576 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27577 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27579 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27582 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27584 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27585 special handling for target option conflicts between dform
27586 options (-mpower9-dform, -mpower9-dform-vector,
27587 -mpower9-dform-scalar) and -mno-direct-move.
27589 2017-03-27 Richard Biener <rguenther@suse.de>
27591 PR tree-optimization/80181
27592 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27594 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27596 * config/arc/predicates.md (move_double_src_operand): Replace the
27597 call to move_double_src_operand with a call to address_operand.
27599 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27601 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27602 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27603 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27605 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27607 * config/arc/predicates.md (long_immediate_loadstore_operand):
27608 Consider scaled addresses cases.
27610 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27612 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27613 restored when in interrupt.
27614 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27615 doesn't have delay slot.
27617 2017-03-27 Richard Biener <rguenther@suse.de>
27620 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27621 inlined thunk clones.
27623 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27626 * asan.c (instrument_derefs): Copy over last operand from
27627 original COMPONENT_REF to the new COMPONENT_REF with
27628 DECL_BIT_FIELD_REPRESENTATIVE.
27629 * ubsan.c (instrument_object_size): Likewise.
27631 2017-03-27 Richard Biener <rguenther@suse.de>
27633 PR tree-optimization/80170
27634 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27635 sure DR/SCEV didnt fold in constants we do not see when looking
27636 at the reference base alignment.
27638 2017-03-27 Richard Biener <rguenther@suse.de>
27640 PR middle-end/80171
27641 * gimple-fold.c (fold_ctor_reference): Properly guard against
27642 NULL return value from canonicalize_constructor_val.
27644 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27647 * config/i386/i386.c (ix86_expand_builtin)
27648 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27649 flags reg setting and flags reg using instructions.
27650 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27651 clobbering instructions to zero extend op2.
27653 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27655 * doc/install.texi (Configuration) <--with-aix-soname>:
27656 Update link to AIX ld.
27658 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27660 PR rtl-optimization/80160
27661 PR rtl-optimization/80159
27662 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27663 reg_alternate_class into account.
27665 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27668 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27669 to consider in curr_insn_transform.
27671 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27673 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27674 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27675 and emit_mode_inner.
27677 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27679 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27680 argument to the overloaded builtin variants. Use the new flag to
27681 deprecate certain builtin variants.
27682 * config/s390/s390-builtin-types.def: Add new builtin types.
27683 * config/s390/s390-builtins.h: Support new flags field for
27684 overloaded builtins.
27685 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27686 (s390_macro_to_expand): Enable vector float data type.
27687 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27688 builtins by incrementing the __VEC__ version number.
27689 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27691 (s390_resolve_overloaded_builtin): Emit error messages depending
27692 on the builtin flags.
27693 * config/s390/s390.c (s390_expand_builtin): Support additional
27694 flags argument. Change error message to match the messages
27695 emitted in s390-c.c.
27696 * config/s390/s390.md: New UNSPEC_* constants.
27697 (op_type): Add new instruction types.
27698 * config/s390/vecintrin.h: Add new builtins and test data class
27700 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27701 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27702 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27703 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27704 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27705 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27707 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27708 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27709 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27710 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27712 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27713 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27714 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27715 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27717 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27718 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27719 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27720 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27721 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27722 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27723 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27725 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27726 ("vec_scatter_element<V_HW_4:mode>_DI")
27727 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27728 ("vec_fpint<mode>", "vflls")
27729 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27730 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27731 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27732 ("*vec_cmphe<mode>_cc"): ... these.
27734 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27735 mode constant instead of magic value.
27737 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27739 * config/s390/s390.c (s390_expand_vec_compare): Support other
27740 vector floating point modes than just V2DF.
27741 (s390_expand_vcond): Likewise.
27742 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27743 (s390_cannot_change_mode_class): Prevent mode changes between TF
27744 and V1TF in vector registers.
27745 * config/s390/s390.md (DF, SF): New mode attributes.
27746 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27747 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27748 SFmode support for VRs.
27749 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27751 (VFT, VF_HW): New mode iterators.
27752 (vw, sdx): New mode attributes.
27753 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27754 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27755 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27756 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27757 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27758 also the new vector floating point modes. Renaming to ...
27760 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27761 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27762 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27763 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27764 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27765 ("vec_unordered<mode>"): ... these.
27767 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27768 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27769 ("*vec_extendv2df"): New insn definitions.
27771 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27773 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27774 ("mulditi3_2", "*muldi3_sign"): New patterns.
27775 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27776 rename the pattern definition.
27778 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27780 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27782 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27784 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27786 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27787 instruction if possible.
27788 * config/s390/vector.md (vec_halfnumelts): New mode
27790 ("*vec_vllezlf<mode>"): New pattern.
27792 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27794 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27795 ("popcountv4si2", "popcountv2di2"): Rename to ...
27796 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27797 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27799 ("popcount<mode>2_vxe"): New pattern.
27801 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27803 * common/config/s390/s390-common.c (processor_flags_table): Add
27805 * config.gcc: Add arch12.
27806 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27807 Default to arch12 for unknown CPU model numbers.
27808 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27809 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27810 PROCESSOR_max sanity check.
27811 * config/s390/s390-opts.h (enum processor_type): Add
27813 * config/s390/s390.c (processor_table): Add arch12.
27814 (s390_expand_builtin): Add check for B_VXE flag.
27815 (s390_issue_rate): Add PROCESSOR_ARCH12.
27816 (s390_get_sched_attrmask): Likewise.
27817 (s390_get_unit_mask): Likewise.
27818 (s390_sched_score): Enable z13 scheduling for arch12.
27819 (s390_sched_reorder): Likewise.
27820 (s390_sched_variable_issue): Likewise.
27821 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27823 (s390_tune_attr): Use z13 scheduling also for arch12.
27824 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27825 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27826 (TARGET_VXE_P): New macros.
27827 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27828 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27829 * config/s390/s390.opt: Add arch12 as processor_type.
27831 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27833 * config/s390/s390.md
27834 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27835 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27836 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27838 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27839 Rename expanders to ...
27841 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27842 ("fixuns_truncdddi2_emu"): ... these.
27844 ("fixuns_trunc<mode>si2_emu"): New expander.
27846 ("*fixuns_truncdfdi2_z13"): Rename to ...
27847 ("*fixuns_truncdfdi2_vx"): ... this.
27849 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27851 * config/s390/2964.md: Remove the single element vector compare
27852 instructions which are no longer used.
27853 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27855 (s390_canonicalize_comparison): Remove handling of DFmode
27857 (s390_expand_vec_compare_scalar): Remove function.
27858 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27859 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27861 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27863 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27865 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27867 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27868 will anyway by matched by mov<mode>_64dfp.
27870 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27872 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27873 vlef/vstef. Add missing operand to vleif.
27875 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27877 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27878 pair for all vector types with 64 bit elements.
27879 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27880 * config/s390/vector.md (V_HW_64): ... here.
27881 (V_128_NOSINGLE): New mode iterator.
27882 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27883 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27884 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27885 ("*vec_load_pairv2di"): Change to ...
27886 ("*vec_load_pair<mode>"): ... this one.
27888 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27890 * config/s390/constraints.md: Add comments.
27891 (jKK): Reject element sizes > 8 bytes.
27892 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27894 * config/s390/s390.md: Add the s_operand checks formerly in
27895 s390_split_ok_p to various splitters where they are still
27897 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27898 for 128 bit vectors. Plus two splitters.
27900 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27902 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27905 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27908 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27909 error if the boundary argument is not constant.
27911 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27913 PR rtl-optimization/80112
27914 * loop-doloop.c (doloop_condition_get): Don't check condition
27915 if cmp isn't SET with IF_THEN_ELSE src.
27917 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27919 PR tree-optimization/80158
27920 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27921 replacing a candidate statement, also replace it for the
27922 candidate's alternate interpretation.
27923 (replace_rhs_if_not_dup): Likewise.
27924 (replace_one_candidate): Likewise.
27926 2017-03-24 Richard Biener <rguenther@suse.de>
27928 PR tree-optimization/80167
27929 * graphite-isl-ast-to-gimple.c
27930 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27932 (translate_isl_ast_to_gimple::get_rename): Likewise.
27934 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27936 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27937 handling of certain combinations of target options, including the
27938 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27939 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27941 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27944 * config/arm/arm.md (*load_multiple): Add reload_completed to
27945 matching condition.
27947 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27948 Richard Biener <rguenth@suse.de>
27950 PR tree-optimization/79908
27951 PR tree-optimization/80136
27952 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27953 been cast away, gimplify_and_add suffices.
27955 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27957 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27959 2017-03-23 Richard Biener <rguenther@suse.de>
27961 PR tree-optimization/80032
27962 * gimplify.c (gimple_push_cleanup): Forced unconditional
27963 cleanups still have to go to the conditional_cleanups
27966 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27968 PR tree-optimization/80072
27969 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27971 (next_operand_entry_id): Change type to unsigned int.
27972 (sort_by_operand_rank): Make sure to return the right return value
27973 even if unsigned fields are bigger than INT_MAX.
27974 (struct oecount): Change cnt and id type to unsigned int.
27975 (oecount_hasher::equal): Formatting fix.
27976 (oecount_cmp): Make sure to return the right return value
27977 even if unsigned fields are bigger than INT_MAX.
27978 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27981 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27982 TREE_READONLY on result if writing it more than once.
27985 * doc/invoke.texi (-fsanitize=thread): Document that with
27986 -fnon-call-exceptions atomics are not able to throw
27990 * tsan.c: Include tree-eh.h.
27991 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27992 maybe_clean_or_replace_eh_stmt where needed.
27993 (instrument_memory_accesses): Add cfg_changed argument.
27994 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27995 if it returned true.
27996 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27998 PR rtl-optimization/63191
27999 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28000 wrapper function, moved the whole old content into ...
28001 (ix86_delegitimize_address_1): ... this. New inline function.
28002 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28003 true as last argument instead of ix86_delegitimize_address.
28005 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28007 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28008 cortexa57_branch_cost.
28010 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28012 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28014 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28017 * doc/md.texi (Constraints): Document wA constraint.
28018 * config/rs6000/constraints.md (wA): New.
28019 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28020 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28021 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28022 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28024 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28027 * gimplify.c (is_oacc_declared): New function.
28028 (oacc_default_clause): Use it to set default flags for acc declared
28029 variables inside parallel regions.
28030 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28031 declared variables.
28032 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28033 declare attribute to any decl as necessary.
28035 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28038 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28039 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28040 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28041 (arm_arch_lpae): This.
28042 * config/arm/arm.c (arm_arch7ve): Rename into ...
28043 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28044 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28047 2017-03-22 Martin Liska <mliska@suse.cz>
28050 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28051 error message instead of an ICE.
28053 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28055 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28057 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28059 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28061 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28064 2017-03-21 Martin Sebor <msebor@redhat.com>
28066 * doc/extend.texi: Use "cannot" instead of "can't."
28067 * doc/hostconfig.texi: Same.
28068 * doc/install.texi: Same.
28069 * doc/invoke.texi: Same.
28070 * doc/loop.texi: Same.
28071 * doc/md.texi: Same.
28072 * doc/objc.texi: Same.
28073 * doc/rtl.texi: Same.
28074 * doc/tm.texi: Same.
28075 * doc/tm.texi.in: Same.
28076 * doc/trouble.texi: Same.
28078 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28081 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28082 (collect_checksum_attributes): Set it.
28083 (die_checksum_ordered): Use it.
28085 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28087 PR tree-optimization/79908
28088 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28089 change: For a VA_ARG whose LHS has been cast away, use
28090 force_gimple_operand to construct the side effects.
28092 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28094 PR translation/80001
28095 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28096 more amenable to translation.
28097 (oacc_loop_auto_partitions): Likewise.
28099 2017-03-21 Marek Polacek <polacek@redhat.com>
28100 Martin Sebor <msebor@redhat.com>
28102 PR tree-optimization/80109
28103 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28104 on INTEGRAL_TYPE_P.
28106 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28107 Segher Boessenkool <segher@kernel.crashing.org>
28110 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28111 check reg_used_between_p between insn and one of succ or succ2
28112 depending on if succ is artificial insn not inserted into insn
28115 2017-03-21 Martin Liska <mliska@suse.cz>
28117 PR gcov-profile/80081
28118 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28119 * doc/gcc.texi: Include gcov-dump stuff.
28120 * doc/gcov-dump.texi: New file.
28122 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28124 PR rtl-optimization/79150
28125 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28126 conditional jump, if the jump is the last insn of the loop.
28128 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28129 Richard Biener <rguenth@suse.de>
28131 PR tree-optimization/79908
28132 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28133 been cast away, use force_gimple_operand to construct the side
28136 2017-03-21 Martin Liska <mliska@suse.cz>
28138 PR libfortran/79956
28139 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28142 2017-03-21 Brad Spengler <spender@grsecurity.net>
28145 * plugin.c (htab_hash_plugin): New function.
28146 (add_new_plugin): Use it and adjust.
28147 (parse_plugin_arg_opt): Adjust.
28148 (init_one_plugin): Likewise.
28150 2017-03-21 Richard Biener <rguenther@suse.de>
28152 PR tree-optimization/80032
28153 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28154 if set force the cleanup to happen unconditionally.
28155 (gimplify_target_expr): Push inserted clobbers with force_uncond
28156 to avoid them being removed by control-dependent DCE.
28158 2017-03-21 Richard Biener <rguenther@suse.de>
28160 PR tree-optimization/80122
28161 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28162 va_arg_pack_len when the inlined call stmt requires pack
28164 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28166 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28169 * tsan.c (instrument_builtin_call): If the memory model argument
28170 is not a constant, assume it is valid.
28173 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28176 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28178 PR rtl-optimization/79910
28179 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28180 if its dest is used by an insn before I2 (other than the combined
28181 insns themselves, which are properly handled already).
28183 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28186 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28188 * combine.c (record_used_regs): New static function.
28189 (try_combine): Handle situations where there is an additional
28190 instruction between I2 and I3 which needs to have a LOG_LINK
28194 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28196 * combine.c (try_combine): Delete redundant i1 test. Call
28197 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28199 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28202 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28203 alternatives 13/14.
28205 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28207 PR tree-optimization/80054
28208 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28209 the optimization if a PHI or any of its arguments is not dominated
28210 by the candidate's basis. Use gphi* rather than gimple* as
28212 (replace_profitable_candidates): Clean up a gimple* variable that
28213 should be a gphi* variable.
28215 2017-03-20 Martin Sebor <msebor@redhat.com>
28218 * doc/extend.texi (attribute constructor): Document present limitation.
28220 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28223 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28224 __POWER9_VECTOR__ #ifdef control, change template definition to
28225 use Power9-specific built-in function.
28226 (vec_any_eq): Likewise.
28227 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28228 to control outcomes from this test.
28229 (vector_ae_<mode>p): For VEC_F modes, likewise.
28231 2017-03-20 Ian Lance Taylor <iant@google.com>
28233 * config/i386/i386.c (ix86_function_regparm): Save an extra
28234 register for -fsplit-stack with DECL_STATIC_CHAIN.
28236 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28239 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28240 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28242 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28244 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28246 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28249 2017-03-20 Marek Polacek <polacek@redhat.com>
28252 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28254 2017-03-20 Richard Biener <rguenther@suse.de>
28256 PR tree-optimization/80113
28257 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28258 allocate extra SSA name for PHI def.
28259 (add_close_phis_to_outer_loops): Likewise.
28260 (add_close_phis_to_merge_points): Likewise.
28261 (copy_loop_close_phi_args): Likewise.
28262 (copy_cond_phi_nodes): Likewise.
28264 2017-03-20 Martin Liska <mliska@suse.cz>
28266 PR middle-end/79753
28267 * tree-chkp.c (chkp_build_returned_bound): Do not build
28268 returned bounds for a LHS that's not a BOUNDED_P type.
28270 2017-03-20 Martin Liska <mliska@suse.cz>
28274 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28275 COMPLEX_CST and VECTOR_CST.
28277 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28280 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28281 target operand. A new splitter adds the clobber statement in case
28282 the target operand is dead anyway.
28284 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28286 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28287 to age-old versions of binutils and glibc.
28289 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28291 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28293 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28295 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28297 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28299 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28300 requirement for binutils 2.13.
28302 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28304 * combine.c (try_combine): Delete redundant i1 test. Call
28305 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28307 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28309 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28310 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28312 <riscv64-*-elf>: Re-arrange section
28313 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28314 <riscv32-*-linux>: Likewise.
28315 <riscv64-*-elf>: Likewise
28316 <riscv64-*-linux>: Likewise.
28318 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28321 * aarch64.opt(verbose-cost-dump): Fix typo.
28323 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28326 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28327 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28329 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28331 * reload.c (find_reloads): When reloading a nonoffsettable address,
28332 use RELOAD_OTHER for it and its address reloads.
28334 PR rtl-optimization/79910
28335 * combine.c (record_used_regs): New static function.
28336 (try_combine): Handle situations where there is an additional
28337 instruction between I2 and I3 which needs to have a LOG_LINK
28340 2017-03-17 Jeff Law <law@redhat.com>
28342 PR tree-optimization/71437
28343 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28344 conditional in the hash table first.
28345 (vrp_dom_walker::before_dom_children): Extract condition from
28346 ASSERT_EXPR. Record condition, its inverion and any implied
28347 conditions as well.
28349 2017-03-17 Marek Polacek <polacek@redhat.com>
28350 Markus Trippelsdorf <markus@trippelsdorf.de>
28352 PR tree-optimization/80079
28353 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28356 2017-03-17 Richard Biener <rguenther@suse.de>
28358 PR middle-end/80075
28359 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28360 Properly verify the LHS before the RHS possibly claims to be
28362 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28365 2017-03-17 Martin Jambor <mjambor@suse.cz>
28367 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28368 (List of -O2 options): Likewise.
28369 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28372 2017-03-17 Tom de Vries <tom@codesourcery.com>
28374 * gcov-dump.c (print_usage): Print bug_report_url.
28376 2017-03-17 Richard Biener <rguenther@suse.de>
28378 PR middle-end/80050
28379 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28380 (parser::peek): Likewise.
28382 2017-03-17 Richard Biener <rguenther@suse.de>
28384 PR tree-optimization/80048
28385 * sese.c (free_sese_info): Properly release rename_map and
28386 copied_bb_map elements.
28388 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28390 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28391 Add linked-list forward and backlinks. Insert on
28392 construction, remove on destruction.
28393 (class pass_store_merging): Add m_stores_head field.
28394 (pass_store_merging::terminate_and_process_all_chains):
28395 Iterate over m_stores_head list.
28396 (pass_store_merging::terminate_all_aliasing_chains):
28398 (pass_store_merging::execute): Check for debug stmts first.
28399 Push new chains onto the m_stores_head stack.
28401 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28404 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28405 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28406 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28408 2017-03-16 Jeff Law <law@redhat.com>
28410 PR tree-optimization/71437
28411 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28412 member function. Implementation moved into after_dom_children
28413 member function and into the threader's thread_outgoing_edges
28415 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28416 some code into new thread_outgoing_edges.
28417 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28418 definition. Simplify marker handling (do it here). Assume we always
28419 have the available expression and the const/copies tables.
28420 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28422 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28423 * tree-vrp.c (equiv_stack): No longer file scoped.
28424 (vrp_dom_walker): New class.
28425 (vrp_dom_walker::before_dom_children): New member function.
28426 (vrp_dom_walker::after_dom_children): Likewise.
28427 (identify_jump_threads): Setup domwalker. Use it rather than
28428 walking edges in a random order by hand. Simplify setup/finalization.
28429 (finalize_jump_threads): Remove.
28430 (vrp_finalize): Do not call identify_jump_threads here.
28431 (execute_vrp): Do it here instead and call thread_through_all_blocks
28434 PR tree-optimization/71437
28435 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28437 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28439 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28440 (dom_opt_dom_walker::thread_across_edge): Remove
28441 handle_dominating_asserts argument. All callers changed.
28442 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28443 changes. Remove calls to lhs_of_dominating_assert. Other
28444 uses of handle_dominating_asserts turn into unconditional code
28445 (simplify_control_stmt_condition_1): Likewise.
28446 (simplify_control_stmt_condition): Likewise.
28447 (thread_through_normal_block, thread_across_edge): Likewise.
28448 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28449 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28450 object if it is not an SSA_NAME.
28451 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28452 before calling into the VRP specific simplifiers.
28453 (identify_jump_threads): Remove handle_dominating_asserts
28456 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28459 * tree-diagnostic.c (default_tree_printer): No longer static.
28460 * tree-diagnostic.h (default_tree_printer): New prototype.
28462 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28464 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28465 Change ins into fmov.
28467 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28469 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28470 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28471 Use h_con constraint for operand 1.
28472 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28473 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28475 2017-03-15 Jeff Law <law@redhat.com>
28477 PR tree-optimization/71437
28478 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28479 (record_temporary_equivalences): Use it.
28481 PR tree-optimization/71437
28482 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28483 tree-ssa-scopedtables.
28484 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28485 (record_conditions, record_cond, vuse_eq): Likewise.
28486 (record_edge_info): Adjust to API tweak of record_conditions.
28487 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28488 (record_temporary_equivalences, optimize_stmt): Likewise.
28489 (eliminate_redundant_computations): Likewise.
28490 (record_equivalences_from_stmt): Likewise.
28491 * tree-ssa-scopedtables.c: Include options.h and params.h.
28492 (vuse_eq): New function, moved from tree-ssa-dom.c
28493 (build_and_record_new_cond): Likewise.
28494 (record_conditions): Likewise. Accept vector of conditions rather
28495 than edge_equivalence structure for first argument.
28496 for the first argument.
28497 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28498 from tree-ssa-dom.c.
28499 (avail_exprs_stack::record_cond): Likewise.
28500 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28501 from tree-ssa-dom.c.
28502 (avail_exprs_stack): Add new member functions lookup_avail_expr
28504 (record_conditions): Declare.
28506 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28509 * lra-constraints.c (process_alt_operands): Increase reject for
28510 reloading an input/output operand.
28512 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28515 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28516 insns to convert from signed/unsigned char/short to IEEE 128-bit
28518 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28520 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28523 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28524 subreg of inner mode for values already in registers.
28526 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28528 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28529 iteration reg is used after the loop.
28531 2017-03-14 Martin Sebor <msebor@redhat.com>
28533 PR tree-optimization/79800
28534 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28535 precision in negative-positive range.
28536 (format_floating): Call non-const overload with adjusted precision.
28538 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28541 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28544 2017-03-14 Martin Sebor <msebor@redhat.com>
28546 PR middle-end/80020
28547 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28548 * builtins.def (aligned_alloc): Use it.
28551 * Makefile.in (GTFILES): Add calls.c.
28552 * calls.c: Include "gt-calls.h".
28554 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28556 PR rtl-optimization/79728
28557 * regs.h (struct target_regs): New field
28558 x_contains_allocatable_regs_of_mode.
28559 (contains_allocatable_regs_of_mode): New macro.
28560 * reginfo.c (init_reg_sets_1): Initialize it, and change
28561 contains_reg_of_mode so it includes global regs as well.
28562 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28563 rather than contains_regs_of_mode.
28565 2017-03-14 Martin Liska <mliska@suse.cz>
28567 * doc/invoke.texi: Document options that can't be combined with
28568 -fcheck-pointer-bounds.
28570 2017-03-14 Martin Liska <mliska@suse.cz>
28572 PR middle-end/79831
28573 * doc/invoke.texi (-Wchkp): Document the option.
28575 2017-03-14 Martin Liska <mliska@suse.cz>
28577 * Makefile.in: Install gcov-dump.
28579 2017-03-14 Martin Liska <mliska@suse.cz>
28581 * multiple_target.c (expand_target_clones): Bail out for
28582 an invalid attribute.
28584 2017-03-14 Richard Biener <rguenther@suse.de>
28586 * alias.c (struct alias_set_entry): Pack properly.
28587 * cfgloop.h (struct loop): Likewise.
28588 * cse.c (struct set): Likewise.
28589 * ipa-utils.c (struct searchc_env): Likewise.
28590 * loop-invariant.c (struct invariant): Likewise.
28591 * lra-remat.c (struct cand): Likewise.
28592 * recog.c (struct change_t): Likewise.
28593 * rtl.h (struct address_info): Likewise.
28594 * symbol-summary.h (function_summary): Likewise.
28595 * tree-loop-distribution.c (struct partition): Likewise.
28596 * tree-object-size.c (struct object_size_info): Likewise.
28597 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28598 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28599 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28600 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28601 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28602 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28603 (struct _stmt_vec_info): Likewise.
28605 2017-03-14 Martin Liska <mliska@suse.cz>
28608 * multiple_target.c (create_dispatcher_calls): Check that
28609 a target can create a function dispatcher.
28611 2017-03-14 Martin Liska <mliska@suse.cz>
28614 * multiple_target.c (expand_target_clones): Drop local.local
28615 flag for default implementation.
28617 2017-03-14 Richard Biener <rguenther@suse.de>
28619 PR tree-optimization/80030
28620 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28622 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28624 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28625 gcc_fallthrough() instead of __attribute__((fallthrough));
28627 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28629 * doc/gcc.texi: Remove "up" link to (DIR).
28630 * doc/gccint.texi: Ditto.
28632 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28634 * doc/install.texi (Specific) <avr>: Remove reference to
28637 2017-03-13 Jeff Law <law@redhat.com>
28639 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28640 attribute rather than comments.
28642 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28643 match_scratch operand is highest.
28645 2017-03-13 Martin Liska <mliska@suse.cz>
28647 PR middle-end/78339
28648 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28649 is a CHKP clone, use original declaration.
28651 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28653 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28654 (arc_conditional_register_usage): Use a different allocation order
28655 when optimizing for size.
28656 * common/config/arc/arc-common.c (arc_option_optimization_table):
28657 Section anchors default on when optimizing for size.
28659 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28661 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28663 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28665 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28666 * config/arc/arc.md (cpu_facility): Add cd variant.
28667 (*movqi_insn): Add code density variant.
28668 (*movhi_insn): Likewise.
28669 (*movqi_insn): Likewise.
28670 (*addsi3_mixed): Likewise.
28671 (subsi3_insn): Likewise.
28673 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28675 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28677 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28679 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28680 expressions with MINUS and UNARY ops.
28682 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28685 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28687 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28688 between vec_select and vector argument.
28689 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28690 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28691 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28692 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28693 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28694 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28696 2017-03-13 Richard Biener <rguenther@suse.de>
28699 * params.def (vect-max-peeling-for-alignment): Fix typo.
28701 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28703 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28704 issue that only occurred with binutils below 2.18.
28706 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28708 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28709 refer to binutils 2.11/2.12 minimum.
28711 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28713 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28714 ftp.kernel.org and simplify binutils requirement.
28716 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28718 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28720 (Optimize Options): Ditto. Also remove redundancy.
28722 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28724 PR translation/79848
28725 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28727 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28728 to G_ to avoid double translation.
28730 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28732 PR translation/79923
28733 * auto-profile.c (get_combined_location): Convert leading
28734 character of diagnostics to lower case and remove trailing period.
28735 (read_profile): Likewise for various diagnostics.
28736 * config/arm/arm.c (arm_option_override): Remove trailing period
28737 from various diagnostics.
28738 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28739 (msp430_expand_delay_cycles): Likewise.
28741 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28744 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28745 full command-line argument, rather than just "str".
28746 (aarch64_validate_march): Likewise.
28747 (aarch64_validate_mtune): Likewise.
28749 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28751 PR rtl-optimization/78911
28752 * lra-assigns.c (must_not_spill_p): New function.
28753 (spill_for): Use it.
28755 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28757 PR tree-optimization/79981
28758 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28759 ATOMIC_COMPARE_EXCHANGE ifn result.
28760 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28761 IFN_ATOMIC_COMPARE_EXCHANGE.
28763 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28766 * opts.c (parse_sanitizer_options): Add missing question mark to
28767 "did you mean" message.
28769 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28771 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28773 (VMULEUH_UNS): Likewise.
28774 (VMULOUB_UNS): Likewise.
28775 (VMULOUH_UNS): Likewise.
28776 * config/rs6000/rs6000.c (builtin_function_type): Remove
28777 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28779 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28782 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28783 x with result of extra_parsing_for_operand_code_0.
28784 (function_reader::extra_parsing_for_operand_code_0): Convert
28785 return type from void to rtx, returning x. When reading
28786 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28787 larger size containing struct block_symbol.
28789 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28791 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28792 -mfloat128-hardware without -m64.
28794 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28797 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28798 entries to the case statement that marks unsigned arguments to
28799 overloaded functions.
28801 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28803 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28804 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28806 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28809 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28810 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28812 2017-03-10 Martin Liska <mliska@suse.cz>
28816 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28817 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28820 2017-03-10 Olivier Hainque <hainque@adacore.com>
28822 * tree-switch-conversion (array_value_type): Start by resetting
28823 candidate type to it's main variant.
28825 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28827 PR rtl-optimization/79909
28828 * combine.c (try_combine): Use simplify_replace_rtx on individual
28829 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28830 of the whole CALL_INSN_FUNCTION_USAGE.
28832 PR tree-optimization/79972
28833 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28834 get_range_info on SSA_NAMEs. Formatting fixes.
28836 2017-03-10 Richard Biener <rguenther@suse.de>
28837 Jakub Jelinek <jakub@redhat.com>
28839 PR tree-optimization/77975
28840 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28841 edge to be constant.
28842 (get_val_for): For constant x return it. Formatting fix.
28843 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28844 would use the same bases as the current one.
28846 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28848 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28849 instead of vec_select for V1TImode.
28850 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28852 (VSX_LE_128): Add V1TI to this mode iterator.
28853 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28854 (*vsx_le_perm_store_<mode>): Likewise.
28855 (pre-reload splitter for VSX stores): Likewise.
28856 (post-reload splitter for VSX stores): Likewise.
28857 (*vsx_xxpermdi2_le_<mode>): Likewise.
28858 (*vsx_lxvd2x2_le_<mode>): Likewise.
28859 (*vsx_stxvd2x2_le_<mode>): Likewise.
28861 2017-03-09 Michael Eager <eager@eagercon.com>
28863 Correct failures with --enable-checking=yes,rtl.
28865 * config/microblaze/microblaze.c (microblaze_expand_shift):
28866 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28867 test for const0_rtx.
28868 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28869 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28871 2017-03-09 Richard Biener <rguenther@suse.de>
28873 PR tree-optimization/79977
28874 * graphite-scop-detection.c (scop_detection::merge_sese):
28875 Handle the case of extra exits to blocks dominating the entry.
28877 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28879 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28882 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28884 PR rtl-optimization/79949
28885 * lra-constraints.c (process_alt_operands): Check memory when
28886 trying to predict a cycle. Print about the overall increase.
28888 2017-03-09 Richard Biener <rguenther@suse.de>
28890 PR middle-end/79971
28891 * gimple-expr.c (useless_type_conversion_p): Preserve
28892 TYPE_SATURATING for fixed-point types.
28894 2017-03-09 Richard Biener <rguenther@suse.de>
28897 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28898 alignment of BLKmode params.
28900 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28903 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28904 (VALL_NO_V2Q): Likewise.
28906 * config/aarch64/aarch64-simd.md
28907 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28909 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28910 VALL_NO_V2Q mode iterator.
28911 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28913 2017-03-09 Martin Liska <mliska@suse.cz>
28915 PR tree-optimization/79631
28916 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28917 tree_int_cst_sign_bit just for INTEGER constants.
28919 2017-03-09 Martin Liska <mliska@suse.cz>
28923 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28926 2017-03-09 Marek Polacek <polacek@redhat.com>
28929 * tree.c (inchash::add_expr): Handle TREE_VEC.
28931 2017-03-09 Martin Liska <mliska@suse.cz>
28934 (chkp_narrow_size_and_offset): New function.
28935 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28936 (void chkp_parse_bit_field_ref): New function.
28937 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28938 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28940 2017-03-09 Martin Liska <mliska@suse.cz>
28943 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28944 (chkp_find_bounds_1): Remove gcc_unreachable.
28946 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28949 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28950 BUILT_IN_SYNC*, determine the access type from the size suffix and
28951 always build a MEM_REF with that type. Handle forgotten
28952 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28955 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28956 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28957 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28958 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28959 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28960 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28961 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28962 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28963 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28964 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28965 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28966 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28967 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28968 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28969 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28970 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28971 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28972 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28973 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28974 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28975 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28976 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28977 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28978 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28979 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28980 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28981 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28982 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28983 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28984 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28985 definitions outside of __OPTIMIZE__ guarded section.
28988 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28989 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28990 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28991 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28994 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28996 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28997 ("vfenez<mode>"): Add missing constraints.
28999 2017-03-08 Martin Sebor <msebor@redhat.com>
29002 * config/nds32/nds32.c (nds32_option_override):
29003 Fix misspelled diagnostic.
29005 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29008 * gimplify.c (gimplify_omp_for): Replace index var in outer
29009 taskloop statement with an artificial variable and add
29010 OMP_CLAUSE_PRIVATE clause for it.
29012 2017-03-08 Richard Biener <rguenther@suse.de>
29014 PR tree-optimization/79955
29015 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29016 for accesses that are completely outside of the variable.
29018 2017-03-08 Andrew Haley <aph@redhat.com>
29020 PR tree-optimization/79943
29021 * tree-ssa-loop-split.c (compute_new_first_bound): When
29022 calculating the new upper bound, (END-BEG) should be added, not
29025 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29027 * config/avr/avr.md (setmemhi): Make sure match_dup
29028 operand number comes before match_scratch.
29030 2017-03-08 Richard Biener <rguenther@suse.de>
29032 PR tree-optimization/79920
29033 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29034 with ncopies == 1 to ...
29035 (vect_transform_slp_perm_load): ... here. Properly compute
29036 all element loads by iterating VF times over the group. Do
29037 not handle ncopies (computed in a broken way) in
29038 vect_create_mask_and_perm.
29040 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29043 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29044 is a uniform vector, use uniform_vector_p return value instead of
29045 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29047 2017-03-07 Marek Polacek <polacek@redhat.com>
29049 PR middle-end/79809
29050 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29051 (alloca_call_type): Likewise.
29053 2017-03-07 Martin Liska <mliska@suse.cz>
29055 * gcov.c (process_args): Put comment to correct location.
29057 2017-03-07 Martin Liska <mliska@suse.cz>
29059 PR middle-end/68270
29060 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29061 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29062 (chkp_narrow_bounds_for_field): Likewise.
29063 (chkp_parse_array_and_component_ref): Pass one more argument to
29066 2017-03-07 Richard Biener <rguenther@suse.de>
29068 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29071 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29073 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29074 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29076 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29079 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29080 to end of description.
29081 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29083 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29085 PR rtl-optimization/79901
29086 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29088 (*avx512f_<code><mode>3<mask_name>): ... this.
29089 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29090 iterator instead of VI8_AVX2_AVX512BW.
29092 PR rtl-optimization/79901
29093 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29094 min/max expander, expand it using expand_vec_cond_expr.
29097 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29100 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29103 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29104 to void * for PCH reasons.
29105 * dwarf2out.c (output_loc_operands, output_die): Cast
29106 v.val_vec.array to unsigned char *.
29108 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29111 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29114 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29116 PR rtl-optimization/79571
29117 * lra-constraints.c (process_alt_operands): Calculate static
29118 reject and subtract it from overall when only addresses will be
29121 2017-03-06 Julia Koval <julia.koval@intel.com>
29124 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29125 incoming stack boundary to 128 for 64-bit targets.
29127 2017-03-06 Richard Biener <rguenther@suse.de>
29129 PR tree-optimization/79894
29130 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29131 to NULL after folding it.
29133 2017-03-06 Richard Biener <rguenther@suse.de>
29135 PR tree-optimization/79824
29136 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29137 check disabling peeling for gaps.
29139 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29141 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29142 attributes): Document gettimeofday.
29144 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29146 * config/s390/s390.c (s390_option_override_internal): Set
29147 PARAM_MIN_VECT_LOOP_BOUND
29149 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29151 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29152 * config/s390/s390.md: Likewise.
29154 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29157 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29158 (<avx2_avx512>_perm<mode>): Rename to ...
29159 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29161 (<avx512>_perm<mode>_mask): Rename to ...
29162 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29164 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29165 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29166 instead of VI8F_256_512.
29167 (avx512f_perm<mode>): New define_expand.
29168 (avx512f_perm<mode>_mask): Likewise.
29169 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29170 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29172 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29174 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29175 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29177 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29179 2017-03-06 Martin Liska <mliska@suse.cz>
29182 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29183 when having a SSA NAME w/o VAR_DECL assigned to it.
29185 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29187 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29188 msa_dpsub_<su>_d): Fix MODE for vec_select.
29190 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29192 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29194 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29196 2017-03-06 Richard Biener <rguenther@suse.de>
29198 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29199 * plugin.c (register_plugin_info): Likewise.
29200 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29202 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29204 * config/i386/sse.md (sse_storehps, sse_storelps,
29205 avx_<castmode><avxsizesuffix>_<castmode>,
29206 avx512f_<castmode><avxsizesuffix>_<castmode>,
29207 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29208 in condition that at least one operand is not a MEM.
29210 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29212 PR middle-end/79805
29213 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29214 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29216 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29217 gimple_call_nothrow_p flag based on whether original builtin can throw.
29218 If it can, emit following stmts on the fallthrough edge.
29219 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29220 don't create new bb if inserting just debug stmts on the edge, try to
29221 insert them on the fallthru bb or just reset debug stmts.
29223 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29226 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29227 restore recog_data (including the operand rtxes inside it) around
29228 the call to get_insn_template.
29230 2017-03-03 Martin Sebor <msebor@redhat.com>
29232 PR tree-optimization/79699
29233 * context.c (context::~context): Free MPFR caches to avoid
29234 a memory leak on program exit.
29236 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29238 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29239 Use wide_int::ulow () instead of .elt (0).
29241 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29243 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29244 (*pushxf): Limit oF constraint to 32bit targets and add oC
29245 constraint for 64bit targets.
29246 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29247 (*pushdf): Change rmF constraint to rmC.
29249 2017-03-03 Martin Liska <mliska@suse.cz>
29251 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29252 Remove unused variable.
29254 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29257 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29258 is a memory operand, increase num_memory.
29259 (ix86_expand_args_builtin): Likewise.
29261 2017-03-03 Jan Hubicka <jh@suse.cz>
29264 * ipa-devirt.c (maybe_record_node): Properly handle
29265 __cxa_pure_virtual visibility.
29267 2017-03-03 Martin Liska <mliska@suse.cz>
29269 PR tree-optimization/79803
29270 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29272 (pass_loop_prefetch::execute): Disabled optimization if an
29273 assumption about L1 cache size is not met.
29275 2017-03-03 Martin Liska <mliska@suse.cz>
29277 PR rtl-optimization/79574
29278 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29279 (hash_scan_set): Likewise.
29280 (dump_hash_table): Likewise.
29281 (hoist_code): Likewise.
29283 2017-03-03 Richard Biener <rguenther@suse.de>
29285 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29286 in place of uhigh ().
29287 (fixed_convert_from_real): Likewise.
29289 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29292 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29294 2017-03-03 Richard Biener <rguenther@suse.de>
29296 PR middle-end/79818
29297 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29298 TYPE_OVERFLOW_UNDEFINED check.
29300 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29302 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29304 (vector_ae_<mode>_p): Likewise.
29305 (vector_nez_<mode>_p): Likewise.
29306 (vector_ne_v2di_p): Likewise.
29307 (vector_ae_v2di_p): Likewise.
29308 (vector_ne_<mode>_p): Likewise.
29309 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29311 (vsx_tsqrt<mode>2_fe): Likewise.
29313 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29316 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29318 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29320 PR rtl-optimization/79780
29321 * cprop.c (one_cprop_pass): When second and further conditional trap
29322 in a single basic block is turned into an unconditional trap, turn it
29323 into a deleted note to avoid RTL verification failures.
29325 2017-03-02 Richard Biener <rguenther@suse.de>
29327 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29329 2017-03-02 Richard Biener <rguenther@suse.de>
29331 PR tree-optimization/79345
29333 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29334 param and abort the walk, returning -1 if it is hit.
29335 (walk_aliased_vdefs): Take a limit param and pass it on.
29336 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29337 defaulting to 0 and return a signed int.
29338 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29339 (check_defs): New helper.
29340 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29341 about uninitialized memory.
29342 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29343 bogus uninitialized warning.
29344 (fixed_convert_from_real): Likewise.
29346 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29348 PR tree-optimization/66768
29349 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29350 iv_use if base object can't be determined.
29352 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29354 PR tree-optimization/79345
29355 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29356 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29357 (get_pattern_stats): Initialize it.
29358 * genemit.c (gen_expand): Verify match_scratch numbers come after
29359 match_operand/match_dup numbers.
29360 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29361 match_scratch numbers.
29362 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29364 * config/s390/s390.md (trunctdsd2): Likewise.
29366 2017-03-02 Richard Biener <rguenther@suse.de>
29368 * wide-int.h (wide_int_storage::operator=): Implement in terms
29371 2017-03-02 Richard Biener <rguenther@suse.de>
29373 PR tree-optimization/79777
29374 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29375 the to insert expression to sth existing.
29377 2017-03-01 Martin Sebor <msebor@redhat.com>
29379 PR middle-end/79692
29380 * gimple-ssa-sprintf.c
29381 (directive::known_width_and_precision): New function.
29382 (format_integer): Use it.
29383 (get_mpfr_format_length): Consider the full range of precision
29384 when computing %g output with the # flag. Set the likely byte
29385 count to 3 rather than 1 when precision is indeterminate.
29386 (format_floating): Correct the lower bound of precision.
29388 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29390 * doc/invoke.texi: Document default code model for 64-bit Linux.
29392 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29395 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29396 udiv rather than div since input pattern is unsigned.
29398 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29400 * config/i386/i386.c (print_reg): Warn for values of
29401 unsupported size in integer register.
29403 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29406 * config/rs6000/predicates.md (current_file_function_operand): Do
29407 not allow self calls to be local if the function is replaceable.
29409 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29412 * config/rs6000/altivec.h (vec_ctz and others): Change the
29413 preprocessor macro that controls conditional compilation from
29414 _ARCH_PWR9 to __POWER9_VECTOR__.
29415 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29416 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29417 control (instead of _ARCH_PWR9 control) so that template
29418 definition uses power9-specific function.
29419 (vec_any_eq): Likewise.
29420 (vec_all_ne): Change macro definition to use a power9-specific
29421 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29422 _ARCH_PWR9 control).
29423 (vec_any_eq) Likewise.
29424 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29425 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29426 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29427 support for xvcmpnedp instruction.
29428 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29429 macro expansion so that Power9 implementation of vec_all_ne does
29430 not use the AltiVec predicate framework.
29431 (VCMPNEH_P): Likewise.
29432 (VCMPNEW_P): Likewise.
29433 (VCMPNED_P): Likewise.
29434 (VCMPNEFP_P): Likewise.
29435 (VCMPNEDP_P): Likewise.
29436 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29437 implementation of vec_any_eq to not use AltiVec predicate
29439 (VCMPAEH_P): Likewise.
29440 (VCMPAEW_P): Likewise.
29441 (VCMPAED_P): Likewise.
29442 (VCMPAEFP_P): Likewise.
29443 (VCMPAEDP_P): Likewise.
29444 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29445 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29446 not use the AltiVec predicate framework.
29447 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29448 of vec_any_eq to not use AltiVec predicate framework.
29449 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29450 support for predefined __POWER9_VECTOR__ macro to indicate that
29451 Power9 instruction selection is enabled.
29452 (altivec_overloaded_builtins): Remove extraneous
29453 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29454 function argument types RS6000_BTI_bool_V16QI and
29455 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29456 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29457 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29458 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29459 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29460 Power9 for implementations of vec_cmpne. Change the signature for
29461 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29462 (representing vec_all_ne) to remove the previously described first
29463 argument of type RS6000_BTI_INTSI, as this was an artifact of
29464 reliance on the AltiVec predicate framework, which is no longer
29465 used in the implementation of these functions. Add
29466 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29467 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29468 since, unlike the AltiVec predicate framework implementation, we
29469 do not share function descriptors between vec_alle and vec_anyeq.
29470 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29471 set of modes that receive special treatment even when
29472 TARGET_P9_VECTOR is true. The special treatment emits code that
29473 does not depend on Power9 instructions.
29474 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29475 define_expand to not rely on AltiVec predicate framework.
29476 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29478 (vector_ne_v2di_p): Change this define_expand to not rely on
29479 AltiVec predicate framework.
29480 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29482 (vector_ne_<mode>_p): Change this define_expand to not rely on
29483 AltiVec predicate framework.
29484 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29486 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29487 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29488 define_insn pattern.
29489 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29490 define_insn pattern because the xvcmpne<VSs>. instruction is not
29492 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29493 instruction is not supported.
29495 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29497 * config/nvptx/nvptx.c: Include intl.h.
29499 2017-03-01 Martin Jambor <mjambor@suse.cz>
29502 * ipa-prop.h (ipa_bits): Removed field known.
29503 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29504 to pointers. Adjusted their comments to warn about their sharing.
29505 (ipcp_transformation_summary): Change bits to a vector of pointers.
29506 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29507 (ipa_get_ipa_bits_for_value): Declare.
29508 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29509 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29510 (ipa_bits_hash_table): Likewise.
29511 (ipa_vr_ggc_hash_traits): Likewise.
29512 (ipa_vr_hash_table): Likewise.
29513 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29514 being pointers and vr_known being removed.
29515 (ipa_set_jf_unknown): Likewise.
29516 (ipa_get_ipa_bits_for_value): New function.
29517 (ipa_set_jfunc_bits): Likewise.
29518 (ipa_get_value_range): New overloaded functions.
29519 (ipa_set_jfunc_vr): Likewise.
29520 (ipa_compute_jump_functions_for_edge): Use the above functions to
29521 construct bits and vr parts of jump functions.
29522 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29523 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29525 (ipcp_grow_transformations_if_necessary): Also allocate
29526 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29528 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29529 them. Fix too long lines.
29530 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29531 vr_known being removed.
29532 (ipa_read_jump_function): Use new setter functions to construct bits
29533 and vr parts of jump functions or set them to NULL.
29534 (write_ipcp_transformation_info): Adjust for bits being pointers.
29535 (read_ipcp_transformation_info): Likewise.
29536 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29538 Include gt-ipa-prop.h.
29539 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29541 (ipcp_store_bits_results): Likewise.
29542 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29543 Do not write to existing jump functions but use a temporary instead.
29545 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29548 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29549 attempt to use its first operand as BIT_FIELD_REF base.
29551 2017-03-01 Richard Biener <rguenther@suse.de>
29553 PR middle-end/79721
29554 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29555 interpolating formula in wrapping arithmetic.
29556 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29558 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29560 PR tree-optimization/79734
29561 * tree-vect-generic.c (expand_vector_condition): Optimize
29562 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29563 Handle VEC_COND_EXPR where comparison has different inner width from
29564 type's inner width.
29566 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29568 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29569 markup, and similar issues. Remove @opindex entries for things
29570 that aren't options. Add missing -mmpy-option entries.
29572 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29574 PR tree-optimization/79737
29575 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29576 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29577 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29578 instead of byte_size. Formatting fix.
29579 (shift_bytes_in_array_right): Formatting fix.
29581 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29584 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29585 condition on optimize for the leaf function test.
29587 2017-02-28 Martin Liska <mliska@suse.cz>
29590 * read-rtl-function.c (function_reader::handle_unknown_directive):
29591 Bail out when one uses -flto.
29593 2017-02-28 Martin Liska <mliska@suse.cz>
29595 * common.opt: Replace space with tabular for options of <number>
29597 * config/i386/i386.opt: Show <number> value for
29598 -mlarge-data-threshold.
29599 * opts.c (print_filtered_help): Do not display number in hexadecimal
29602 2017-02-28 Martin Liska <mliska@suse.cz>
29604 * common.opt: Fix --help=option -Q for options which are of
29607 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29609 * config/i386/i386.c (print_reg): Error out for values
29610 of 8-bit size in invalid integer register.
29612 2017-02-28 Martin Sebor <msebor@redhat.com>
29614 PR tree-optimization/79691
29615 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29617 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29620 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29621 gcc_unreachable with output_operand_lossage.
29623 2017-02-28 Richard Biener <rguenther@suse.de>
29625 PR tree-optimization/79740
29626 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29628 (visit_nary_op): Insert the nary into the hashtable if we
29629 pattern-matched sth.
29630 * tree-ssa-pre.c (eliminate_insert): Robustify.
29632 2017-02-28 Richard Biener <rguenther@suse.de>
29634 PR middle-end/79731
29635 * fold-const.c (decode_field_reference): Reject out-of-bound
29638 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29640 * config/i386/i386.c: Include intl.h.
29641 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29642 instead of just cond ? "..." : "...".
29643 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29644 * coverage.c (read_counts_file): Likewise.
29645 * omp-offload.c: Include intl.h.
29646 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29647 of just cond ? "..." : "...".
29648 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29649 of just cond ? "..." : "...".
29651 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29654 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29656 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29657 'tune for' CPU name.
29658 * config/arm/arm-cpu-data.h: Regenerated.
29660 2017-02-28 Richard Biener <rguenther@suse.de>
29662 PR tree-optimization/79732
29663 * tree-inline.c (expand_call_inline): Do not shadow var.
29665 2017-02-28 Richard Biener <rguenther@suse.de>
29667 PR tree-optimization/79723
29668 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29669 address-space properly.
29671 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29673 * doc/optinfo.texi (Optimization groups): Fix option used for
29675 * doc/invoke.texi (-fopt-info): Document "omp".
29676 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29677 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29678 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29679 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29680 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29682 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29684 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29685 OPTGROUP_OMP with "omp".
29687 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29690 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29691 for arithmetic shift of unsigned V2DI.
29693 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29695 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29696 arc/linux.h headers.
29697 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29698 (LINK_SPEC): Likewise.
29699 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29700 (EXTRA_SPECS): Likewise.
29701 (STARTFILE_SPEC): Likewise.
29702 (ENDFILE_SPEC): Likewise.
29703 (LIB_SPEC): Likewise.
29704 (TARGET_SDATA_DEFAULT): Likewise.
29705 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29706 (MULTILIB_DEFAULTS): Likewise.
29707 (DWARF2_UNWIND_INFO): Likewise.
29708 * config/arc/big.h: New file.
29709 * config/arc/elf.h: Likewise.
29710 * config/arc/linux.h: Likewise.
29711 * config/arc/t-uClibc: Remove.
29713 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29715 PR tree-optimization/77536
29716 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29717 (tree_transform_and_unroll_loop): Use above function to compute the
29718 estimated niter of unrolled loop and use it when scaling profile.
29719 Also use count info rather than frequency if it's non-zero.
29720 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29721 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29722 (vect_transform_loop): Call above function.
29724 2017-02-27 Richard Biener <rguenther@suse.de>
29726 PR tree-optimization/45397
29727 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29728 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29729 (visit_nary_op): Add pattern matching for CSEing sign-changed
29730 or truncated operations with wider ones.
29732 2017-02-27 Richard Biener <rguenther@suse.de>
29734 PR tree-optimization/79690
29735 * tree-vect-stmts.c (vectorizable_store): Use vector type
29736 built from the DR with address-space.
29738 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29740 * doc/invoke.texi (Optimize Options): Refine the description
29741 of asan-use-after-return.
29743 2017-02-25 Alan Modra <amodra@gmail.com>
29745 PR rtl-optimization/79584
29746 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29747 base, not ad->base_term, the reg within base. Remove assertion
29748 that ad->base == ad->base_term. Replace gen_int_mode using
29749 bogus mode with const0_rtx.
29751 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29753 PR middle-end/79396
29754 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29755 FMA_EXPR like tcc_binary or tcc_unary.
29757 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29760 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29762 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29763 (output_loc_operands): Handle DW_OP_call_ref and
29764 DW_OP_GNU_variable_value.
29765 (struct variable_value_struct): New type.
29766 (struct variable_value_hasher): Likewise.
29767 (variable_value_hash): New variable.
29768 (string_types): Remove.
29769 (copy_loc_descr): New function.
29770 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29771 (prepend_loc_descr_to_each): New function.
29772 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29773 instead of add_loc_descr_to_each if the first argument is single
29774 location list and the second has multiple.
29775 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29776 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29777 when looking for variable value which doesn't have other location info.
29778 (loc_list_from_tree): Formatting fix.
29779 (gen_array_type_die): Simplify DW_AT_string_length handling.
29780 (adjust_string_types): Remove.
29781 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29782 string_types. Call resolve_variable_values.
29783 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29784 (resolve_addr_in_expr): Likewise. Add A argument.
29785 (copy_deref_exprloc): Remove deref argument. Adjust for the
29786 original expression being DW_OP_GNU_variable_value with optionally
29787 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29788 optionally after it.
29789 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29790 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29791 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29792 (variable_value_hasher::hash, variable_value_hasher::equal): New
29794 (resolve_variable_value_in_expr, resolve_variable_value,
29795 resolve_variable_values, note_variable_value_in_expr,
29796 note_variable_value): New functions.
29797 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29800 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29803 * opts.h (handle_generated_option): Add GENERATED_P argument.
29804 * opts-common.c (handle_option): Adjust function comment.
29805 (handle_generated_option): Add GENERATED_P argument, pass it to
29807 (control_warning_option): Pass false to handle_generated_option
29809 * opts.c (maybe_default_option): Pass true to handle_generated_option
29811 * optc-gen.awk: Likewise.
29813 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29815 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29816 a REG, look at the REG it is a SUBREG of.
29817 (splitter for cmpeqsi_t): Ditto.
29819 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29821 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29822 the special USEs with the pattern of the insn, not the insn itself.
29824 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29827 * doc/invoke.texi: Document -mload-store-pairs.
29829 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29830 Sandra Loosemore <sandra@codesourcery.com>
29832 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29833 argument isn't a CONST_INT.
29834 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29835 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29836 (nios2_validate_compare): Bypass alternate compare logic if *op2
29837 is not a CONST_INT.
29838 (ldstwm_operation_p): Return false if first_base is not a REG or
29839 if first_offset is not a CONST_INT.
29841 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29843 * config/cris/cris.md: Use correct operand in a define_peephole2.
29845 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29847 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29849 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29851 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29852 this_insn if it is an INSN or JUMP_INSN.
29853 (force_offsettable): Look at base, not at addr.
29854 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29855 on things that aren't necessarily CONST_INTs.
29857 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29859 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29860 -mfpmath=sse is the default also for x86-32 targets with SSE2
29861 instruction set when @option{-ffast-math} is enabled
29863 2017-02-24 Jeff Law <law@redhat.com>
29865 PR rtl-optimizatoin/79286
29866 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29869 2017-02-24 Richard Biener <rguenther@suse.de>
29871 PR tree-optimization/79389
29872 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29875 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29877 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29878 function comment to reflect reality.
29879 (loop_exits_before_overflow): Fix typo in function description.
29881 2017-02-24 Richard Biener <rguenther@suse.de>
29883 PR tree-optimization/79389
29884 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29885 properly that a threading opportunity exists. Detect conditional
29886 copy/constant propagation opportunities.
29888 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29890 * config/visium/visium.md (type): Add trap.
29891 (b): New mode attribute.
29892 (*btst): Rename into...
29893 (*btst<mode>): ...this and adjust.
29894 (*cbranchsi4_btst_insn): Rename into...
29895 (*cbranch<mode>4_btst_insn): ...this and adjust.
29896 (trap): New define_insn.
29898 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29900 PR tree-optimization/79389
29901 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29902 (noce_reversed_cond_code): New function.
29903 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29904 reversed_comparison_code. Formatting fix.
29905 (noce_try_store_flag): Test rev_cond != NULL in addition to
29906 reversed_comparison_code.
29907 (noce_try_store_flag_constants): Likewise.
29908 (noce_try_store_flag_mask): Likewise.
29909 (noce_try_addcc): Use rev_cond if non-NULL instead of
29910 reversed_comparison_code.
29911 (noce_try_cmove_arith): Likewise. Formatting fixes.
29912 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29913 (noce_find_if_block): Initialize rev_cond.
29914 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29915 instead of false as last argument never attempt to reverse it
29918 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29920 PR tree-optimization/79663
29921 * tree-predcom.c (combine_chains): Process refs in reverse order
29922 only for ZERO length chains, and add explaining comment.
29924 2017-02-23 Jeff Law <law@redhat.com>
29926 PR tree-optimization/79578
29927 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29928 in call to operand_equal_p.
29930 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29933 * config/i386/cpuid.h: Fix another undefined behavior.
29935 2017-02-23 Richard Biener <rguenther@suse.de>
29937 PR tree-optimization/79683
29938 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29939 vector types for data-refs.
29941 2017-02-23 Martin Liska <mliska@suse.cz>
29943 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29945 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29947 PR middle-end/79665
29948 * internal-fn.c (get_range_pos_neg): Moved to ...
29949 * tree.c (get_range_pos_neg): ... here. No longer static.
29950 * tree.h (get_range_pos_neg): New prototype.
29951 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29952 are known to be in between 0 and signed maximum inclusive, try to
29953 expand both unsigned and signed divmod and use the cheaper one from
29956 2017-02-22 Jeff Law <law@redhat.com>
29958 PR tree-optimization/79578
29959 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29960 to compare base operands.
29962 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29965 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29966 gpc_reg_operand instead of fpr_reg_operand.
29968 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29970 * config/mips/mips.c (mips_return_in_memory): Force FP
29971 vector types to be returned in memory for o32 ABI.
29973 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29975 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29976 instead of DW_TAG_member for static data member declarations and don't
29977 set no_linkage_name for static inline data members.
29978 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29981 2017-02-22 Martin Liska <mliska@suse.cz>
29983 * doc/invoke.texi: Replace inequality signs with square brackets
29986 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29989 * lra-constraints.c (simplify_operand_subreg): Handle
29990 WORD_REGISTER_OPERATIONS targets.
29992 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29995 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29996 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29997 elimination by swapping fld*.
29999 2017-02-22 Richard Biener <rguenther@suse.de>
30001 PR tree-optimization/79673
30002 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30003 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30004 irrelevant address-space qualifiers and avoiding a
30005 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30007 2017-02-22 Richard Biener <rguenther@suse.de>
30009 PR tree-optimization/79666
30010 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30011 to not symbolically negate if that may introduce undefined
30014 2017-02-22 Martin Liska <mliska@suse.cz>
30017 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30018 * data-streamer-out.c (streamer_write_gcov_count_stream):
30020 * value-prof.c (stream_out_histogram_value): Make assert more
30021 precise based on type of counter.
30023 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30026 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30027 Use nonimmediate_operand instead of memory_operand for operand 1.
30028 (float-extend standard_x87sse_constant_load splitter): Ditto.
30030 2017-02-21 Jeff Law <law@redhat.com>
30032 PR tree-optimization/79621
30033 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30034 blocks with edges to themselves.
30036 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30039 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30040 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30041 Use gimple_call_builtin_p.
30044 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30045 on temporarily removed DEBUG_INSNs.
30047 PR tree-optimization/79649
30048 * tree-loop-distribution.c (classify_partition): Give up on
30049 non-generic address space loads/stores.
30051 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30053 * doc/loop.texi (Loop manipulation): Remove nonexistent
30054 tree_ssa_loop_version from the documentation.
30055 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30057 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30060 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30061 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30062 * config/rs6000/rs6000.c: Include except.h.
30063 (rs6000_expand_split_stack_prologue): Call
30064 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30066 2017-02-21 Martin Jambor <mjambor@suse.cz>
30069 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30070 have been analyzed.
30072 2017-02-21 Martin Jambor <mjambor@suse.cz>
30074 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30075 for backward compatibility only.
30076 * doc/invoke.texi (Option Summary): Remove all references to
30077 -fipa-cp-alignment.
30079 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30083 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30085 * lra-constraints.c (curr_insn_transform): Handle
30086 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30088 2017-02-21 Martin Liska <mliska@suse.cz>
30090 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30092 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30094 PR translation/79638
30095 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30097 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30100 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30101 (arm_function_ok_for_sibcall): Return false for an indirect call by
30102 descriptor if all the argument registers are used.
30103 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30104 alignment of the function.
30106 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30108 PR tree-optimization/61441
30109 * simplify-rtx.c (simplify_const_unary_operation): For
30110 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30111 the sNaN unmodified.
30113 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30115 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30116 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30117 instead of SYSTEM_HEADER_DIR.
30119 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30120 Martin Liška <mliska@suse.cz>
30122 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30123 Fix typos and grammar, use active voice, and clarify.
30125 2017-02-20 Marek Polacek <polacek@redhat.com>
30127 PR middle-end/79537
30128 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30131 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30133 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30136 * config/i386/i386.c (ix86_expand_builtin): Handle
30137 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30138 ix86_builtins_isa[fcode].isa as a requirement of those
30139 flags and any other flag in the bitmask.
30140 (ix86_init_mmx_sse_builtins): Use 0 instead of
30141 ~OPTION_MASK_ISA_64BIT as mask.
30142 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30143 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30144 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30145 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30147 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30150 * lra-constraints.c (split_reg): Check requested split mode
30151 is supported by the register.
30153 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30155 * lra-constraints.c (simplify_operand_subreg): Remove early
30158 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30161 * lra-constraints.c (curr_insn_transform): Tighten condition
30162 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30164 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30167 * lra-constraints.c (curr_insn_transform): Handle
30168 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30170 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30173 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30175 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30177 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30180 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30183 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30185 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30188 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30190 * final.c (last_columnnum, override_columnnum): New variables.
30191 (final_start_function): Set last_columnnum, pass it to begin_prologue
30192 hook and pass 0 to dwarf2out_begin_prologue.
30193 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30194 to source_line debug hook.
30195 (notice_source_line): Compute last_columnnum and for debug_column_info
30196 return true on column changes.
30197 * debug.h (struct gcc_debug_hooks): Add column argument to
30198 source_line and begin_prologue hooks.
30199 (debug_nothing_int_charstar_int_bool): Remove prototype.
30200 (debug_nothing_int_int_charstar,
30201 debug_nothing_int_int_charstar_int_bool): New prototypes.
30202 (dwarf2out_begin_prologue): Add column argument.
30203 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30204 begin_prologue hooks.
30205 (debug_nothing_int_charstar_int_bool): Remove.
30206 (debug_nothing_int_int_charstar,
30207 debug_nothing_int_int_charstar_int_bool): New functions.
30208 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30209 through to dwarf2out_source_line.
30210 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30211 (dwarf2out_source_line): Add column argument, emit it if requested.
30212 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30214 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30215 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30216 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30217 through to dwarf2out_begin_prologue.
30218 (vmsdbgout_source_line): Add column argument, pass it through to
30219 dwarf2out_source_line.
30220 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30221 dbxout_source_line caller.
30222 (dbxout_source_line): Add column argument.
30224 * common.opt (gno-column-info, gcolumn-info): New options.
30225 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30226 (check_die): Also test for multiple DW_AT_decl_column attributes.
30227 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30228 DW_AT_decl_column if requested.
30229 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30231 (gen_variable_die): Likewise.
30232 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30233 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30236 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30237 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30238 (ix86_handle_option): Handle OPT_m3dnowa.
30239 * doc/invoke.texi (-m3dnowa): Document.
30240 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30241 -m3dnowa instead of -m3dnow -march=athlon.
30244 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30245 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30247 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30250 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30251 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30252 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30253 generator for vsx_xxpermdi_<mode>_be.
30254 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30255 force big-endian semantics.
30256 (vsx_xxpermdi_<mode>_be): New define_expand with same
30257 implementation as previous version of vsx_xxpermdi_<mode>.
30259 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30261 PR tree-optimization/79327
30262 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30263 variable, its initialization and use.
30265 2017-02-17 Julia Koval <julia.koval@intel.com>
30267 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30268 (OPTION_MASK_ISA_PKU_UNSET): New.
30269 (ix86_handle_option): Handle -mrdpid.
30270 * config/i386/cpuid.h (bit_RDPID): New.
30271 * config/i386/driver-i386.c (host_detect_local_cpu):
30272 Detect RDPID feature.
30273 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30274 * config/i386/i386-c.c (ix86_target_macros_internal):
30276 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30277 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30278 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30279 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30280 * config/i386/i386.md (define_insn "rdpid"): New.
30281 * config/i386/i386.opt Add -mrdpid.
30282 * config/i386/immintrin.h (_rdpid_u32): New.
30284 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30286 PR rtl-optimization/79541
30287 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30288 instead of transforming it into USE.
30290 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30292 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30293 If HONOR_SNANS (SFmode) force the input to a register.
30294 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30295 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30296 an frsp or similar insn.
30298 2017-02-17 Martin Liska <mliska@suse.cz>
30300 PR rtl-optimization/79577
30301 * params.def (selsched-max-sched-times): Increase minimum to 1.
30303 2017-02-17 Martin Liska <mliska@suse.cz>
30305 PR rtl-optimization/79574
30306 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30308 2017-02-17 Martin Liska <mliska@suse.cz>
30310 PR tree-optimization/79529
30311 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30312 ssa_defined_default_def_p to handle cases which are implicitly
30314 * tree-ssa.c (ssa_defined_default_def_p): New function.
30315 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30316 which are implicitly defined.
30317 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30319 2017-02-17 Richard Biener <rguenther@suse.de>
30321 PR middle-end/79576
30322 * params.def (max-ssa-name-query-depth): Limit to 10.
30324 2017-02-17 Richard Biener <rguenther@suse.de>
30326 PR tree-optimization/79552
30327 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30330 2017-02-17 Richard Biener <rguenther@suse.de>
30333 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30335 2017-02-17 Marek Polacek <polacek@redhat.com>
30337 PR middle-end/79536
30338 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30339 (fold_negate_expr): New wrapper.
30341 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30343 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30344 Correct terminology and de-emphasize pre-standard behavior.
30346 2017-02-16 Alan Modra <amodra@gmail.com>
30348 PR rtl-optimization/79286
30349 * ira.c (def_dominates_uses): New function.
30350 (update_equiv_regs): Don't create an equivalence for insns that
30351 may trap where the register def does not dominate the use.
30353 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30355 PR rtl-optimization/78127
30356 * lra.c (lra): Call lra_eliminate before finish the loop after
30359 2017-02-16 Richard Biener <rguenther@suse.de>
30361 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30363 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30364 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30365 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30366 (isl_val_int_from_wi): New function.
30367 (extract_affine_gmp): Rename to ...
30368 (extract_affine_wi): ... this, take a widest_int.
30369 (extract_affine_int): Just wrap extract_affine_wi.
30370 (add_param_constraints): Use isl_val_int_from_wi.
30371 (add_loop_constraints): Likewise, and extract_affine_wi.
30373 2017-02-15 Jeff Law <law@redhat.com>
30375 PR middle-end/79521
30376 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30377 ira_init_register_move_cost_if_necessary.
30379 2017-02-15 Martin Sebor <msebor@redhat.com>
30381 PR middle-end/32003
30382 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30383 removed in a prior commit.
30385 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30387 PR tree-optimization/79347
30388 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30389 counters during peeling.
30391 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30393 * Makefile.in (site.exp): Remove "set ISLVER".
30395 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30398 * real.c (real_from_integer): Call real_convert even for decimal.
30400 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30403 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30405 2017-02-14 Andrew Pinski <apinski@cavium.com>
30407 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30408 cores and change the partno/implementer to be correct.
30409 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30410 the 'B" as the implementer.
30411 * config/aarch64/aarch64-tune.md: Regenerate.
30413 2017-02-14 Carl Love <cel@us.ibm.com>
30415 * config/rs6000/rs6000.c: Add case statement entry to make the
30416 xvcvuxdsp built-in argument unsigned.
30417 * config/rs6000/vsx.md: Fix the source and return operand types so they
30418 match the instruction definitions from the ISA document. Fix typo
30419 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30422 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30425 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30426 member early_clobber_alts.
30427 * lra-lives.c (reg_early_clobber_p): New.
30428 (process_bb_lives): Use it.
30429 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30430 (debug_operand_data): Initialize early_clobber_alts.
30431 (setup_operand_alternative): Set up early_clobber_alts.
30432 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30433 alternatives to new_insn_reg.
30434 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30436 (lra_update_insn_regno_info): Pass the new arg.
30438 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30440 PR middle-end/79505
30441 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30442 (new_oacc_loop_raw): Don't clear already cleared fields.
30445 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30446 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30447 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30449 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30452 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30454 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30457 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30458 the extra instruction to the right place to store 128-bit constant
30461 2017-02-14 Martin Sebor <msebor@redhat.com>
30463 PR middle-end/79448
30464 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30465 warning for strings of unknown length.
30467 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30469 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30471 2017-02-14 Jeff Law <law@redhat.com>
30474 * ira-costs.c (scan_one_insn): Initialize register move costs
30475 for pseudos seen in USE/CLOBBER insns.
30477 PR tree-optimization/79095
30478 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30479 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30480 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30481 if the operands are known to be not equal, then the resulting range
30483 (intersect_ranges): If the new range is ~[0,0] and the old range is
30484 wide, then prefer ~[0,0].
30485 * tree-vrp.c (overflow_comparison_p_1): New function.
30486 (overflow_comparison_p): New function.
30487 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30488 if NAME is used in an overflow test.
30489 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30490 overflow check that can be expressed as an equality test, then adjust
30491 ops to be that equality test.
30493 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30495 * config/s390/s390-builtin-types.def: Remove flags argument.
30496 * config/s390/s390.c (s390_init_builtins): Likewise.
30498 2017-02-14 Martin Liska <mliska@suse.cz>
30500 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30501 vector. Fix trailing white spaces.
30503 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30505 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30508 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30510 PR rtl-optimization/68664
30511 * config/arm/arm.c (arm_sched_can_speculate_insn):
30512 New function. Declare prototype.
30513 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30515 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30517 PR rtl-optimization/68664
30518 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30520 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30522 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30524 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30525 max skip bytes for function, loop and jump.
30527 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30529 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30530 ABS_EXPR for gimple dump.
30532 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30535 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30537 PR tree-optimization/79408
30538 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30539 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30540 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30541 also if rhs1 is INTEGER_CST.
30543 2017-02-14 Richard Biener <rguenther@suse.de>
30545 PR middle-end/79432
30546 * tree-into-ssa.c (insert_phi_nodes): When the function can
30547 have abnormal edges rewrite SSA names with broken use-def
30548 dominance out of SSA and register them for PHI insertion.
30550 2017-02-13 Martin Sebor <msebor@redhat.com>
30552 PR middle-end/79496
30553 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30554 clearing info.nowrite flag when snprintf size argument is a range.
30556 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30558 * cprop.c (cprop_jump): Add missing space in string literal.
30559 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30560 (get_constraint_for_component_ref): Likewise.
30561 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30562 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30563 * lra-constraints.c (process_alt_operands): Likewise.
30564 * ipa-inline.c (inline_small_functions): Likewise.
30565 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30566 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30567 * trans-mem.c (diagnose_tm_1_op): Likewise.
30568 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30569 (grid_parallel_clauses_gridifiable): Likewise.
30571 * config/nvptx/mkoffload.c (process): Add space in between
30574 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30575 "MOD4_SSE_REGS" and "ALL_REGS".
30577 * spellcheck.c (test_data): Add , in between "foo" and "food".
30579 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30582 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30583 boundary crossing check and subsequent code generation agree.
30585 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30587 * config/aarch64/aarch64.c (has_memory_op): Delete.
30588 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30591 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30593 PR rtl-optimization/79388
30594 PR rtl-optimization/79450
30595 * combine.c (distribute_notes): When removing TEM_INSN for which
30596 corresponding dest has last value recorded, invalidate that last
30599 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30601 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30602 of explicit '@'. Add missing assembly comment marker on branch costs
30605 2017-02-13 Nathan Sidwell <nathan@acm.org>
30607 * gengtype-lex.l (<in_struct>): Add '/'.
30609 2017-02-13 Martin Liska <mliska@suse.cz>
30612 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30614 2017-02-13 Richard Biener <rguenther@suse.de>
30616 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30618 * configure: Re-generate.
30619 * config.in: Likewise.
30620 * graphite-dependences.c: Simplify as if
30621 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30622 * graphite-isl-ast-to-gimple.c: Likewise.
30623 * graphite-optimize-isl.c: Likewise.
30624 * graphite-poly.c: Likewise.
30625 * graphite-sese-to-poly.c: Likewise.
30626 * graphite.h: Likewise.
30627 * toplev.c: Include isl/version.h and use isl_version () for
30628 printing the ISL version.
30629 * doc/install.texi: Update ISL requirement.
30631 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30633 * doc/standards.texi (Standards): Update reference to
30636 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30638 * doc/extend.texi (Named Address Spaces): sourceware.org now
30640 * doc/install.texi (Binaries): Ditto.
30643 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30645 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30646 standard terminology "stringize"/"stringizing" throughout.
30647 * doc/cppinternals.texi: Likewise.
30649 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30651 * doc/extend.texi: Fix some spelling mistakes and typos.
30652 * doc/invoke.texi: Likewise.
30654 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30657 * params.def (inline-min-speedup) Change from 10 to 8.
30659 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30661 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30664 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30667 * ipa-inline-analysis.c (get_minimal_bb): New function.
30668 (record_modified): Use it.
30669 (remap_edge_change_prob): Handle also ancestor functions.
30671 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30673 * doc/contrib.texi (Contributors): Remove broken link into
30674 the Mauve CVS repository.
30676 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30678 PR middle-end/79454
30679 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30680 result computation whenever lhs doesn't have vector mode, not
30681 just when it has BLKmode.
30683 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30685 * doc/makefile.texi (profiledbootstrap): Refer to the
30686 installation instructions only in textual form.
30688 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30691 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30693 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30695 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30696 (Specific): Update mingw-w64 reference.
30698 (Specific): Remove broken link to Renesas RX processor.
30700 2017-02-10 Richard Biener <rguenther@suse.de>
30702 * toplev.c (process_options): Do not mention obsolete graphite
30703 options when printing sorry message about missing graphite support.
30704 Mention -floop-nest-optimize.
30706 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30708 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30709 (vtst_p16): Likewise.
30710 (vtstq_p8): Likewise.
30711 (vtstq_p16): Likewise.
30713 (vtstq_p64): Likewise.
30714 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30715 (vset_lane_p64): New.
30716 (vsetq_lane_p64): New.
30718 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30720 PR tree-optimization/79411
30721 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30722 stmt operands are SSA_NAMEs used in abnormal phis.
30723 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30726 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30729 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30732 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30734 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30736 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30738 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30742 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30743 not arbitrary TREE_CONSTANT.
30746 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30747 "omp declare target link" attribute unless is_global_var.
30748 * omp-offload.c (find_link_var_op): Likewise.
30750 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30751 Chung-Lin Tang <cltang@codesourcery.com>
30753 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30755 (gimplify_adjust_omp_clauses): Don't delete TILE.
30756 (gimplify_omp_for): Deal with TILE.
30757 * internal-fn.c (expand_GOACC_TILE): New function.
30758 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30760 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30761 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30763 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30764 avoid DIV for outermost collapse var.
30765 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30766 Remove out of date comments, fix whitespace.
30767 * omp-general.c (omp_extract_for_data): Deal with tiling.
30768 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30769 adjust OLF_DIM_BASE value.
30770 (struct omp_for_data): Add tiling field.
30771 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30772 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30773 for auto loops. Remove default auto determining, moved to
30774 oacc_loop_fixed_partitions.
30775 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30776 stmts, add e_mask field.
30777 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30778 (oacc_thread_numbers): Use oacc_dim_call.
30779 (oacc_xform_tile): New.
30780 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30781 (finish_oacc_loop): Adjust for ifns vector.
30782 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30783 add case for GOACC_TILE fns.
30784 (oacc_loop_xform_loop): Delete.
30785 (oacc_loop_process): Iterate over call list directly, and add
30786 handling for GOACC_TILE fns.
30787 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30789 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30790 vector partitioning to outer loops. Assign 2 partitions to loops
30791 when available. Add TILE handling.
30792 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30793 (execite_oacc_device_lower): Process GOACC_TILE fns,
30794 ignore unknown specs.
30795 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30796 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30797 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30799 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30801 * configure.ac (ACX_BUGURL): Update.
30802 * configure: Regenerate.
30804 2017-02-09 Richard Biener <rguenther@suse.de>
30806 PR tree-optimization/69823
30807 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30808 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30810 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30812 * config/arc/arc-c.def: Add __NPS400__ definition.
30813 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30814 (TARGET_NPS400): Define.
30816 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30818 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30820 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30821 pointer, arch_info.
30822 (arc_cpu_types): Fill the arch_info field with a pointer into the
30823 arc_arch_types table.
30824 (arc_selected_cpu): Declare.
30825 * config/arc/arc.c (arc_selected_cpu): Make global.
30826 (arc_selected_arch): Delete.
30827 (arc_base_cpu): Delete.
30828 (arc_override_options): Remove references to deleted variables,
30829 update access to arch information.
30830 (ARC_OPT): Update access to arch information.
30831 (ARC_OPTX): Likewise.
30832 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30833 (TARGET_ARC600): Update access to arch information.
30834 (TARGET_ARC601): Likewise.
30835 (TARGET_ARC700): Likewise.
30836 (TARGET_EM): Likewise.
30837 (TARGET_HS): Likewise.
30838 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30841 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30844 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30845 condition/operands for integer GE/LE/GEU/LEU operations.
30847 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30849 PR translation/79397
30850 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30853 2017-02-08 Martin Jambor <mjambor@suse.cz>
30856 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30857 whether allocation happened.
30858 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30859 nothing was allocated.
30861 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30863 PR tree-optimization/79408
30864 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30865 constant, but SSA_NAME with a known integer range, use the minimum
30866 of that range instead of op1 to determine if modulo can be replaced
30867 with its first operand.
30869 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30871 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30873 2017-02-08 Richard Biener <rguenther@suse.de>
30875 PR tree-optimization/71824
30876 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30877 Check all loops contained in the merged region.
30879 2017-02-07 Andrew Pinski <apinski@cavium.com>
30881 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30883 2017-02-07 Andrew Pinski <apinski@cavium.com>
30885 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30886 (thunderxt88): Likewise.
30887 (thunderxt81): Disable LSE and change v8.1 to v8.
30888 (thunderxt83): Likewise.
30890 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30891 Richard Biener <rguenther@suse.de>
30893 PR middle-end/79399
30894 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30895 type from int to size_t.
30896 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30898 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30900 PR rtl-optimization/79386
30901 * cprop.c (bypass_conditional_jumps): Initialize
30902 bypass_last_basic_block already before splitting bbs after
30903 unconditional traps...
30904 (bypass_conditional_jumps): ... rather than here.
30907 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30908 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30909 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30910 fix -masm=intel patterns.
30912 2017-02-07 Richard Biener <rguenther@suse.de>
30914 PR tree-optimization/79256
30915 PR middle-end/79278
30916 * builtins.c (get_object_alignment_2): Use min_align_of_type
30917 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30918 and ADJUST_FIELD_ALIGN.
30920 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30922 * doc/tm.texi: Regenerate.
30923 * stor-layout.c (layout_decl): Adjust.
30924 (update_alignment_for_field): Likewise.
30925 (place_field): Likewise.
30926 (min_align_of_type): Likewise.
30927 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30928 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30929 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30930 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30931 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30932 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30933 * config/i386/i386.c (x86_field_alignment): Likewise.
30934 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30935 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30936 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30937 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30938 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30939 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30943 2017-01-30 Richard Biener <rguenther@suse.de>
30945 PR tree-optimization/79256
30946 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30947 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30950 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30952 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30953 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30954 builtins to SImode and emit a zero-extend, if necessary.
30956 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30958 * docs/invoke.texi (RISC-V Options): Alphabetize.
30960 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30962 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30965 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30967 * config/riscv/riscv.c: New file.
30968 * common/config/riscv/riscv-common.c: Likewise.
30969 * config.gcc: Likewise.
30970 * config/riscv/constraints.md: Likewise.
30971 * config/riscv/elf.h: Likewise.
30972 * config/riscv/generic.md: Likewise.
30973 * config/riscv/linux.h: Likewise.
30974 * config/riscv/multilib-generator: Likewise.
30975 * config/riscv/peephole.md: Likewise.
30976 * config/riscv/pic.md: Likewise.
30977 * config/riscv/predicates.md: Likewise.
30978 * config/riscv/riscv-builtins.c: Likewise.
30979 * config/riscv/riscv-c.c: Likewise.
30980 * config/riscv/riscv-ftypes.def: Likewise.
30981 * config/riscv/riscv-modes.def: Likewise.
30982 * config/riscv/riscv-opts.h: Likewise.
30983 * config/riscv/riscv-protos.h: Likewise.
30984 * config/riscv/riscv.h: Likewise.
30985 * config/riscv/riscv.md: Likewise.
30986 * config/riscv/riscv.opt: Likewise.
30987 * config/riscv/sync.md: Likewise.
30988 * config/riscv/t-elf-multilib: Likewise.
30989 * config/riscv/t-linux: Likewise.
30990 * config/riscv/t-linux-multilib: Likewise.
30991 * config/riscv/t-riscv: Likewise.
30992 * configure.ac: Likewise.
30993 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30994 Waterman as RISC-V maintainers.
30995 * doc/install.texi: Add RISC-V entries.
30996 * doc/invoke.texi: Add RISC-V options section.
30997 * doc/md.texi: Add RISC-V constraints section.
30998 * configure: Regenerated.
31000 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31003 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31004 false values to be constant vectors with all 0 or all 1 bits set.
31005 (vcondu<mode><mode>): Likewise.
31006 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31008 (fpmask_comparison_operator): Update comment.
31009 (vecint_comparison_operator): New predicate.
31010 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31011 vector conditionals when the true and false values are constant
31012 vectors with all 0 bits or all 1 bits set.
31014 2017-02-06 Martin Sebor <msebor@redhat.com>
31016 PR tree-optimization/79376
31017 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31019 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31021 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31022 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31023 to simplify split condition.
31025 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31027 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31028 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31031 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31033 PR rtl-optimization/68664
31034 * target.def (can_speculate_insn): New hook.
31035 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31036 * doc/tm.texi: Regenerate.
31037 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31038 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31039 (rs6000_sched_can_speculate_insn): New function.
31041 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31043 PR tree-optimization/79284
31044 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31045 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31046 vectorizable_mask_load_store, vectorizable_operation,
31047 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31048 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31049 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31050 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31051 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31052 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31053 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31054 is_gimple_assign (stmt). Replace another such test with
31055 is_gimple_assign (stmt).
31057 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31060 * config/avr/avr.c (rtl-iter.h): Include it.
31061 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31062 (avr_legitimate_combined_insn): ...and implementation.
31064 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31066 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31067 * config/s390/s390.c (s390_const_operand_ok)
31068 (s390_canonicalize_comparison, s390_extract_part)
31069 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31070 (s390_contiguous_bitmask_p, s390_rtx_costs)
31071 (legitimize_pic_address): Likewise.
31072 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31073 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31074 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31075 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31076 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31078 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31080 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31081 REGNO($0) == REGNO($1).
31083 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31085 * config/s390/linux.h(SIZE_TYPE): Add comment.
31087 2017-02-06 Julian Brown <julian@codesourcery.com>
31088 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31089 Virendra Pathak <virendra.pathak@broadcom.com>
31091 * config/aarch64/aarch64-cores.def: Change the scheduler
31093 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31094 * config/aarch64/thunderx2t99.md: New file.
31096 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31098 * doc/standards.texi (Go Language): Update link to language
31101 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31103 * tree-eh.c (lower_resx): Sanitize profile.
31104 (cleanup_empty_eh_move_lp): Likewise.
31106 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31109 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31111 * cfgloopmanip.h (loop_version): Update prototype.
31112 * modulo-sched.c (sms_schedule): Update call of loop_version.
31113 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31114 * tree-parloops.c (gen_parallel_loop): Likewise.
31115 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31116 * tree-ssa-loop-split.c (split_loop): Likewise.
31117 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31118 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31120 2017-02-05 Martin Liska <mliska@suse.cz>
31123 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31125 (print_operand_address): Initialize a struct to zero.
31127 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31129 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31130 garbage collector only in textual form.
31132 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31134 * doc/extend.texi (x86 specific memory model extensions for
31135 transactional memory): Simplify a phrase.
31137 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31140 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31141 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31142 (atomic_storedi_1): Likewise.
31144 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31146 PR tree-optimization/79338
31147 * tree-parloops.c (gather_scalar_reductions): Don't call
31148 vect_analyze_loop_form for loop->inner before destroying loop's
31151 2017-02-03 Martin Sebor <msebor@redhat.com>
31153 PR tree-optimization/79327
31154 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31155 when precision has resulted in leading zeros.
31156 (format_integer): Adjust the likely counter to assume an unknown
31157 argument that may be zero is non-zero.
31159 2017-02-03 Jason Merrill <jason@redhat.com>
31162 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31163 avoid copying non-taken branch.
31165 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31167 PR tree-optimization/79340
31168 * tree-vect-loop.c (vectorizable_reduction): Release
31169 vec_defs elements after safe_splicing them into other vectors.
31172 PR tree-optimization/79327
31173 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31174 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31176 (format_integer): Use wide_int_to_tree instead of build_int_cst
31177 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31178 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31179 of shortest and longest sequence.
31181 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31183 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31184 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31186 2017-02-03 Walter Lee <walt@tilera.com>
31189 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31190 after initial stackframe link reg save.
31191 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31193 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31196 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31197 wu for stxssp alternative.
31199 2017-02-03 Martin Sebor <msebor@redhat.com>
31201 PR tree-optimization/79352
31202 * gimple-fold.c (get_range_strlen): Add argument.
31203 (get_range_strlen): Change return type to bool.
31204 (get_maxval_strlen): Pass in a dummy argument.
31205 * gimple-fold.h (get_range_strlen): Change return type to bool.
31206 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31207 * tree.h (array_at_struct_end_p): Add argument.
31208 * tree.c (array_at_struct_end_p): Handle it.
31210 2017-02-03 Martin Liska <mliska@suse.cz>
31213 * multiple_target.c (create_dispatcher_calls): Redirect edge
31214 from a caller of a dispatcher.
31215 (expand_target_clones): Make the clones local.
31216 (ipa_target_clone): Do both target clones and resolvers.
31217 (ipa_dispatcher_calls): Remove the pass.
31218 (pass_dispatcher_calls::gate): Likewise.
31219 (make_pass_dispatcher_calls): Likewise.
31220 * passes.def (pass_target_clone): Put as very first IPA early
31223 2017-02-03 Martin Liska <mliska@suse.cz>
31225 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31226 in case of a function with ifunc attribute.
31228 2017-02-03 Martin Liska <mliska@suse.cz>
31230 * cgraph.c (cgraph_node::dump): Dump function version info.
31231 * symtab.c (symtab_node::dump_base): Add missing new line.
31233 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31235 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31236 (ifcombine_ifandif): Use it.
31238 2017-02-03 Martin Liska <mliska@suse.cz>
31240 * doc/invoke.texi: Document default value for
31241 use-after-scope-direct-emission-threshold.
31243 2017-02-03 Martin Liska <mliska@suse.cz>
31245 PR tree-optimization/79339
31246 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31247 (format_floating): Likewise.
31249 2017-02-03 Martin Liska <mliska@suse.cz>
31252 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31254 (ipa_node_params_t::remove): Likewise.
31255 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31256 initialization from removed ipa_node_params_t::insert.
31257 (ipa_node_params::~ipa_node_params): Move from removed
31258 ipa_node_params_t::release.
31259 * symbol-summary.h (symbol_summary::m_released): New member.
31260 Do not release a summary twice. Do not allow to call finalizer
31261 for types of a summary that live in GGC memory.
31263 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31265 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31268 2017-02-02 Martin Sebor <msebor@redhat.com>
31270 PR middle-end/79275
31271 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31272 (format_string): Tighten up the range of output for non-constant
31273 strings and correct the expected range for wide non-constant strings.
31275 2017-02-02 Martin Sebor <msebor@redhat.com>
31277 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31279 PR middle-end/32003
31280 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31282 (-fdump-tree-@var): Add to index and document how to come up
31283 with pass-specific option and dump file names.
31284 (-fdump-passes): Clarify where to look for output.
31286 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31288 PR middle-end/77445
31289 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31290 statistics of the analyzed path; allow threading for speed when
31291 any of BBs along the path are optimized for speed.
31293 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31295 PR middle-end/78468
31296 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31297 settings of the virtual registers.
31300 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31302 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31303 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31306 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31308 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31309 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31311 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31313 * config/s390/s390.md: Add missing comments with the expanded
31315 * config/s390/vector.md: Likewise.
31316 * config/s390/vx-builtins.md: Likewise.
31318 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31321 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31322 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31323 conditions on a single line.
31325 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31327 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31328 __S390_VX__ to __VX__.
31330 2017-02-01 Andrew Pinski <apinski@cavium.com>
31332 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31333 stmt_info to record_stmt_cost.
31334 (vect_get_known_peeling_cost): Pass stmt_info if known to
31336 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31337 cpu_vector_cost field into
31338 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31339 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31340 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31341 splitting of scalar_stmt_cost and vec_stmt_cost.
31342 (thunderx_vector_cost): Likewise.
31343 (cortexa57_vector_cost): LIkewise.
31344 (exynosm1_vector_cost): Likewise.
31345 (xgene1_vector_cost): Likewise.
31346 (thunderx2t99_vector_cost): Improve after the splitting of the two
31348 (aarch64_builtin_vectorization_cost): Update for the splitting of
31349 scalar_stmt_cost and vec_stmt_cost.
31351 2017-02-01 Torvald Riegel <triegel@redhat.com>
31352 Richard Henderson <rth@redhat.com>
31354 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31355 conditional on existance of a fast atomic load.
31356 * optabs-query.c (can_atomic_load_p): New function.
31357 * optabs-query.h (can_atomic_load_p): Declare it.
31358 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31359 no fast atomic load is available for the particular size of access.
31360 (expand_atomic_compare_and_swap): Likewise.
31361 (expand_atomic_load): Likewise.
31362 (expand_atomic_store): Likewise.
31363 (expand_atomic_fetch_op): Likewise.
31364 * testsuite/lib/target-supports.exp
31365 (check_effective_target_sync_int_128): Remove x86 because it provides
31366 no fast atomic load.
31367 (check_effective_target_sync_int_128_runtime): Likewise.
31369 2017-02-01 Richard Biener <rguenther@suse.de>
31371 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31372 (graphite_transform_loops): Provide opt-info for optimized nests.
31373 * tree-parloop.c (parallelize_loops): Provide opt-info for
31374 parallelized loops.
31376 2017-02-01 Richard Biener <rguenther@suse.de>
31378 PR middle-end/79315
31379 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31380 was not set before.
31382 2017-02-01 Richard Biener <rguenther@suse.de>
31384 PR tree-optimization/71824
31385 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31386 Verify the loops are valid in the merged SESE region.
31387 (scop_detection::can_represent_loop_1): Check analyzing the
31388 evolution of the number of iterations in the region succeeds.
31390 2017-01-31 Ian Lance Taylor <iant@golang.org>
31392 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31393 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31395 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31397 PR preprocessor/79210
31398 * input.c (get_substring_ranges_for_loc): Replace line_width
31399 assertion with error-handling.
31401 2017-01-31 Richard Biener <rguenther@suse.de>
31403 PR tree-optimization/77318
31404 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31405 (create_pw_aff_from_tree): Take loop parameter.
31406 (add_condition_to_pbb): Pass loop of the condition to
31407 create_pw_aff_from_tree.
31409 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31411 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31412 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31414 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31418 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31420 (convert_int_to_float128): Likewise.
31421 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31422 (convert_int_to_float128): Likewise.
31423 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31424 (UNSPEC_IEEE128_CONVERT): Likewise.
31425 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31426 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31427 Use local variables for IBM extended format.
31428 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31429 (fix_trunc<mode>si2_fprs): Likewise.
31430 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31431 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31432 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31433 to know that we can now have integers of all sizes in vector
31435 (fix<uns>_<mode>di2_hw): Likewise.
31436 (float<uns>_<mode>si2_hw): Likewise.
31437 (fix_<mode>si2_hw): Likewise.
31438 (fixuns_<mode>si2_hw): Likewise.
31439 (float<uns>_<mode>di2_hw): Likewise.
31440 (float_<mode>di2_hw): Likewise.
31441 (float_<mode>si2_hw): Likewise.
31442 (floatuns_<mode>di2_hw): Likewise.
31443 (floatuns_<mode>si2_hw): Likewise.
31444 (xscvqp<su>wz_<mode>): Delete, no longer used.
31445 (xscvqp<su>dz_<mode>): Likewise.
31446 (xscv<su>dqp_<mode>): Likewise.
31447 (ieee128_mfvsrd_64bit): Likewise.
31448 (ieee128_mfvsrd_32bit): Likewise.
31449 (ieee128_mfvsrwz): Likewise.
31450 (ieee128_mtvsrw): Likewise.
31451 (ieee128_mtvsrd_64bit): Likewise.
31452 (ieee128_mtvsrd_32bit): Likewise.
31454 2017-01-31 Martin Liska <mliska@suse.cz>
31457 * ipa-prop.c (ipa_free_all_node_params): Call release method
31458 instead of ~sumbol_summary to not to trigger double times
31461 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31463 PR tree-optimization/71691
31464 * bitmap.h (class auto_bitmap): New.
31465 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31466 is_maybe_undefined instead of ssa_undefined_value_p.
31468 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31470 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31471 __S390_ARCH_LEVEL__ to __ARCH__.
31473 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31475 PR tree-optimization/79267
31476 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31477 if should_remove_lhs_p is true.
31479 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31482 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31483 (add_alignment_attribute): New.
31484 (base_type_die): Add alignment attribute.
31485 (subrange_type_die): Likewise.
31486 (modified_type_die): Likewise.
31487 (gen_array_type_die): Likewise.
31488 (gen_descr_array_type_die: Likewise.
31489 (gen_enumeration_type_die): Likewise.
31490 (gen_subprogram_die): Likewise.
31491 (gen_variable_die): Likewise.
31492 (gen_field_die): Likewise.
31493 (gen_ptr_to_mbr_type_die): Likewise.
31494 (gen_struct_or_union_type_die): Likewise.
31495 (gen_subroutine_type_die): Likewise.
31496 (gen_typedef_die): Likewise.
31497 (base_type_cmp): Compare alignment attribute.
31499 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31502 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31503 (setb_unsigned) New pattern for setb with CCUNS.
31504 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31505 subfc./subfe sequence to avoid overflow problems. Generate a
31506 shorter sequence with cmpld/setb for power9.
31507 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31508 for generating subfc. instruction.
31509 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31510 now uses this instruction.
31512 2017-01-30 Ian Lance Taylor <iant@google.com>
31515 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31516 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31518 2017-01-30 Martin Sebor <msebor@redhat.com>
31520 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31521 Move constant to the right of a relational operator.
31522 (get_mpfr_format_length, format_character, format_string): Ditto.
31523 (should_warn_p, maybe_warn): Same.
31525 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31527 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31530 * asan.c (get_translation_unit_decl): Remove function.
31531 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31533 2017-01-30 Martin Liska <mliska@suse.cz>
31535 PR gcov-profile/79259
31536 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31537 -fprofile-generate.
31539 2017-01-30 Martin Liska <mliska@suse.cz>
31542 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31543 Initialize variables with NULL value.
31545 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31548 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31550 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31552 2017-01-30 Richard Biener <rguenther@suse.de>
31554 PR tree-optimization/79276
31555 * tree-vrp.c (process_assert_insertions): Properly adjust common
31556 when removing a duplicate.
31558 2017-01-30 Richard Biener <rguenther@suse.de>
31560 PR tree-optimization/79256
31561 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31562 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31564 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31566 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31569 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31570 ("*r<noxa>sbg_<mode>_sll_bitmask")
31571 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31572 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31573 Use contiguous_bitmask_nowrap_operand.
31575 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31578 * config/rs6000/altivec.h (vec_xl): Revise #define.
31579 (vec_xst): Likewise.
31581 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31583 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31585 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31587 PR rtl-optimization/79194
31588 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31589 traps before call to bypass_conditional_jumps.
31591 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31593 PR tree-optimization/71374
31594 * lra-constraints.c (check_conflict_input_operands): New.
31595 (match_reload): Use it.
31597 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31600 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31601 account to calculate conflict_set.
31603 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31605 PR rtl-optimization/78559
31606 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31607 other_insn in combine.
31609 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31611 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31612 uint16_type_node for BT_UINT16.
31614 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31616 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31617 "RTL Tests" to menu.
31618 (GIMPLE Tests): New node.
31619 (RTL Tests): New node.
31621 2017-01-27 Richard Biener <rguenther@suse.de>
31623 PR tree-optimization/79245
31624 * tree-loop-distribution.c (distribute_loop): Apply cost
31625 modeling also to detected patterns.
31627 2017-01-27 Richard Biener <rguenther@suse.de>
31629 PR tree-optimization/71433
31630 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31631 (compare_assert_loc): New function.
31632 (process_assert_insertions): Sort and optimize assert locations
31633 to remove duplicates and push down identical assertions on
31634 edges to their destination block.
31636 2017-01-27 Richard Biener <rguenther@suse.de>
31638 PR tree-optimization/79244
31639 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31640 out SSA names even if abnormal.
31642 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31644 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31645 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31646 instead of MPFR_RNDN.
31648 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31651 * arm.c (arm_option_override): Don't call build_target_option_node
31652 until after doing all option overrides.
31653 (arm_valid_target_attribute_tree): Likewise.
31655 2017-01-27 Martin Liska <mliska@suse.cz>
31657 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31658 for {cd}tors and C++ {cd}tors.
31660 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31662 * config/s390/s390.md ("*setmem_long_and")
31663 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31665 2017-01-26 Martin Sebor <msebor@redhat.com>
31667 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31670 2017-01-26 Martin Sebor <msebor@redhat.com>
31672 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31673 HAVE_DFmode before using XFmode or DFmode.
31674 (parse_directive): Avoid using the z length modifier to avoid
31675 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31677 PR middle-end/78703
31678 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31679 to accept adjustment as an array.
31680 (get_int_range): New function.
31681 (struct directive): Make width and prec arrays.
31682 (directive::set_width, directive::set_precision): Call get_int_range.
31683 (format_integer, format_floating): Handle width and precision ranges.
31684 (format_string, parse_directive): Same.
31686 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31689 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31690 comdat_type_p set, just clone them, but keep the children in the
31694 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31695 which have direct callers with -fvar-tracking-assignments enabled
31697 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31698 inside of type units.
31700 2017-01-26 Martin Sebor <msebor@redhat.com>
31702 PR middle-end/78703
31703 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31705 (struct format_result): Replace number_chars, number_chars_min,
31706 and number_chars_max with a single member of struct result_range.
31708 (format_result::operator+=): Adjust.
31709 (struct fmtresult): Remove bounded. Handle likely and unlikely
31711 (fmtresult::adjust_for_width_or_precision): New function.
31712 (fmtresult:type_max_digits): New function.
31713 (bytes_remaining): Handle likely and unlikely counters.
31714 (min_bytes_remaining): Remove.
31715 (format_percent): Simplify.
31716 (format_integer, format_floating): Set likely and unlikely counters.
31717 (get_string_length, format_character, format_string): Same.
31718 (format_plain, should_warn_p): New function.
31719 (maybe_warn): Call should_warn_p. Update diagnostic messages
31720 and handle those for all directives, including plain strings.
31721 (format_directive): Handle likely and unlikely counters.
31722 Remove unnecessary quoting from diagnostics. Add an informational
31724 (add_bytes): Remove.
31725 (pass_sprintf_length::compute_format_length): Simplify.
31726 (try_substitute_return_value): Handle likely and unlikely counters.
31728 2017-01-26 Carl Love <cel@us.ibm.com>
31730 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31731 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31733 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31736 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31737 endianess for subregs into account.
31738 * lra-constraints.c (lra_constraints): Do risky transformations
31739 always on the first iteration.
31740 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31741 last_call_used_reg_set.
31742 (process_bb_lives): Define and use last_call_used_reg_set.
31743 * lra.c (lra): Always continue after lra_constraints on the first
31746 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31748 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31750 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31752 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31754 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31755 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31756 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31757 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31758 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31759 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31760 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31761 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31762 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31764 2017-01-26 Marek Polacek <polacek@redhat.com>
31767 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31768 for the third operand.
31770 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31772 PR middle-end/79236
31773 * omp-low.c (struct omp_context): Add simt_stmt field.
31774 (scan_omp_for): Return omp_context *.
31775 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31776 context to the _simt_ SIMD stmt.
31777 (lower_omp_for): For combined SIMD with sibling _simt_
31778 SIMD, make sure to use the same decls in _looptemp_
31779 clauses as in the sibling.
31781 2017-01-26 David Sherwood <david.sherwood@arm.com>
31783 PR middle-end/79212
31784 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31787 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31790 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31791 emit fld b; fld a; if possible.
31793 * brig-builtins.def: Update copyright years.
31794 * config/arm/arm_acle_builtins.def: Update copyright years.
31796 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31799 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31800 constraint instead of o for the stxsd instruction.
31802 2017-01-25 Carl Love <cel@us.ibm.com>
31804 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31805 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31807 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31809 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31811 2017-01-25 Richard Biener <rguenther@suse.de>
31813 PR tree-optimization/69264
31814 * target.def (vector_alignment_reachable): Improve documentation.
31815 * doc/tm.texi: Regenerate.
31816 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31818 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31819 earlier changes with respect to TYPE_USER_ALIGN.
31820 (vector_alignment_reachable_p): Likewise. Improve dumping.
31822 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31825 * config/arm/arm.md (xordi3): Force constant operand into a register
31828 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31830 * doc/invoke.texi (-fstore-merging): Correct default optimization
31831 levels at which it is enabled.
31832 (-O): Move -fstore-merging from list to...
31835 2017-01-25 Richard Biener <rguenther@suse.de>
31838 * omp-expand.c: Include debug.h.
31839 (expand_omp_taskreg): Make sure to generate early debug before
31840 outlining anything from a function.
31841 (expand_omp_target): Likewise.
31842 (grid_expand_target_grid_body): Likewise.
31844 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31847 * asan.c (get_translation_unit_decl): New function.
31848 (asan_add_global): Extract modules file name from globals
31849 TRANSLATION_UNIT_DECL name.
31851 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31854 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31855 for long calls with APCS frame and VFP.
31857 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31859 * cfg.c (original_copy_tables_initialized_p): New function.
31860 * cfg.h (original_copy_tables_initialized_p): New decl.
31861 * cfgrtl.c (relink_block_chain): Guard the call to
31862 free_original_copy_tables with a call to
31863 original_copy_tables_initialized_p.
31864 * cgraph.h (symtab_node::native_rtl_p): New decl.
31865 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31866 (symtab_node::needed_p): Don't assert for early assembly output
31867 for __RTL functions.
31868 (cgraph_node::finalize_function): Set "force_output" for __RTL
31870 (cgraph_node::analyze): Bail out early for __RTL functions.
31871 (analyze_functions): Update assertion to support __RTL functions.
31872 (cgraph_node::expand): Bail out early for __RTL functions.
31873 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31875 * function.h (struct function): Update comment for field
31877 * gimple-expr.c: Include "tree-pass.h".
31878 (gimple_has_body_p): Return false for __RTL functions.
31879 * Makefile.in (OBJS): Add run-rtl-passes.o.
31880 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31882 (gcc::pass_manager::get_clean_slate): New accessor.
31883 * passes.c: Include "insn-addr.h".
31884 (should_skip_pass_p): Add logging. Update logic for running
31885 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31886 property-provider override so it is only done for gimple passes.
31888 (skip_pass): New function.
31889 (execute_one_pass): Call skip_pass when skipping passes.
31890 * read-md.c (md_reader::read_char): Support filtering
31891 the input to a subset of line numbers.
31892 (md_reader::md_reader): Initialize fields
31893 m_first_line and m_last_line.
31894 (md_reader::read_file_fragment): New function.
31895 * read-md.h (md_reader::read_file_fragment): New decl.
31896 (md_reader::m_first_line): New field.
31897 (md_reader::m_last_line): New field.
31898 * read-rtl-function.c (function_reader::create_function): Only
31899 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31900 curr_properties. Set DECL_INITIAL to a dummy block.
31901 (read_rtl_function_body_from_file_range): New function.
31902 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31904 * run-rtl-passes.c: New file.
31905 * run-rtl-passes.h: New file.
31907 2017-01-24 Jeff Law <law@redhat.com>
31909 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31912 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31914 PR tree-optimization/79159
31915 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31916 (record_nonwrapping_iv): Improve boundary using above function if no
31917 value range information.
31919 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31920 Martin Jambor <mjambor@suse.cz>
31922 * brig-builtins.def: New file.
31923 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31924 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31925 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31926 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31927 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31928 * builtin-types.def (BT_INT8): New.
31929 (BT_INT16): Likewise.
31930 (BT_UINT8): Likewise.
31931 (BT_UINT16): Likewise.
31932 (BT_FN_ULONG): Likewise.
31933 (BT_FN_UINT_INT): Likewise.
31934 (BT_FN_UINT_ULONG): Likewise.
31935 (BT_FN_UINT_LONG): Likewise.
31936 (BT_FN_UINT_PTR): Likewise.
31937 (BT_FN_ULONG_PTR): Likewise.
31938 (BT_FN_INT8_FLOAT): Likewise.
31939 (BT_FN_INT16_FLOAT): Likewise.
31940 (BT_FN_UINT32_FLOAT): Likewise.
31941 (BT_FN_UINT16_FLOAT): Likewise.
31942 (BT_FN_UINT8_FLOAT): Likewise.
31943 (BT_FN_UINT64_FLOAT): Likewise.
31944 (BT_FN_UINT16_UINT32): Likewise.
31945 (BT_FN_UINT32_UINT16): Likewise.
31946 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31947 (BT_FN_INT_PTR_INT): Likewise.
31948 (BT_FN_UINT_PTR_UINT): Likewise.
31949 (BT_FN_LONG_PTR_LONG): Likewise.
31950 (BT_FN_ULONG_PTR_ULONG): Likewise.
31951 (BT_FN_VOID_UINT64_UINT64): Likewise.
31952 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31953 (BT_FN_INT8_INT8_INT8): Likewise.
31954 (BT_FN_INT16_INT16_INT16): Likewise.
31955 (BT_FN_INT_INT_INT): Likewise.
31956 (BT_FN_UINT_FLOAT_UINT): Likewise.
31957 (BT_FN_FLOAT_UINT_UINT): Likewise.
31958 (BT_FN_ULONG_UINT_UINT): Likewise.
31959 (BT_FN_ULONG_UINT_PTR): Likewise.
31960 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31961 (BT_FN_UINT_UINT_UINT): Likewise.
31962 (BT_FN_VOID_UINT_PTR): Likewise.
31963 (BT_FN_UINT_UINT_PTR: Likewise.
31964 (BT_FN_UINT32_UINT64_PTR): Likewise.
31965 (BT_FN_INT_INT_UINT_UINT): Likewise.
31966 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31967 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31968 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31969 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31970 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31971 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31972 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31973 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31974 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31975 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31976 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31977 * doc/frontends.texi: List BRIG FE.
31978 * doc/install.texi (Testing): Add BRIG tesring requirements.
31979 * doc/invoke.texi (Overall Options): Mention BRIG.
31980 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31982 2017-01-24 Richard Biener <rguenther@suse.de>
31984 PR translation/79208
31985 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31987 2017-01-24 Martin Jambor <mjambor@suse.cz>
31990 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31991 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31992 and known_contexts.
31994 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31996 PR middle-end/79123
31997 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31998 casts from signed to unsigned really don't have a range.
32000 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32002 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32003 GMP_RNDx for compatiblity.
32005 2017-01-24 Martin Liska <mliska@suse.cz>
32008 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32009 that would prevent us to call alloca with -1 as argument.
32011 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32013 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32014 Avoid -Wformat-security warning.
32016 2017-01-23 Andrew Pinski <apinski@cavium.com>
32018 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32021 2017-01-23 Martin Sebor <msebor@redhat.com>
32023 PR middle-end/78703
32024 * gimple-ssa-sprintf.c (warn_level): New global.
32025 (format_integer): Use it here and throughout the rest of the file.
32026 Use the same switch to compute sign as base.
32027 (maybe_warn): New function.
32028 (format_directive): Factor out warnings into maybe_warn.
32029 Add debugging output. Use warn_level.
32030 (add_bytes): Use warn_level.
32031 (pass_sprintf_length::compute_format_length): Add debugging output.
32032 (try_substitute_return_value): Same.
32033 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32035 PR middle-end/78703
32036 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32037 (struct fmtresult, format_integer, format_floating): Adjust.
32038 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32039 (get_string_length, format_string,format_directive): Same.
32040 (pass_sprintf_length::compute_format_length): Same.
32041 (try_substitute_return_value): Simplify slightly.
32043 PR middle-end/78703
32044 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32045 (fmtresult::operator+=): Outlined.
32046 (struct fmtresult): Add ctors.
32047 (struct conversion_spec): Rename...
32048 (struct directive): ...to this. Add and remove data members.
32049 (directive::set_width, directive::set_precision): New functions.
32050 (format_percent): Use fmtresult ctor.
32051 (get_width_and_precision): Remove.
32052 (format_integer): Make naming changes. Avoid computing width and
32054 (format_floating): Same. Adjust indentation.
32055 (format_character, format_none): New functions.
32056 (format_string): Moved character handling to format_character.
32057 (format_directive): Remove arguments, change return type.
32058 (parse_directive): New function.
32059 (pass_sprintf_length::compute_format_length): Move directive
32060 parsing to parse_directive.
32062 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32064 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32065 (assign_assembler_name_if_needed): ... this.
32066 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32067 (assign_assembler_name_if_needed): ... this.
32068 (free_lang_data_in_cgraph): Adjust callers.
32069 * cgraphunit.c (cgraph_node::analyze): Likewise.
32070 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32073 2017-01-23 Richard Biener <rguenther@suse.de>
32075 PR tree-optimization/79088
32076 PR tree-optimization/79188
32077 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32078 resetting loop bounds after last path deletion. Reset loop
32079 bounds of the target loop, make code match the comments.
32080 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32081 Make sure loops need no fixups.
32083 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32085 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32086 exponent support with double type for first argument.
32087 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32088 type returned by __builtin_vec_extract_sig,
32089 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32090 functions from "vector int" to "vector unsigned int" or from
32091 "vector long long int" to "vector unsigned long long int".
32092 Changed type returned by __builtin_vec_extract_exp,
32093 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_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 return type of __builtin_vec_test_data_class,
32097 __builtin_vec_test_data_class_sp, and
32098 __builtin_vec_test_data_class_dp from "vector int" to
32099 "vector bool int" or from "vector long long int" to "vector bool
32100 long long int" and changed second argument type from "unsigned
32101 int" to "int". Added new overloaded function forms "vector float
32102 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32103 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32104 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32105 double, vector unsigned long long int)" and "vector double
32106 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32107 long int)". Changed return type of
32108 __builtin_scalar_test_data_class and
32109 __builtin_scalar_test_data_class_sp and
32110 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32111 int" and changed second argument from "unsigned int" to "int".
32112 Changed type returned by __builtin_scalar_test_neg,
32113 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32114 from "int" to "bool int". Added new overloaded function form
32115 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32116 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32117 exponent double-precision with floating point first argument.
32118 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32119 documentation of scalar_test_data_class, scalar_test_neg,
32120 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32121 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32122 vec_test_data_class built-in functions to reflect refinements in
32123 their type signatures.
32125 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32127 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32129 (aarch64_elf_asm_destructor): Likewise.
32131 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32133 PR rtl-optimization/78634
32134 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32135 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32136 * ifcvt.c (noce_try_cmove): Add missing cost check.
32138 PR rtl-optimization/71724
32139 * combine.c (if_then_else_cond): Look for situations where it is
32140 beneficial to undo the work of one of the recursive calls.
32142 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32144 PR tree-optimization/70754
32145 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32146 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32147 combined stmt before it if not NULL.
32148 (combine_chains): Process refs reversely and compute dominance point
32151 2017-01-23 Martin Liska <mliska@suse.cz>
32153 PR tree-optimization/79196
32154 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32155 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32157 (strlen_optimize_stmt): Call the renamed function.
32159 2017-01-23 Michael Matz <matz@suse.de>
32161 PR tree-optimization/78384
32162 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32164 2017-01-23 Richard Biener <rguenther@suse.de>
32166 PR tree-optimization/79186
32167 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32168 both incoming edges before moving an assert.
32170 2017-01-23 Martin Jambor <mjambor@suse.cz>
32172 * ipa-prop.c (load_from_param_1): Removed.
32173 (load_from_unmodified_param): Bits from load_from_param_1 put back
32175 (load_from_param): Removed.
32176 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32177 with stmt. Reverted back to use of load_from_unmodified_param.
32179 2017-01-23 Martin Jambor <mjambor@suse.cz>
32182 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32183 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32184 field a pointer to garbage collected vector, mark lattices and
32185 ipcp_orig_node with GTY((skip)).
32186 (ipa_get_param_count): Adjust to descriptors being a pointer.
32187 (ipa_get_param): Likewise.
32188 (ipa_get_type): Likewise.
32189 (ipa_get_param_move_cost): Likewise.
32190 (ipa_set_param_used): Likewise.
32191 (ipa_get_controlled_uses): Likewise.
32192 (ipa_set_controlled_uses): Likewise.
32193 (ipa_is_param_used): Likewise.
32194 (ipa_node_params_t): Move into garbage collector. New methods insert
32196 (ipa_node_params_sum): Annotate wth GTY(()).
32197 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32199 (ipa_load_from_parm_agg): Adjust declaration.
32200 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32201 * ipa-profile.c (ipa_profile): Likewise.
32202 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32203 (ipa_populate_param_decls): Make descriptors parameter garbage
32205 (ipa_dump_param): Adjust to descriptors being a pointer.
32206 (ipa_alloc_node_params): Likewise.
32207 (ipa_initialize_node_params): Likewise.
32208 (load_from_param_1): Make descriptors parameter garbage collected.
32209 (load_from_unmodified_param): Likewise.
32210 (load_from_param): Likewise.
32211 (ipa_load_from_parm_agg): Likewise.
32212 (ipa_node_params::~ipa_node_params): Removed.
32213 (ipa_free_all_node_params): Remove call to delete operator.
32214 (ipa_node_params_t::insert): New.
32215 (ipa_node_params_t::remove): Likewise.
32216 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32217 copy known_csts and known_contexts vectors.
32218 (ipa_read_node_info): Adjust to descriptors being a pointer.
32219 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32221 (ipcp_transform_function): Make descriptors variable garbage
32224 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32226 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32227 * config/i386/avx512dqintrin.h: Ditto.
32228 * config/i386/avx512fintrin.h: Ditto.
32229 * config/i386/i386.c: Handle new builtins.
32230 * config/i386/i386-builtin.def: Add new builtins.
32231 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32232 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32234 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32235 Martin Liska <mliska@suse.cz>
32237 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32238 * asan.c (asan_expand_poison_ifn): Support stores and use
32239 appropriate ASAN report function.
32240 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32241 * internal-fn.def (ASAN_POISON_USE): Declare.
32242 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32243 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32244 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32245 ASAN_POISON calls w/o LHS.
32246 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32247 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32248 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32249 * gimplify.c (asan_poison_variables): Add attribute
32250 use_after_scope_memory to variables that really needs to live
32252 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32253 having the attribute.
32255 2017-01-23 Martin Liska <mliska@suse.cz>
32257 * asan.c (create_asan_shadow_var): New function.
32258 (asan_expand_poison_ifn): Likewise.
32259 * asan.h (asan_expand_poison_ifn): New declaration.
32260 * internal-fn.c (expand_ASAN_POISON): Likewise.
32261 * internal-fn.def (ASAN_POISON): New builtin.
32262 * sanopt.c (pass_sanopt::execute): Expand
32263 asan_expand_poison_ifn.
32264 * tree-inline.c (copy_decl_for_dup_finish): Make function
32266 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32267 * tree-ssa.c (is_asan_mark_p): New function.
32268 (execute_update_addresses_taken): Rewrite local variables
32269 (identified just by use-after-scope as addressable) into SSA.
32271 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32273 * doc/install.texi (Specific): opensource.apple.com uses https
32274 now. Remove trailing slash.
32276 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32278 * README.Portability: Remove note on an Irix compatibility issue.
32280 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32282 * gcov.c (INCLUDE_ALGORITHM): Define.
32283 (INCLUDE_VECTOR): Define.
32284 No longer include <vector> and <algorithm> directly.
32286 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32288 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32290 * doc/invoke.texi (Code Gen Options): Ditto.
32292 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32295 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32297 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32299 rtl-optimization/79125
32300 * cprop.c (local_cprop_pass): Handle cases where we make an
32301 unconditional trap.
32303 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32307 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32308 read from, for big endian.
32310 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32312 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32313 register pauth builtins for LP64 only.
32315 2017-01-20 Marek Polacek <polacek@redhat.com>
32318 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32321 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32323 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32325 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32326 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32327 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32329 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32332 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32333 in big-endian mode when they are not a single duplicated value.
32335 2017-01-20 Richard Biener <rguenther@suse.de>
32337 * BASE-VER: Bump to 7.0.1.
32339 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32341 * omp-low.c (omplow_simd_context): New struct. Use it...
32342 (lower_rec_simd_input_clauses): ...here and...
32343 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32344 references to idx, lane, max_vf, is_simt.
32346 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32348 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32351 2017-01-20 Martin Jambor <mjambor@suse.cz>
32353 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32354 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32356 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32357 (GTFILES): Rename hsa.c to hsa-common.c.
32358 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32359 * hsa-dump.c: Likewise.
32360 * hsa-gen.c: Likewise.
32361 * hsa-regalloc.c: Likewise.
32362 * ipa-hsa.c: Likewise.
32363 * omp-expand.c: Likewise.
32364 * omp-low.c: Likewise.
32365 * toplev.c: Likewise.
32367 2017-01-20 Marek Polacek <polacek@redhat.com>
32370 * doc/invoke.texi: Document -Wduplicated-branches.
32371 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32372 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32373 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32374 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32375 return 0 only when not OEP_LEXICOGRAPHIC.
32376 (fold_build_cleanup_point_expr): Use the expression
32377 location when building CLEANUP_POINT_EXPR.
32378 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32379 * tree.c (add_expr): Handle error_mark_node.
32381 2017-01-20 Martin Liska <mliska@suse.cz>
32384 * tree-profile.c (init_ic_make_global_vars): Do not call
32386 (gimple_init_gcov_profiler): Likewise.
32388 2017-01-20 Martin Liska <mliska@suse.cz>
32391 * cgraph.h (maybe_create_reference): Remove argument and
32393 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32395 * ipa-cp.c (create_specialized_node): Likewise.
32396 * symtab.c (symtab_node::maybe_create_reference): Handle
32397 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32399 2017-01-20 Martin Liska <mliska@suse.cz>
32401 * read-rtl-function.c (function_reader::create_function): Use
32402 build_decl instread of build_decl_stat.
32404 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32406 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32407 * config/i386/avx512dqintrin.h: Ditto.
32408 * config/i386/avx512fintrin.h: Ditto.
32409 * config/i386/i386-builtin-types.def: Add new types.
32410 * config/i386/i386.c: Handle new types.
32411 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32412 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32413 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32414 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32415 (__builtin_ia32_kshiftridi): New.
32416 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32418 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32422 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32423 define to rs6000_init_stack_protect_guard.
32424 (rs6000_init_stack_protect_guard): New function.
32426 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32427 Yunqiang Su <yunqiang.su@imgtec.com>
32429 * config.gcc (supported_defaults): Add madd4.
32430 (with_madd4): Add validation.
32431 (all_defaults): Add madd4.
32432 * config/mips/mips.opt (mmadd4): New option.
32433 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32435 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32437 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32438 (ISA_HAS_FUSED_MADD4): Likewise.
32439 * doc/invoke.texi (-mmadd4): Document the new option.
32440 * doc/install.texi (--with-madd4): Document the new option.
32442 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32444 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32445 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32446 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32447 (aarch64_init_pauth_hint_builtins): New.
32448 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32449 (aarch64_expand_builtin): Expand new builtins.
32451 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32453 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32454 * combine-stack-adj.c (no_unhandled_cfa): Handle
32455 REG_CFA_TOGGLE_RA_MANGLE.
32456 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32457 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32458 info for return address signing.
32459 (aarch64_expand_epilogue): Likewise.
32461 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32463 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32464 * config/aarch64/aarch64-protos.h
32465 (aarch64_return_address_signing_enabled): New declaration.
32466 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32468 (aarch64_expand_prologue): Sign return address before it's pushed onto
32470 (aarch64_expand_epilogue): Authenticate return address fetched from
32472 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32473 (aarch64_attributes): New function attributes for "sign-return-address".
32474 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32475 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32476 ("*do_return"): Generate combined instructions according to key index.
32477 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32478 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32480 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32481 * config/aarch64/aarch64.opt (msign-return-address=): New.
32482 * doc/extend.texi (AArch64 Function Attributes): Documents
32483 "sign-return-address=".
32484 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32486 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32488 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32489 overall option summary.
32491 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32493 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32494 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32495 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32496 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32498 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32500 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32501 -mpower9-minmax by default for -mcpu=power9.
32502 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32503 128-bit floating point.
32505 2017-01-20 Alan Modra <amodra@gmail.com>
32507 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32508 optimizing for size.
32510 2017-01-20 Alan Modra <amodra@gmail.com>
32513 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32514 for strcmp and strncmp from corresponding builtin decl.
32516 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32518 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32519 instead of i386/rtems-64.h.
32520 * config/i386/rtems-64.h: Remove.
32522 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32526 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32528 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32530 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32532 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32533 Change int to HOST_WIDE_INT.
32534 * config/aarch64/aarch64-protos.h
32535 (aarch64_simd_gen_const_vector_dup): Likewise.
32536 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32538 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32540 * langhooks-def.h (lhd_type_for_size): New decl.
32541 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32542 * langhooks.c (lhd_type_for_size): New function, taken from
32545 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32547 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32548 define_bypass for CR latency.
32549 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32550 (power9-alu2): Add define_bypass for CR latency.
32552 (power9-mul): Update insn latency.
32553 (power9-mul-compare): Update insn latency, bypass latency and remove
32556 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32558 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32560 * config/aarch64/aarch64.md
32561 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32562 aarch64_nopcrelative_literal_loads.
32563 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32565 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32567 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32568 TARGET_LOONGSON_3A.
32569 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32571 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32574 * config.gcc (supported_defaults): Add lxc1-sxc1.
32575 (with_lxc1_sxc1): Add validation.
32576 (all_defaults): Add lxc1-sxc1.
32577 * config/mips/mips.opt (mlxc1-sxc1): New option.
32578 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32580 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32581 __mips_no_lxc1_sxc1.
32582 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32583 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32584 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32586 2017-01-19 Richard Biener <rguenther@suse.de>
32588 PR tree-optimization/72488
32589 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32590 sure to restore SSA info.
32591 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32593 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32595 PR rtl-optimization/79121
32596 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32597 of the inner type when shifting an extended value.
32599 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32602 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32603 interposable aliases.
32605 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32608 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32609 Use the evmergelohi instruction.
32610 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32611 (mov_sitf_e500_subreg8_2_be): Likewise.
32612 (mov_sitf_e500_subreg12_2_le): Likewise.
32613 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32614 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32615 (mov_sitf_e500_subreg8_2_le): Likewise.
32616 (mov_sitf_e500_subreg12_2_be): Likewise.
32618 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32620 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32621 attribute from vecsimple to vecperm.
32622 (altivec_vbpermq2): Likewise.
32624 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32627 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32629 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32630 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32631 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32632 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32633 case where N arg is SIZE_MAX.
32634 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32635 (cmpstrsi): Add pattern.
32637 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32639 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32640 __builtin_vec_revb builtins.
32641 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32642 built-in functions to support generation of the ISA 3.0 XXBR<x>
32643 vector byte reverse instructions.
32644 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32645 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32646 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32647 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32648 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32649 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32650 (P9V_BUILTIN_VEC_REVB): Likewise.
32651 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32652 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32653 (p9_xxbrq_v16qi): Likewise.
32654 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32655 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32656 (p9_xxbrh_v8hi): Likewise.
32657 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32658 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32659 vec_revb built-in functions.
32661 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32663 PR rtl-optimization/78952
32664 * config/i386/i386.md (any_extract): New code iterator.
32665 (*insvqi_2): Use any_extract for source operand.
32666 (*insvqi_3): Use any_shiftrt for source operand.
32668 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32670 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32672 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32674 2017-01-18 Matthias Klose <doko@ubuntu.com>
32676 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32678 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32680 * config/rs6000/altivec.h (vec_bperm): Change #define.
32681 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32682 (altivec_vbpermq2): New define_insn.
32683 (altivec_vbpermd): Likewise.
32684 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32685 function interface.
32686 (VBPERMD): Likewise.
32687 (VBPERM): New polymorphic function interface.
32688 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32689 Add entries for P9V_BUILTIN_VEC_VBPERM.
32690 * doc/extend.texi: Add interfaces for vec_bperm.
32692 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32694 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32695 first letter of error messages.
32696 (s390_resolve_overloaded_builtin): Likewise.
32697 * config/s390/s390.c (s390_expand_builtin): Likewise.
32698 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32699 (s390_valid_target_attribute_inner_p): Likewise.
32700 * config/s390/s390.md ("tabort"): Likewise.
32702 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32704 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32705 (ISA_AVOID_DIV_HILO): New macro.
32706 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32707 (ISA_HAS_DDIV): Likewise.
32709 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32711 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32713 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32715 * doc/invoke.texi (fabi-version): Spelling fix.
32717 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32720 * doc/invoke.texi (fabi-version): Mention mangling fix for
32723 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32726 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32728 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32731 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32732 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32734 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32736 (stack_protect_test): Ditto.
32737 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32738 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32740 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32741 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32742 -mstack-protector-guard-offset=.
32743 (RS/6000 and PowerPC Options): Ditto.
32745 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32747 * config/i386/i386.h (MASK_CLASS_P): New define.
32748 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32749 there are no registers from different register sets also when
32750 mask registers are used. Update function comment.
32751 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32752 to (*k/*r) and (*k/*km) alternatives.
32754 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32756 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32757 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32758 (EH_RETURN_HANDLER_RTX): New define.
32759 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32760 Force frame pointer in EH return functions.
32761 (aarch64_expand_epilogue): Add barrier for eh_return.
32762 (aarch64_final_eh_return_addr): Remove.
32763 (aarch64_eh_return_handler_rtx): New function.
32764 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32766 (aarch64_eh_return_handler_rtx): New prototype.
32768 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32770 * config/rs6000/altivec.h (vec_rlmi): New #define.
32771 (vec_vrlnm): Likewise.
32772 (vec_rlnm): Likewise.
32773 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32774 (UNSPEC_VRLNM): Likewise.
32775 (VIlong): New mode iterator.
32776 (altivec_vrl<VI_char>mi): New define_insn.
32777 (altivec_vrl<VI_char>nm): Likewise.
32778 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32780 (VRLDNM): Likewise.
32781 (RLNM): New polymorphic function entry.
32782 (VRLWMI): New monomorphic function entry.
32783 (VRLDMI): Likewise.
32784 (RLMI): New polymorphic function entry.
32785 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32786 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32787 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32790 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32793 * dwarf2out.c (field_byte_offset): Restore the
32794 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32795 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32796 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32799 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32802 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32804 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32807 * dwarf2out.c (add_data_member_location_attribute): For constant
32808 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32809 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32810 DW_AT_byte_size attributes.
32812 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32814 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32815 after forcing to constant memory when the code model is medium.
32817 2017-01-17 Julia Koval <julia.koval@intel.com>
32820 * config/i386/avx512fintrin.h
32821 (_mm512_i32gather_ps): Change __addr type to void const*.
32822 (_mm512_mask_i32gather_ps): Ditto.
32823 (_mm512_i32gather_pd): Ditto.
32824 (_mm512_mask_i32gather_pd): Ditto.
32825 (_mm512_i64gather_ps): Ditto.
32826 (_mm512_mask_i64gather_ps): Ditto.
32827 (_mm512_i64gather_pd): Ditto.
32828 (_mm512_mask_i64gather_pd): Ditto.
32829 (_mm512_i32gather_epi32): Ditto.
32830 (_mm512_mask_i32gather_epi32): Ditto.
32831 (_mm512_i32gather_epi64): Ditto.
32832 (_mm512_mask_i32gather_epi64): Ditto.
32833 (_mm512_i64gather_epi32): Ditto.
32834 (_mm512_mask_i64gather_epi32): Ditto.
32835 (_mm512_i64gather_epi64): Ditto.
32836 (_mm512_mask_i64gather_epi64): Ditto.
32837 (_mm512_i32scatter_ps): Change __addr type to void*.
32838 (_mm512_mask_i32scatter_ps): Ditto.
32839 (_mm512_i32scatter_pd): Ditto.
32840 (_mm512_mask_i32scatter_pd): Ditto.
32841 (_mm512_i64scatter_ps): Ditto.
32842 (_mm512_mask_i64scatter_ps): Ditto.
32843 (_mm512_i64scatter_pd): Ditto.
32844 (_mm512_mask_i64scatter_pd): Ditto.
32845 (_mm512_i32scatter_epi32): Ditto.
32846 (_mm512_mask_i32scatter_epi32): Ditto.
32847 (_mm512_i32scatter_epi64): Ditto.
32848 (_mm512_mask_i32scatter_epi64): Ditto.
32849 (_mm512_i64scatter_epi32): Ditto.
32850 (_mm512_mask_i64scatter_epi32): Ditto.
32851 (_mm512_i64scatter_epi64): Ditto.
32852 (_mm512_mask_i64scatter_epi64): Ditto.
32853 * config/i386/avx512pfintrin.h
32854 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32855 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32856 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32857 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32858 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32859 (_mm512_prefetch_i32scatter_ps): Ditto.
32860 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32861 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32862 (_mm512_prefetch_i64scatter_pd): Ditto.
32863 (_mm512_prefetch_i64scatter_ps): Ditto.
32864 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32865 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32866 * config/i386/avx512vlintrin.h
32867 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32868 (_mm_mmask_i32gather_ps): Ditto.
32869 (_mm256_mmask_i32gather_pd): Ditto.
32870 (_mm_mmask_i32gather_pd): Ditto.
32871 (_mm256_mmask_i64gather_ps): Ditto.
32872 (_mm_mmask_i64gather_ps): Ditto.
32873 (_mm256_mmask_i64gather_pd): Ditto.
32874 (_mm_mmask_i64gather_pd): Ditto.
32875 (_mm256_mmask_i32gather_epi32): Ditto.
32876 (_mm_mmask_i32gather_epi32): Ditto.
32877 (_mm256_mmask_i32gather_epi64): Ditto.
32878 (_mm_mmask_i32gather_epi64): Ditto.
32879 (_mm256_mmask_i64gather_epi32): Ditto.
32880 (_mm_mmask_i64gather_epi32): Ditto.
32881 (_mm256_mmask_i64gather_epi64): Ditto.
32882 (_mm_mmask_i64gather_epi64): Ditto.
32883 (_mm256_i32scatter_ps): Change __addr type to void*.
32884 (_mm256_mask_i32scatter_ps): Ditto.
32885 (_mm_i32scatter_ps): Ditto.
32886 (_mm_mask_i32scatter_ps): Ditto.
32887 (_mm256_i32scatter_pd): Ditto.
32888 (_mm256_mask_i32scatter_pd): Ditto.
32889 (_mm_i32scatter_pd): Ditto.
32890 (_mm_mask_i32scatter_pd): Ditto.
32891 (_mm256_i64scatter_ps): Ditto.
32892 (_mm256_mask_i64scatter_ps): Ditto.
32893 (_mm_i64scatter_ps): Ditto.
32894 (_mm_mask_i64scatter_ps): Ditto.
32895 (_mm256_i64scatter_pd): Ditto.
32896 (_mm256_mask_i64scatter_pd): Ditto.
32897 (_mm_i64scatter_pd): Ditto.
32898 (_mm_mask_i64scatter_pd): Ditto.
32899 (_mm256_i32scatter_epi32): Ditto.
32900 (_mm256_mask_i32scatter_epi32): Ditto.
32901 (_mm_i32scatter_epi32): Ditto.
32902 (_mm_mask_i32scatter_epi32): Ditto.
32903 (_mm256_i32scatter_epi64): Ditto.
32904 (_mm256_mask_i32scatter_epi64): Ditto.
32905 (_mm_i32scatter_epi64): Ditto.
32906 (_mm_mask_i32scatter_epi64): Ditto.
32907 (_mm256_i64scatter_epi32): Ditto.
32908 (_mm256_mask_i64scatter_epi32): Ditto.
32909 (_mm_i64scatter_epi32): Ditto.
32910 (_mm_mask_i64scatter_epi32): Ditto.
32911 (_mm256_i64scatter_epi64): Ditto.
32912 (_mm256_mask_i64scatter_epi64): Ditto.
32913 (_mm_i64scatter_epi64): Ditto.
32914 (_mm_mask_i64scatter_epi64): Ditto.
32915 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32916 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32917 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32918 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32919 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32920 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32921 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32922 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32923 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32924 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32925 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32926 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32927 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32928 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32929 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32930 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32931 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32932 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32933 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32934 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32935 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32936 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32937 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32938 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32939 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32940 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32941 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32942 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32943 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32944 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32945 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32946 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32947 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32948 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32949 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32950 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32951 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32952 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32953 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32954 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32955 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32956 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32957 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32958 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32959 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32960 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32961 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32962 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32963 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32964 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32965 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32966 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32967 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32968 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32969 definitions accordingly.
32971 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32972 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32975 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32978 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32981 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32982 hard regs for inner regno.
32984 2017-01-17 Martin Liska <mliska@suse.cz>
32987 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32988 assumption and add comment.
32990 2017-01-17 Nathan Sidwell <nathan@acm.org>
32992 * ipa-visibility.c (localize_node): New function, broken out of ...
32993 (function_and_variable_visibility): ... here. Call it.
32995 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32997 PR middle-end/77445
32998 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32999 correctly set frequency of oudgoing edge.
33000 (duplicate_thread_path): Fix profile updating.
33002 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33005 * configure.ac: Add GCC_BASE_VER.
33006 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33007 version from BASE-VER file.
33008 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33009 (gcc.o): Depend on $(BASEVER).
33010 * common.opt (dumpfullversion): New option.
33011 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33012 * doc/invoke.texi: Document -dumpfullversion.
33013 * doc/install.texi: Document --with-gcc-major-version-only.
33014 * configure: Regenerated.
33016 2017-01-17 Richard Biener <rguenther@suse.de>
33018 PR tree-optimization/71433
33019 * tree-vrp.c (register_new_assert_for): Merge same asserts
33020 on all incoming edges.
33021 (process_assert_insertions_for): Handle insertions at the
33024 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33026 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33027 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33029 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33032 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33035 2017-01-17 Alan Modra <amodra@gmail.com>
33038 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33039 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33040 symbolic stack limit when pic.
33042 2017-01-16 Martin Sebor <msebor@redhat.com>
33044 PR tree-optimization/78608
33045 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33047 2017-01-16 Jeff Law <law@redhat.com>
33050 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33051 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33052 for several include directories that may be relative to sysroot.
33053 * config/i386/x-mingw32 (gplus_includedir): Define.
33054 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33055 (native_system_includedir): Likewise.
33056 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33057 override if TARGET_SYSTEM_ROOT is defined.
33058 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33060 PR tree-optimization/79090
33061 PR tree-optimization/33562
33062 PR tree-optimization/61912
33063 PR tree-optimization/77485
33064 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33065 and computed trims into the dump file.
33067 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33069 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33071 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33074 * gimplify.c (gimplify_init_constructor): If want_value and
33075 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33079 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33080 sequence. Formatting fixes.
33081 (doloop_optimize): Formatting fixes.
33084 * gcc.c (debug_level_greater_than_spec_func): New function.
33085 (static_spec_functions): Add debug-level-gt spec function.
33086 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33088 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33089 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33090 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33091 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33092 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33093 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33095 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33097 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33098 QImode fixups to general and mask registers only.
33100 2017-01-16 Carl Love <cel@us.ibm.com>
33102 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33103 for built-in functions
33104 vector signed char vec_nabs (vector signed char)
33105 vector signed short vec_nabs (vector signed short)
33106 vector signed int vec_nabs (vector signed int)
33107 vector signed long long vec_nabs (vector signed long long)
33108 vector float vec_nabs (vector float)
33109 vector double vec_nabs (vector double)
33110 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33112 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33113 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33114 * doc/extend.texi: Update the documentation file for the new built-in
33117 2017-01-16 Martin Sebor <msebor@redhat.com>
33119 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33122 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33124 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33125 UNSPEC_VSX__XXSPLTD to require special splat handling.
33127 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33130 * system.h: Poison strndup.
33132 2017-01-16 Alan Modra <amodra@gmail.com>
33135 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33138 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33140 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33142 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33144 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33145 call recog here. Assert that INSN_CODE (insn) is non-negative.
33147 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33150 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33152 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33153 in the currently scheduled RTL fragment.
33155 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33157 PR rtl-optimization/78751
33158 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33161 2017-01-14 Jeff Law <law@redhat.com>
33163 PR tree-optimization/79090
33164 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33165 variable length stores.
33166 (compute_trims): Delete dead assignment to *trim_tail.
33167 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33170 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33172 PR rtl-optimization/78626
33173 PR rtl-optimization/78727
33174 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33175 of a block, and split such blocks after everything else is finished.
33177 2017-01-14 Alan Modra <amodra@gmail.com>
33180 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33181 target legitimate_combined_insn.
33182 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33183 (rs6000_legitimate_combined_insn): New function.
33184 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33186 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33187 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33188 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33190 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33192 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33194 2017-01-13 Jeff Law <law@redhat.com>
33196 PR tree-optimization/33562
33197 PR tree-optimization/61912
33198 PR tree-optimization/77485
33199 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33201 (delete_dead_assignment): Likewise.
33202 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33203 statement to delete_dead_call and delete_dead_assignment.
33205 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33208 * substring-locations.c (format_warning_va): Strengthen case 1 so
33209 that both endpoints of the substring must be within the format
33210 range for just the substring to be printed.
33212 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33214 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33215 * config/i386/i386.c (ix86_target_string): Add missing options
33216 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33217 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33218 flags_other and ix86_target_other to flags2_other. Display unknown
33220 (ix86_valid_target_attribute_inner_p): Add missing options and
33221 reorder options by implied ISAs, as in ix86_target_string.
33223 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33225 * hash-table.h (hash_table::too_empty_p): New function.
33226 (hash_table::expand): Use it.
33227 (hash_table::traverse): Likewise.
33228 (hash_table::empty_slot): Use sizeof (value_type) instead of
33229 sizeof (PTR) to convert bytes to elements. Shrink the table
33230 if the current size is excessive for the current number of
33233 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33235 * ira-costs.c (record_reg_classes): Break from the inner loop
33236 early once alt_fail is known to be true. Update outer loop
33237 handling accordingly.
33239 2017-01-13 Jeff Law <law@redhat.com>
33241 * tree-ssa-dse.c (decrement_count): New function.
33242 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33243 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33244 when we know the partially dead statement is a mem* function.
33246 PR tree-optimization/61912
33247 PR tree-optimization/77485
33248 * tree-ssa-dse.c: Include expr.h.
33249 (maybe_trim_constructor_store): New function.
33250 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33252 PR tree-optimization/33562
33253 PR tree-optimization/61912
33254 PR tree-optimization/77485
33255 * doc/invoke.texi: Document new dse-max-object-size param.
33256 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33257 * tree-ssa-dse.c: Include params.h.
33258 (dse_store_status): New enum.
33259 (initialize_ao_ref_for_dse): New, partially extracted from
33261 (valid_ao_ref_for_dse, normalize_ref): New.
33262 (setup_live_bytes_from_ref, compute_trims): Likewise.
33263 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33264 (maybe_trim_partially_dead_store): Likewise.
33265 (maybe_trim_complex_store): Likewise.
33266 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33267 Track what bytes live from the original store. Return tri-state
33268 for dead, partially dead or live.
33269 (dse_dom_walker): Add constructor, destructor and new private members.
33270 (delete_dead_call, delete_dead_assignment): New extracted from
33272 (dse_optimize_stmt): Make a member of dse_dom_walker.
33273 Use initialize_ao_ref_for_dse.
33275 PR tree-optimization/33562
33276 PR tree-optimization/61912
33277 PR tree-optimization/77485
33278 * sbitmap.h (bitmap_count_bits): Prototype.
33279 (bitmap_clear_range, bitmap_set_range): Likewise.
33280 * sbitmap.c (bitmap_clear_range): New function.
33281 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33283 2017-01-13 Martin Liska <mliska@suse.cz>
33286 * function.c (set_cfun): Add new argument force.
33287 * function.h (set_cfun): Likewise.
33288 * ipa-inline-transform.c (inline_call): Use the function when
33289 strict alising from is dropped for function we inline to.
33291 2017-01-13 Richard Biener <rguenther@suse.de>
33293 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33294 for dumping GIMPLE INTEGER_CSTs.
33296 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33298 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33299 to 201112L since C++17.
33301 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33304 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33305 if -fsanitize=kernel-address is present.
33307 2017-01-13 Richard Biener <rguenther@suse.de>
33309 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33310 as _Literal ( type ) number in case usual suffixes do not
33311 preserve all information.
33313 2017-01-13 Richard Biener <rguenther@suse.de>
33315 PR tree-optimization/77283
33316 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33317 and ssa-iterators.h.
33318 (is_feasible_trace): Implement a cost model based on joiner
33321 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33324 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33325 char or short to __float128/_Float128 directly.
33327 2017-01-12 Martin Sebor <msebor@redhat.com>
33329 to -Wformat-overflow.
33330 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33331 (min_bytes_remaining): Same.
33332 (get_string_length): Same.
33333 (format_string): Same.
33334 (format_directive): Same.
33336 (pass_sprintf_length::handle_gimple_call): Same.
33338 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33340 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33341 info.nowrite calls with no lhs that can't throw. Return bool
33342 whether gsi_remove has been called or not.
33343 (pass_sprintf_length::handle_gimple_call): Return bool whether
33344 try_substitute_return_value called gsi_remove. Formatting fix.
33345 (pass_sprintf_length::execute): Don't use gsi_remove if
33346 handle_gimple_call returned true.
33349 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33350 be removed due to side-effects, don't remove following barrier nor
33351 turn the successor edge into fallthru edge.
33353 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33356 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33357 element-reversing loads and stores as not swappable.
33359 2017-01-12 Nathan Sidwell <nathan@acm.org>
33360 Nicolai Stange <nicstange@gmail.com>
33362 * combine.c (try_combine): Don't ignore result of overlap checking
33363 loop. Combine overlap & asm check into single loop.
33365 2017-01-12 Richard Biener <rguenther@suse.de>
33367 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33368 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33370 2017-01-12 Richard Biener <rguenther@suse.de>
33372 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33373 and TS_TARGET_OPTION directly derive from TS_BASE.
33374 * tree-core.h (tree_optimization_option): Derive from tree_base.
33375 (tree_target_option): Likewise.
33377 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33379 * config/i386/i386.c (memory_address_length): Increase len
33380 only when rip_relative_addr_p returns false.
33382 2017-01-11 Julia Koval <julia.koval@intel.com>
33384 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33385 (OPTION_MASK_ISA_SGX_SET): New.
33386 (ix86_handle_option): Handle OPT_msgx.
33387 * config.gcc: Added sgxintrin.h.
33388 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33389 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33390 * config/i386/i386.c (ix86_target_string): Add -msgx.
33392 (ix86_option_override_internal): Handle new options.
33393 (ix86_valid_target_attribute_inner_p): Add sgx.
33394 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33395 * config/i386/i386.opt: Add msgx.
33396 * config/i386/sgxintrin.h: New file.
33397 * config/i386/x86intrin.h: Add sgxintrin.h.
33399 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33402 * fold-const.c (maybe_nonzero_address): Return 1 for function
33404 (tree_single_nonzero_warnv_p): Don't handle function local objects
33408 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33411 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33414 * opts.c: Include "spellcheck.h"
33415 (struct string_fragment): New struct.
33416 (struct edit_distance_traits<const string_fragment &>): New
33418 (get_closest_sanitizer_option): New function.
33419 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33421 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33423 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33425 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33426 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33427 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33428 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33429 for initial die_offset if dwarf_split_debug_info.
33430 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33431 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33433 (output_skeleton_debug_sections): Formatting fix. Use
33434 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33435 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33437 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33439 * config/arm/cortex-a53.md: Add bypasses for
33440 cortex_a53_r2f_cvt.
33441 (cortex_a53_r2f): Only use for transfers.
33442 (cortex_a53_f2r): Likewise.
33443 (cortex_a53_r2f_cvt): Add reservation for conversions.
33444 (cortex_a53_f2r_cvt): Likewise.
33446 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33448 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33449 to all inlined functions, change static to extern.
33451 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33454 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33456 (arm_assemble_integer): Likewise.
33458 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33460 * config.gcc: Use new awk script to check CPU, FPU and architecture
33461 parameters for --with-... options.
33462 * config/arm/parsecpu.awk: New file
33463 * config/arm/arm-cpus.in: New file.
33464 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33466 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33468 * config/arm/t-arm: Update dependency rules.
33469 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33470 of processing .def files.
33471 * config/arm/genopt.sh: Deleted.
33472 * config/arm/gentune.sh: Deleted.
33473 * config/arm/arm-cores.def: Deleted.
33474 * config/arm/arm-arches.def: Deleted.
33475 * config/arm/arm-fpus.def: Deleted.
33476 * config/arm/arm-tune.md: Regenerated.
33477 * config/arm/arm-tables.opt: Regenerated.
33478 * config/arm/arm-cpu.h: New generated file.
33479 * config/arm/arm-cpu-data.h: New generated file.
33480 * config/arm/arm-cpu-cdata.h: New generated file.
33482 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33485 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33487 (input_varpool_node): Unpack dynamically_initialized bit.
33489 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33491 PR rtl-optimization/79032
33492 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33493 the alignment of the adjusted memory reference against that of MODE,
33494 instead of the alignment of the original memory reference.
33496 2017-01-11 Martin Jambor <mjambor@suse.cz>
33498 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33500 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33501 decorated functions.
33503 2017-01-11 Richard Biener <rguenther@suse.de>
33505 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33506 set range/nonnull info for PHI results. Do not set it on
33507 stmts marked for removal.
33509 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33511 * expr.c (store_field): In the bitfield case, fetch the return value
33512 from the registers before applying a single big-endian adjustment.
33513 Always do a final load for a BLKmode value not larger than a word.
33515 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33518 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33519 that we correctly handle column numbers greater than
33520 LINE_MAP_MAX_COLUMN_NUMBER.
33522 2017-01-10 Martin Sebor <msebor@redhat.com>
33524 PR middle-end/78245
33525 * gimple-ssa-sprintf.c (get_destination_size): Call
33526 {init,fini}object_sizes.
33527 * tree-object-size.c (addr_object_size): Adjust.
33528 (pass_through_call): Adjust.
33529 (pass_object_sizes::execute): Adjust.
33530 * tree-object-size.h (fini_object_sizes): Declare.
33532 2017-01-10 Martin Sebor <msebor@redhat.com>
33534 PR tree-optimization/78775
33535 * builtins.c (get_size_range): Move...
33536 * calls.c: ...to here.
33537 (alloc_max_size): Accept zero argument.
33538 (operand_signed_p): Remove.
33539 (maybe_warn_alloc_args_overflow): Call get_size_range.
33540 * calls.h (get_size_range): Declare.
33542 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33544 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33545 from TI's devices.csv file as of September 2016.
33546 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33548 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33550 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33551 * doc/invoke.texi: Likewise.
33552 * doc/md.texi: Likewise.
33553 * doc/objc.texi: Likewise.
33555 2017-01-10 Joshua Conner <joshconner@google.com>
33557 * config/arm/fuchsia-elf.h: New file.
33558 * config/fuchsia.h: New file.
33559 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33560 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33562 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33564 2016-01-10 Richard Biener <rguenther@suse.de>
33566 PR tree-optimization/79034
33567 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33568 Propagate out degenerate PHIs in the joiner.
33570 2017-01-10 Martin Liska <mliska@suse.cz>
33572 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33573 (sort_congruence_classes_by_decl_uid): Likewise.
33574 (sort_congruence_class_groups_by_decl_uid): Likewise.
33575 (sem_item_optimizer::merge_classes): Sort class, groups in these
33576 classes and members in the groups by DECL_UID of declarations.
33577 This would make merge operations stable.
33579 2017-01-10 Martin Liska <mliska@suse.cz>
33581 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33582 usage of m_classes_vec.
33583 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33584 (sem_item_optimizer::get_group_by_hash): Likewise.
33585 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33586 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33587 (sem_item_optimizer::verify_classes): Likewise.
33588 (sem_item_optimizer::process_cong_reduction): Likewise.
33589 (sem_item_optimizer::dump_cong_classes): Likewise.
33590 (sem_item_optimizer::merge_classes): Likewise.
33591 * ipa-icf.h (congruence_class_hash): Rename from
33592 congruence_class_group_hash. Remove declaration of m_classes_vec.
33594 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33596 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33597 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33598 * config.gcc: Add avx512vpopcntdqintrin.h.
33599 * config/i386/avx512vpopcntdqintrin.h: New.
33600 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33601 * config/i386/i386-builtin-types.def: Add new types.
33602 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33603 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33604 __builtin_ia32_vpopcountq_v8di_mask): New.
33605 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33606 __AVX512VPOPCNTDQ__.
33607 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33608 (PTA_AVX512VPOPCNTDQ): Define.
33609 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33610 TARGET_AVX512VPOPCNTDQ_P): Define.
33611 * config/i386/i386.opt: Add mavx512vpopcntdq.
33612 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33613 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33615 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33617 PR middle-end/77484
33618 * predict.def (PRED_CALL): Set to 67.
33620 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33622 * expr.c (store_field): In the bitfield case, if the value comes from
33623 a function call and is of an aggregate type returned in registers, do
33624 not modify the field mode; extract the value in all cases if the mode
33625 is BLKmode and the size is not larger than a word.
33627 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33630 * config/i386/cpuid.h: Fix undefined behavior.
33632 2017-01-04 Jeff Law <law@redhat.com>
33634 PR tree-optimization/79007
33635 PR tree-optimization/67955
33636 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33637 conservative for pt.null when flag_non_call_exceptions is on.
33639 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33641 PR translation/79019
33642 PR translation/79020
33643 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33644 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33645 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33647 * config/avr/avr.opt (maccumulate-args): Likewise.
33648 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33649 * common.opt (freport-bug): Likewise.
33650 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33651 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33652 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33653 translatable string.
33654 * config/i386/i386.c (function_value_32): Likewise.
33655 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33656 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33658 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33659 * common/config/msp430/msp430-common.c (msp430_handle_option):
33661 * symtab.c (symtab_node::verify_base): Likewise.
33662 * opts.c (set_debug_level): Likewise.
33663 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33664 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33665 missing whitespace to translatable strings.
33666 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33667 * config/sh/superh.h: Likewise.
33668 * config/i386/xopintrin.h: Likewise.
33669 * config/i386/znver1.md: Likewise.
33670 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33671 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33672 * double-int.h (struct double_int): Likewise.
33673 * double-int.c (div_and_round_double): Likewise.
33674 * wide-int.cc: Likewise.
33675 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33676 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33677 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33678 (crossjumps_occurred): ... this.
33679 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33682 PR tree-optimization/78899
33683 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33684 returning bool return struct loop *, NULL for failure and the new
33686 (versionable_outer_loop_p): Don't version outer loop if it has
33687 dont_vectorized bit set.
33688 (tree_if_conversion): When versioning outer loop, ensure
33689 tree_if_conversion is performed also on the inner loop of the
33690 non-vectorizable outer loop copy.
33691 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33692 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33693 prevent vectorization of it.
33694 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33695 the outer loop vectorization of the non-scalar version is attempted
33696 before vectorization of the inner loop in scalar version. If
33697 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33698 vectorization of its inner loop.
33699 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33700 has 2 inner loops, rename also on edges from bb whose single pred
33701 is outer_loop->header. Fix typo in function comment.
33703 2017-01-09 Martin Sebor <msebor@redhat.com>
33706 * asan.c (asan_emit_stack_protection): Increase local buffer size
33707 to avoid snprintf truncation warning.
33709 2017-01-09 Andrew Pinski <apinski@cavium.com>
33711 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33712 to reference thunderx2t99 for the tuning structure
33713 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33715 (thunderx2t99_extra_costs): This.
33716 * config/aarch64/aarch64-tune.md: Regenerate.
33717 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33718 (vulcan_addrcost_table): This.
33719 (vulcan_regmove_cost): Rename to ...
33720 (thunderx2t99_regmove_cost): This.
33721 (vulcan_vector_cost): Rename to ...
33722 (thunderx2t99_vector_cost): this.
33723 (vulcan_branch_cost): Rename to ...
33724 (thunderx2t99_branch_cost): This.
33725 (vulcan_tunings): Rename to ...
33726 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33727 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33729 2017-01-09 Martin Jambor <mjambor@suse.cz>
33733 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33734 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33735 (propagate_vr_accross_jump_function): Use the above function for all
33736 value range computations for pass-through jump functions and type
33737 converasion from explicit value range values.
33738 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33739 parameters from TYPE_ARG_TYPES.
33740 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33741 (ipa_write_node_info): Stream type of the actual argument.
33742 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33744 2017-01-09 Martin Liska <mliska@suse.cz>
33747 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33748 (lookup_compiler): Do not show error message with have_E.
33750 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33752 PR tree-optimization/78938
33753 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33754 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33755 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33756 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33759 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33761 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33764 2017-01-09 Richard Biener <rguenther@suse.de>
33766 PR tree-optimization/78997
33767 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33768 name condition properly.
33770 2017-01-09 Richard Biener <rguenther@suse.de>
33773 * dwarf2out.c (is_cxx): New overload with context.
33774 (is_naming_typedef_decl): Use it.
33776 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33778 * invoke.texi (Option Summary): Correct spacing in option lists
33779 and add line breaks to fix over-long lines.
33781 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33783 PR middle-end/17660
33785 * extend.texi (Common Variable Attributes): Add xref to GCC
33786 Internals manual to explain mode attribute keywords.
33788 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33791 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33792 and Preprocessor Options.
33793 (Options for Linking): Document -pthread here....
33794 (RS/6000 and PowerPC Options): ...not here.
33795 (Solaris 2 Options): ...or here.
33796 * doc/cppopts.texi: Document -pthread.
33798 2017-01-08 Martin Sebor <msebor@redhat.com>
33800 PR middle-end/77708
33801 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33802 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33803 New member functions.
33804 (format_directive): Used them.
33806 (pass_sprintf_length::handle_gimple_call): Same.
33807 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33808 to avoid truncation for any argument.
33809 (extract_affine_mul): Same.
33810 * tree.c (get_file_function_name): Same.
33812 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33814 PR middle-end/77484
33815 * predict.def (PRED_INDIR_CALL): Set to 86.
33817 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33819 PR preprocessor/54124
33820 * doc/cppopts.texi: Reformat -d subtable to list the full name
33821 of the options. Add cross-reference to the docs for the general
33822 compiler -d options.
33823 * doc/invoke.texi (Developer Options): Add cross-reference to the
33824 preprocessor-specific -d option documentation.
33826 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33828 PR preprocessor/13498
33829 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33830 redudant material, and reflect new command-line options.
33831 (System Headers): Likewise.
33833 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33835 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33836 -isystem, and -idirafter. Copy-edit.
33837 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33838 default for -ftrack-macro-expansion. Delete obsolete and
33839 badly-formatted implementation details about -fdebug-cpp output.
33840 * doc/cppwarnopts.texi: Copy-edit.
33842 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33845 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33846 that the transition from a max line width >= 1<<10 to narrower
33847 lines works correctly.
33849 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33851 * doc/options.texi (PerFunction): New.
33852 * opt-functions.awk (switch_flags): Map both Optimization and
33853 PerFunction to CL_OPTIMIZATION.
33854 * opth-gen.awk: Test for PerFunction flag along with
33856 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33857 it only when the latter is present. Skip those that don't in
33858 the hash function generator.
33859 * common.opt (fvar-tracking): Mark as PerFunction instead of
33861 (fvar-tracking-assignments): Likewise.
33862 (fvar-tracking-assignments-toggle): Likewise.
33863 (fvar-tracking-uninit): Likewise.
33865 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33867 PR translation/79018
33868 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33871 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33874 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33875 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33876 TARGET_LONG_JUMP_TABLE_OFFSETS.
33877 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33878 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33879 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33880 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33881 * config/m68k/m68k.md (tablejump expander): Likewise.
33882 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33883 TARGET_LONG_JUMP_TABLE_OFFSETS.
33884 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33885 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33887 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33888 David Holsgrove <david.holsgrove@xilinx.com>
33890 * common/config/microblaze/microblaze-common.c
33891 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33892 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33894 * config/microblaze/microblaze.c (microblaze_must_save_register)
33895 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33897 (microblaze_eh_return): New function.
33898 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33899 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33900 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33901 * config/microblaze/microblaze.md (eh_return): New pattern.
33903 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33905 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33906 GCC_DIAGNOSTIC_STRINGIFY): Define.
33908 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33910 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33912 * config/arm/arm.md (<mcrr>): New.
33914 * config/arm/arm.c (arm_arch5te): New.
33915 (arm_option_override): Set arm_arch5te.
33916 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33918 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33919 (arm_mcrr_qualifiers): ... this. New.
33920 (MRRC_QUALIFIERS): Define to...
33921 (arm_mrrc_qualifiers): ... this. New.
33922 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33924 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33925 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33926 (MRRCI, mrrc, MRRC): New.
33927 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33928 VUNSPEC_MRRC2): New.
33930 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33932 * config/arm/arm.md (<mcr>): New.
33934 * config/arm/arm.c (arm_coproc_builtin_available): Add
33935 support for mcr, mrc, mcr2 and mrc2.
33936 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33937 (arm_mcr_qualifiers): ... this. New.
33938 (MRC_QUALIFIERS): Define to ...
33939 (arm_mrc_qualifiers): ... this. New.
33940 (MCR_QUALIFIERS): Define to ...
33941 (arm_mcr_qualifiers): ... this. New.
33942 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33944 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33945 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33946 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33947 VUNSPEC_MRC2): New.
33949 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33951 * config/arm/arm.md (*ldc): New.
33955 * config/arm/arm.c (arm_coproc_builtin_available): Add
33956 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33957 (arm_coproc_ldc_stc_legitimate_address): New.
33958 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33959 'qualifier_const_pointer'.
33960 (LDC_QUALIFIERS): Define to...
33961 (arm_ldc_qualifiers): ... this. New.
33962 (STC_QUALIFIERS): Define to...
33963 (arm_stc_qualifiers): ... this. New.
33964 * config/arm/arm-protos.h
33965 (arm_coproc_ldc_stc_legitimate_address): New.
33966 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33967 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33968 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33969 stc2, stcl, stc2l): New.
33970 * config/arm/constraints.md (Uz): New.
33971 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33972 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33973 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33974 VUNSPEC_STC2L): New.
33976 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33978 * config/arm/arm.md (<cdp>): New.
33979 * config/arm/arm.c (neon_const_bounds): Rename this ...
33980 (arm_const_bounds): ... this.
33981 (arm_coproc_builtin_available): New.
33982 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33983 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33984 (CDP_QUALIFIERS): Define to...
33985 (arm_cdp_qualifiers): ... this. New.
33987 (arm_expand_builtin_args): Add case for 6 arguments.
33988 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33989 (arm_const_bounds): ... this.
33990 (arm_coproc_builtin_available): New.
33991 * config/arm/arm_acle.h (__arm_cdp): New.
33993 * config/arm/arm_acle_builtins.def (cdp): New.
33995 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33996 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33997 'arm_const_bounds'.
33998 * config/arm/types.md (coproc): New.
33999 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34000 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34001 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34002 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34004 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34006 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34007 (UBINOP_QUALIFIERS): New.
34009 (acle_builtin_data): New. Change comment.
34010 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34011 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34012 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34013 arm_acle_builtins.def.
34014 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34015 (arm_init_acle_builtins): New.
34016 (CRC32_BUILTIN): Remove.
34017 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34018 crc32cb, crc32ch and crc32cw.
34019 (arm_init_crc32_builtins): Remove.
34020 (arm_init_builtins): Use arm_init_acle_builtins rather
34021 than arm_init_crc32_builtins.
34022 (arm_expand_acle_builtin): New.
34023 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34024 * config/arm/arm_acle_builtins.def: New.
34026 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34028 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34029 (arm_builtin_datum): ... this.
34030 (arm_init_neon_builtin): Rename to ...
34031 (arm_init_builtin): ... this. Add a new parameters PREFIX
34032 and USE_SIG_IN_NAME.
34033 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34034 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34035 'arm_builtin_datum'.
34036 (arm_init_vfp_builtins): Likewise.
34037 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34038 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34039 (arm_expand_neon_args): Rename to ...
34040 (arm_expand_builtin_args): ... this. Rename builtin_arg
34041 enum values and differentiate between ARG_BUILTIN_MEMORY
34042 and ARG_BUILTIN_NEON_MEMORY.
34043 (arm_expand_neon_builtin_1): Rename to ...
34044 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34045 values, arm_expand_builtin_args and add bool parameter NEON.
34046 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34047 (arm_expand_vfp_builtin): Likewise.
34048 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34050 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34052 PR middle-end/77484
34053 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34054 * predict.c (tree_estimate_probability_bb): Reverse direction of
34055 polymorphic call predictor.
34057 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34059 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34060 (determine_pass_name_match): ...this new function and...
34061 (should_skip_pass_p): ...this new function.
34063 2017-01-06 Nathan Sidwell <nathan@acm.org>
34065 * ipa-visibility.c (function_and_variable_visibility): Reformat
34066 comments and long lines. Remove extrneous if.
34067 * symtab.c (symtab_node::make_decl_local): Fix code format.
34068 (symtab_node::set_section_for_node): Fix comment typo.
34070 2017-01-06 Martin Liska <mliska@suse.cz>
34073 * lra-constraints.c: Rename invariant to lra_invariant.
34074 * predict.c (set_even_probabilities): Initialize e to NULL.
34076 2017-01-05 Martin Sebor <msebor@redhat.com>
34078 PR tree-optimization/78910
34079 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34080 (format_integer): Correct off-by-one error in the handling
34081 of precision with negative numbers in signed conversions..
34083 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34085 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34087 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34089 PR tree-optimization/71016
34090 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34091 factor_out_conditional_conversion. Formatting fix.
34092 (factor_out_conditional_conversion): Add cond_stmt argument.
34093 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34094 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34097 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34099 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34100 read-rtl-function.o, and selftest-rtl.o.
34101 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34102 (selftest::aarch64_test_loading_full_dump): New function.
34103 (selftest::aarch64_run_selftests): New function.
34104 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34105 selftest::aarch64_run_selftests.
34106 * config/i386/i386.c
34107 (selftest::ix86_test_loading_dump_fragment_1): New function.
34108 (selftest::ix86_test_loading_call_insn): New function.
34109 (selftest::ix86_test_loading_full_dump): New function.
34110 (selftest::ix86_test_loading_unspec): New function.
34111 (selftest::ix86_run_selftests): Call the new functions.
34112 * emit-rtl.c (maybe_set_max_label_num): New function.
34113 * emit-rtl.h (maybe_set_max_label_num): New decl.
34114 * function.c (instantiate_decls): Guard call to
34115 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34116 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34118 * gensupport.c (gen_reader::gen_reader): Pass "false"
34119 for new "compact" param of rtx_reader.
34120 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34121 rather than an empty string for NULL strings.
34122 * read-md.c: Potentially include config.h rather than bconfig.h.
34123 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34124 (have_error): New global, copied from errors.c.
34125 (md_reader::read_name): Rename to...
34126 (md_reader::read_name_1): ...this, adding "out_loc" param,
34127 and converting "missing name or number" to returning false, rather
34129 (md_reader::read_name): Reimplement in terms of read_name_1.
34130 (md_reader::read_name_or_nil): New function.
34131 (md_reader::read_string): Handle "(nil)" by returning NULL.
34132 (md_reader::md_reader): Add new param "compact".
34133 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34134 (md_reader::read_file): New method.
34135 * read-md.h (md_reader::md_reader): Add new param "compact".
34136 (md_reader::read_file): New method.
34137 (md_reader::is_compact): New accessor.
34138 (md_reader::read_name): Convert return type from void to file_location.
34139 (md_reader::read_name_or_nil): New decl.
34140 (md_reader::read_name_1): New decl.
34141 (md_reader::m_compact): New field.
34142 (noop_reader::noop_reader): Pass "false" for new "compact" param
34144 (rtx_reader::rtx_reader): Add new "compact" param.
34145 (rtx_reader::read_rtx_operand): Make virtual and convert return
34146 type from void to rtx.
34147 (rtx_reader::read_until): New decl.
34148 (rtx_reader::handle_any_trailing_information): New virtual function.
34149 (rtx_reader::postprocess): New virtual function.
34150 (rtx_reader::finalize_string): New virtual function.
34151 (rtx_reader::m_in_call_function_usage): New field.
34152 (rtx_reader::m_reuse_rtx_by_id): New field.
34153 * read-rtl-function.c: New file.
34154 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34155 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34156 (selftest::verify_three_block_rtl_cfg): New decl.
34157 * read-rtl-function.h: New file.
34158 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34159 For host, include function.h, memmodel.h, and emit-rtl.h.
34160 (one_time_initialization): New function.
34161 (struct compact_insn_name): New struct.
34162 (compact_insn_names): New array.
34163 (find_code): Handle insn codes in compact dumps.
34164 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34165 (bind_subst_iter_and_attr): Likewise.
34166 (add_condition_to_string): Likewise.
34167 (add_condition_to_rtx): Likewise.
34168 (apply_attribute_uses): Likewise.
34169 (add_current_iterators): Likewise.
34170 (apply_iterators): Likewise.
34171 (initialize_iterators): Guard usage of apply_subst_iterator with
34172 #ifdef GENERATOR_FILE.
34173 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34174 (md_reader::read_mapping): Likewise.
34175 (add_define_attr_for_define_subst): Likewise.
34176 (add_define_subst_attr): Likewise.
34177 (read_subst_mapping): Likewise.
34178 (check_code_iterator): Likewise.
34179 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34181 (one_time_initialization): New function.
34182 (rtx_reader::read_until): New method.
34183 (read_flags): New function.
34184 (parse_reg_note_name): New function.
34185 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34186 Handle reuse_rtx ids.
34187 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34188 Add parsing support for RTL dumps, mirroring the special-cases in
34189 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34190 values, and calling handle_any_trailing_information.
34191 (rtx_reader::read_rtx_operand): Convert return type from void
34192 to rtx, returning return_rtx. Handle case 'e'. Call
34193 finalize_string on XSTR and XTMPL fields.
34194 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34195 "(nil)" values were omitted. Call the postprocess vfunc on the
34197 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34198 class ctor. Initialize m_in_call_function_usage. Call
34199 one_time_initialization.
34200 * rtl-tests.c (selftest::test_uncond_jump): Call
34201 set_new_first_and_last_insn.
34202 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34203 * selftest-rtl.c: New file.
34204 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34205 (selftest::get_insn_by_uid): New decl.
34206 * selftest-run-tests.c (selftest::run_tests): Call
34207 read_rtl_function_c_tests.
34208 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34209 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34212 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34214 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34215 operands in a special way. Assert that pos+len <= mode precision.
34217 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34219 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34220 3 argument Alias with unlimited for the negative form.
34221 (fno-vect-cost-model): Removed.
34223 2017-01-05 Martin Liska <mliska@suse.cz>
34225 * hsa-gen.c (gen_hsa_divmod): New function.
34226 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34228 2017-01-05 Martin Liska <mliska@suse.cz>
34231 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34234 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34236 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34237 small constant length operands.
34239 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34241 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34242 between loop iterations.
34244 2017-01-05 Martin Liska <mliska@suse.cz>
34247 * gimplify.c (gimplify_decl_expr): Compare to
34248 asan_poisoned_variables instread of checking flags.
34249 (gimplify_target_expr): Likewise.
34250 (gimplify_expr): Likewise.
34251 (gimplify_function_tree): Conditionally initialize
34252 asan_poisoned_variables.
34254 2017-01-04 Jeff Law <law@redhat.com>
34256 PR tree-optimizatin/78812
34257 * rtl.h (contains_mem_rtx_p): Prototype.
34258 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34259 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34260 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34261 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34262 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34264 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34266 * input.c (assert_char_at_range): Default-initialize actual_range.
34268 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34270 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34271 to match the caller.
34273 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34275 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34276 insns after final jump in test to emit dummy move.
34278 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34280 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34281 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34283 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34285 * multiple_target.c (create_dispatcher_calls): Init e_next.
34286 * tree-ssa-loop-split.c (split_loop): Init border.
34287 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34290 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34295 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34296 (altivec_register_operand): Do not return true if the operand
34297 contains a SUBREG mixing SImode and SFmode.
34298 (vsx_register_operand): Likewise.
34299 (vsx_reg_sfsubreg_ok): New predicate.
34300 (vfloat_operand): Do not return true if the operand contains a
34301 SUBREG mixing SImode and SFmode.
34302 (vint_operand): Likewise.
34303 (vlogical_operand): Likewise.
34304 (gpc_reg_operand): Likewise.
34305 (int_reg_operand): Likewise.
34306 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34307 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34308 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34310 (rs6000_emit_move_si_sf_subreg): New helper function.
34311 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34312 fixup SUBREGs involving SImode and SFmode.
34313 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34314 numbers for the new peephole2 optimization.
34315 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34316 the GLIBC math library that do AND/IOR/XOR operations on single
34317 precision floating point.
34318 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34319 target macros to say whether we need to avoid SUBREGs mixing
34321 (TARGET_ALLOW_SF_SUBREG): Likewise.
34322 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34323 (UNSPEC_SI_FROM_SF): Likewise.
34324 (iorxor): Change spacing.
34325 (and_ior_xor): New iterator for AND, IOR, and XOR.
34326 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34327 (movdi_from_sf_zero_ext): Likewise.
34328 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34329 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34330 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34331 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34332 (fms<mode>4): Likewise.
34333 (fnma<mode>4): Likewise.
34334 (fnms<mode>4): Likewise.
34335 (nfma<mode>4): Likewise.
34336 (nfms<mode>4): Likewise.
34338 2017-01-04 Marek Polacek <polacek@redhat.com>
34341 * doc/invoke.texi: Document -Wpointer-compare.
34343 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34345 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34348 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34349 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34352 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34355 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34356 documentation of the powerpc_popcntb_ok attribute.
34357 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34358 code to issue warning messages if a requested CPU configuration is
34359 not supported by the binary (assembler and loader) toolchain.
34360 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34361 made to define a built-in function that has been disabled.
34362 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34363 made to define a built-in function that has been disabled.
34364 (altivec_init_builtins): Add comment explaining why definition
34365 of the DST built-in functions is not preceded by an assertion
34366 check. Add assertions to prevent ICE if attempts are made to
34367 define an altivec predicate or an abs* built-in function that has
34369 (htm_init_builtins): Add comment explaining why definition of the
34370 htm built-in functions is not preceded by an assertion check.
34372 2017-01-04 Jeff Law <law@redhat.com>
34374 PR tree-optimizatin/67955
34375 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34376 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34377 the points-to solution does not include pt_null. Use DECL_PT_UID
34380 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34382 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34383 Use gen_int_mode instead of gen_lopwart for const_int operands.
34385 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34387 PR tree-optimization/71563
34388 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34389 out of range value - has low bits known to be zero.
34391 2017-01-04 Alan Modra <amodra@gmail.com>
34393 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34394 * configure: Regenerate.
34395 * config.in: Regenerate.
34397 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34400 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34401 a substring of the message, but strcmp with the whole message. Ifdef
34402 ENABLE_NLS, translate the message first using dgettext.
34404 2017-01-03 Jeff Law <law@redhat.com>
34406 PR tree-optimizatin/78856
34407 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34408 (mark_threaded_blocks): Remove code to truncate thread paths that
34409 cross multiple loop headers. Instead invalidate the cached loop
34410 iteration information and handle case of a thread path walking
34411 into an irreducible region.
34413 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34416 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34417 assertions. Add support for doing the signbit if the IEEE 128-bit
34418 floating point value is in a GPR.
34419 * config/rs6000/rs6000.md (Fsignbit): Delete.
34420 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34421 Update the length attribute if the value is in a GPR.
34422 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34423 the sign or zero extension instruction, since the value is always 0/1.
34424 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34427 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34428 extracting SImode to a GPR register so that we can generate a
34429 store, limit the vector to be in a traditional Altivec register
34430 for the vextuwrx instruction.
34432 2017-01-03 Ian Lance Taylor <iant@google.com>
34434 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34436 2017-01-03 Martin Sebor <msebor@redhat.com>
34438 PR tree-optimization/78696
34439 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34440 precision. Use MPFR for %f for greater fidelity. Correct handling
34442 (pass_sprintf_length::compute_format_length): Set width and precision
34443 specified by asrerisk to void_node for vararg functions.
34444 (try_substitute_return_value): Adjust dump output.
34446 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34448 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34450 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34452 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34453 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34454 -mlra/-mno-lra was passed to the compiler.
34456 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34458 PR rtl-optimization/65618
34459 * emit-rtl.c (try_split): Move initialization of "before" and
34460 "after" to just before the call to emit_insn_after_setloc.
34462 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34464 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34466 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34468 * dwarf2out.c (gen_enumeration_type_die): When
34469 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34471 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34473 PR tree-optimization/78965
34474 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34475 Change first argument from const call_info & to call_info &. For %n
34476 set info.nowrite to false.
34478 PR middle-end/78901
34479 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34480 possibly throwing calls.
34482 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34483 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34484 and fns handling, rather than in a separate case SSA_NAME.
34486 2017-01-02 Jeff Law <law@redhat.com>
34488 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34489 fixes for first_period and second_period variables.
34491 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34494 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34495 (*insvqi_1): New insn pattern.
34496 (*insvqi_1_mem_rex64): Ditto.
34497 (*insvqi_2): Ditto.
34498 (*insvqi_3): Rename from *insvqi.
34500 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34502 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34504 * doc/cfg.texi (Edges): Remove reference to Java.
34505 (Maintaining the CFG): Ditto.
34507 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34509 PR middle-end/77674
34510 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34511 transparent aliases.
34513 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34515 PR middle-end/77484
34516 * predict.def (PRED_CALL): Update hitrate.
34517 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34518 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34519 into direct/indirect/polymorphic variants.
34521 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34523 Update copyright years.
34525 * gcc.c (process_command): Update copyright notice dates.
34526 * gcov-dump.c (print_version): Ditto.
34527 * gcov.c (print_version): Ditto.
34528 * gcov-tool.c (print_version): Ditto.
34529 * gengtype.c (create_file): Ditto.
34530 * doc/cpp.texi: Bump @copying's copyright year.
34531 * doc/cppinternals.texi: Ditto.
34532 * doc/gcc.texi: Ditto.
34533 * doc/gccint.texi: Ditto.
34534 * doc/gcov.texi: Ditto.
34535 * doc/install.texi: Ditto.
34536 * doc/invoke.texi: Ditto.
34538 Copyright (C) 2017 Free Software Foundation, Inc.
34540 Copying and distribution of this file, with or without modification,
34541 are permitted in any medium without royalty provided the copyright
34542 notice and this notice are preserved.