1 2017-11-10 Martin Liska <mliska@suse.cz>
3 * coverage.c (coverage_init): Stream information about
4 support of has_unexecuted_blocks.
5 * doc/gcov.texi: Document that.
6 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7 * gcov.c (read_graph_file): Likewise.
8 (output_line_beginning): Fix a small issue with
11 2017-11-10 Bin Cheng <bin.cheng@arm.com>
13 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
14 reference of trivial component.
16 2017-11-10 Jakub Jelinek <jakub@redhat.com>
19 * gimple-ssa-store-merging.c
20 (pass_store_merging::terminate_all_aliasing_chains): For
21 gimple_store_p stmts also call refs_output_dependent_p.
23 PR rtl-optimization/82913
24 * compare-elim.c (try_merge_compare): Punt if def_insn is not
27 2017-11-09 Jeff Law <law@redhat.com>
29 * vr-values.h: New file with vr_values class.
30 * tree-vrp.c: Include vr-values.h
31 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
32 data objects into the vr_values class.
33 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
34 (get_value_range): Make it a member function within vr_values class.
35 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
36 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
37 (op_with_constant_singleton_value_range): Likewise.
38 (extract_range_for_var_from_comparison_expr): Likewise.
39 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
40 (extract_range_from_binary_expr): Likewise.
41 (extract_range_from_unary_expr): Likewise.
42 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
43 (check_for_binary_op_overflow, extract_range_basic): Likewise.
44 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
45 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
46 (compare_name_with_value, compare_names): Likewise.
47 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
48 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
49 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
50 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
51 (extract_range_from_phi_node): Likewise.
52 (simplify_truth_ops_using_ranges): Likewise.
53 (simplify_div_or_mod_using_ranges): Likewise.
54 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
55 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
56 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
57 (simplify_float_conversion_using_ranges): Likewise.
58 (simplify_internal_call_using_ranges): Likewise.
59 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
60 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
61 poitner via x_vr_values for calls into gimple folder.
62 (vrp_initialize_lattice): Make this the vr_values ctor.
63 (vrp_free_lattice): Make this the vr_values dtor.
64 (set_vr_value): New function.
65 (class vrp_prop): Add vr_values data member. Add various member
66 functions as well as member functions that delegate to vr_values.
67 (check_array_ref): Make a member function within vrp_prop class.
68 (search_for_addr_array, vrp_initialize): Likewise.
69 (vrp_finalize): Likewise. Revamp to avoid direct access to
70 vr_value, values_propagated, etc.
71 (check_array_bounds): Extract vrp_prop class instance pointer from
72 walk info structure. Use it to call member functions.
73 (check_all_array_refs): Make a member function within vrp_prop class.
74 Smuggle class instance pointer via walk info structure.
75 (x_vr_values): New local static.
76 (vrp_valueize): Use x_vr_values to get class instance.
77 (vr_valueize_1): Likewise.
78 (class vrp_folder): Add vr_values data member. Add various member
79 functions as well as member functions that delegate to vr_values.
80 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
81 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
82 class instance from vr_values. Use it to call member functions.
83 (vrp_dom_walker): Add vr_values data member.
84 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
85 instance via x_vr_values.
86 (identify_jump_threads): Accept vr_values as argument. Store
87 it into the walker structure.
88 (evrp_dom_walker): Add vr_values class data member. Add various
90 (evrp_dom_walker::try_find_new_range): Use vr_values data
91 member to access the memory allocator.
92 (evrp_dom_walker::before_dom_children): Store vr_values class
93 instance into the vrp_folder class.
94 (evrp_dom_walker::push_value_range): Rework to avoid direct
95 access to num_vr_values and vr_value.
96 (evrp_dom_walker::pop_value_range): Likewise.
97 (execute_early_vrp): Remove call to vrp_initialize_lattice.
98 Use vr_values to get to dump_all_value_ranges member function.
99 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
100 and simplify_cond_using_ranges_2 via vrp_prop class instance.
101 Pass vr_values class instance down to identify_jump_threads.
102 Remove call to vrp_free_lattice.
103 (debug_all_value_ranges): Remove.
105 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
106 (vrp_folder): Likewise.
108 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
109 Get it from the existing bitmap instead.
110 (vrp_intersect_ranges_1): Likewise.
112 2017-11-09 Jakub Jelinek <jakub@redhat.com>
114 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
116 (store_immediate_info::store_immediate_info): Add bitnotp argument,
118 (imm_store_chain_info::coalesce_immediate_stores): Break group
119 if bit_not_p is different.
120 (count_multiple_uses, split_group,
121 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
122 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
123 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
124 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
125 to store_immediate_info ctor.
127 2017-11-09 Jim Wilson <jimw@sifive.com>
129 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
130 (scan_prog_file): Likewise.
132 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
134 * bb-reorder.c (max_entry_frequency): Remove.
135 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
136 connect_traces, push_to_next_round_p): Remove prototypes.
137 (find_traces_1_round): Use counts only.
138 (push_to_next_round_p): Likewise.
139 (find_traces): Likewise.
140 (rotate_loop): Likewise.
141 (find_traces_1_round): Likewise.
142 (connect_traces): Likewise.
143 (edge_order): Likewise.
145 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
147 * config/arm/arm.c (output_return_instruction): Add comments to
148 indicate requirement for cmse_nonsecure_entry return to account
149 for the size of clearing instruction output here.
150 (thumb_exit): Likewise.
151 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
152 return in hardfloat mode.
154 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
156 * config/rs6000/rs6000.c (machine_function): Add a bool,
157 "toc_is_wrapped_separately".
158 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
159 if it wasn't explicitly set or unset, we are optimizing for speed, and
160 doing separate shrink-wrapping.
161 (rs6000_get_separate_components): Enable the TOC component if
162 saving the TOC register in the prologue.
163 (rs6000_components_for_bb): Handle the TOC component.
164 (rs6000_emit_prologue_components): Store the TOC register where needed.
165 (rs6000_set_handled_components): Mark TOC as handled, if handled.
166 (rs6000_emit_prologue): Don't save the TOC if that is already done.
168 2017-11-09 Martin Jambor <mjambor@suse.cz>
170 * ipa-param-manipulation.c: New file.
171 * ipa-param-manipulation.h: Likewise.
172 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
173 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
174 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
175 (ipa_parm_adjustment): Likewise.
176 (ipa_parm_adjustment_vec): Likewise.
177 (ipa_get_vector_of_formal_parms): Moved declaration to
178 ipa-param-manipulation.h.
179 (ipa_get_vector_of_formal_parm_types): Likewise.
180 (ipa_modify_formal_parameters): Likewise.
181 (ipa_modify_call_arguments): Likewise.
182 (ipa_combine_adjustments): Likewise.
183 (ipa_dump_param_adjustments): Likewise.
184 (ipa_modify_expr): Likewise.
185 (ipa_get_adjustment_candidate): Likewise.
186 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
187 ipa-param-manipulation.c.
188 (ipa_get_vector_of_formal_parm_types): Likewise.
189 (ipa_modify_formal_parameters): Likewise.
190 (ipa_modify_call_arguments): Likewise.
191 (ipa_modify_expr): Likewise.
192 (get_ssa_base_param): Likewise.
193 (ipa_get_adjustment_candidate): Likewise.
194 (index_in_adjustments_multiple_times_p): Likewise.
195 (ipa_combine_adjustments): Likewise.
196 (ipa_dump_param_adjustments): Likewise.
197 * tree-sra.c: Also include ipa-param-manipulation.h
198 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
201 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
202 Alan Hayward <alan.hayward@arm.com>
203 David Sherwood <david.sherwood@arm.com>
205 * doc/sourcebuild.texi (vect_masked_store): Document.
207 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
208 Alan Hayward <alan.hayward@arm.com>
209 David Sherwood <david.sherwood@arm.com>
211 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
213 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
214 Alan Hayward <alan.hayward@arm.com>
215 David Sherwood <david.sherwood@arm.com>
217 * doc/sourcebuild.texi (vect_variable_length): Document.
219 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
220 Alan Hayward <alan.hayward@arm.com>
221 David Sherwood <david.sherwood@arm.com>
223 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
225 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
226 Alan Hayward <alan.hayward@arm.com>
227 David Sherwood <david.sherwood@arm.com>
229 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
231 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
232 Alan Hayward <alan.hayward@arm.com>
233 David Sherwood <david.sherwood@arm.com>
235 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
236 previously undocumented selectors.
237 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
239 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
241 * doc/rtl.texi (const_vector): Say that elements can be
243 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
244 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
245 (gen_vec_duplicate): Use it instead of CONSTANT_P.
246 * optabs.c (expand_vector_broadcast): Likewise.
248 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
249 Alan Hayward <alan.hayward@arm.com>
250 David Sherwood <david.sherwood@arm.com>
252 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
253 scaled index even if the unscaled address was invalid.
254 Don't increase the complexity of using a scale in that case.
256 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
257 Alan Hayward <alan.hayward@arm.com>
258 David Sherwood <david.sherwood@arm.com>
260 * doc/rtl.texi: Rewrite the subreg rules so that they partition
261 the inner register into REGMODE_NATURAL_SIZE bytes rather than
262 UNITS_PER_WORD bytes.
263 * emit-rtl.c (validate_subreg): Divide subregs into blocks
264 based on REGMODE_NATURAL_SIZE of the inner mode.
265 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
266 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
267 * expmed.c (lowpart_bit_field_p): Divide the value up into
268 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
269 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
270 whether something is likely to occupy more than one register.
272 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
275 * ipa-inline-transform.c (update_noncloned_frequencies): Use
276 profile_count::adjust_for_ipa_scaling.
277 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
278 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
280 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
282 2017-11-09 Jakub Jelinek <jakub@redhat.com>
284 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
285 (split_group): Add total_orig and total_new arguments, estimate the
286 number of statements related to the store group without store merging
287 and with store merging.
288 (imm_store_chain_info::output_merged_store): Adjust split_group
289 callers, punt if estimated number of statements with store merging
290 is not smaller than estimated number of statements without it.
292 (handled_load): Remove has_single_use checks.
293 (pass_store_merging::process_store): Likewise.
295 2017-11-09 Richard Biener <rguenther@suse.de>
297 PR tree-optimization/82902
298 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
300 2017-11-09 Martin Liska <mliska@suse.cz>
303 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
306 2017-11-09 Martin Liska <mliska@suse.cz>
308 PR tree-optimization/82669
309 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
311 2017-11-09 Martin Liska <mliska@suse.cz>
313 PR gcov-profile/48463
314 * coverage.c (coverage_begin_function): Output also end locus
315 of a function and information whether the function is
317 * gcov-dump.c (tag_function): Parse and print the information.
318 * gcov.c (INCLUDE_MAP): Add include.
319 (INCLUDE_SET): Likewise.
320 (struct line_info): Move earlier in the source file because
321 of vector<line_info> in function_info structure.
322 (line_info::line_info): Likewise.
323 (line_info::has_block): Likewise.
324 (struct source_info): Add new member index.
325 (source_info::get_functions_at_location): New function.
326 (function_info::group_line_p): New function.
327 (output_intermediate_line): New function.
328 (output_intermediate_file): Use the mentioned function.
329 (struct function_start): New.
330 (struct function_start_pair_hash): Likewise.
331 (process_file): Add code that identifies group functions.
332 Assign lines either to global or function scope.
333 (generate_results): Skip artificial functions.
334 (find_source): Assign index for each source file.
335 (read_graph_file): Read new flag artificial and end_line.
336 (add_line_counts): Assign it either to global of function scope.
337 (accumulate_line_counts): Isolate core of the function to
338 accumulate_line_info and call it for both function and global
340 (accumulate_line_info): New function.
341 (output_line_beginning): Fix GNU coding style.
342 (print_source_line): New function.
343 (output_line_details): Likewise.
344 (output_function_details): Likewise.
345 (output_lines): Iterate both source (global) scope and function
347 (struct function_line_start_cmp): New class.
348 * doc/gcov.texi: Reflect changes in documentation.
350 2017-11-09 Jakub Jelinek <jakub@redhat.com>
353 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
354 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
355 and similarly for not instead of neg.
357 2017-11-08 Andi Kleen <ak@linux.intel.com>
359 * config/i386/i386.opt: Add -mforce-indirect-call.
360 * config/i386/predicates.md: Check for flag_force_indirect_call.
361 * doc/invoke.texi: Document -mforce-indirect-call
363 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
365 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
367 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
368 config/riscv/riscv.c (predict.h): New include.
369 (riscv_slow_unaligned_access_p): No longer static.
370 (riscv_block_move_straight): Add require.
371 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
373 2017-11-08 Jakub Jelinek <jakub@redhat.com>
376 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
377 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
378 nonimmediate_operand predicate for operand 1 instead of
381 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
383 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
385 * config/aarch64/constraints.md (Uml): New constraint.
386 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
389 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
391 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
392 of two vec_duplicates into a vec_concat.
394 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
396 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
397 Simplify vec_merge of vec_duplicate and vec_concat.
398 * config/aarch64/constraints.md (Utq): New constraint.
399 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
402 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
404 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
405 Simplify vec_merge of vec_duplicate and const_vector.
406 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
408 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
409 mode iterator. Update predicate on operand 1 to
410 handle non-const_vec constants. Delete constraints.
411 (*aarch64_combinez_be<mode>): Likewise for operand 2.
413 2017-11-08 Jakub Jelinek <jakub@redhat.com>
415 PR tree-optimization/78821
416 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
418 (store_operand_info::store_operand_info): Initialize it to false.
419 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
420 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
421 store in the group, and if chain_info is non-NULL, to ignore altogether
423 (compatible_load_p): Fail if bit_not_p does not match.
424 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
425 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
426 (pass_store_merging::process_store): Adjust
427 terminate_all_aliasing_chains calls to pass NULL in all current spots,
428 call terminate_all_aliasing_chains newly when adding a store into
429 a chain with non-NULL chain_info.
431 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
433 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
435 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
437 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
439 (aarch64_layout_frame): Initialise emit_frame_chain.
440 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
441 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
443 2017-11-08 Martin Liska <mliska@suse.cz>
445 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
446 of gimple_call_internal_p.
448 2017-11-07 Tom de Vries <tom@codesourcery.com>
450 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
452 2017-11-07 Tom de Vries <tom@codesourcery.com>
454 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
457 2017-11-08 Martin Liska <mliska@suse.cz>
460 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
462 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
464 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
466 2017-11-07 Jakub Jelinek <jakub@redhat.com>
469 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
470 (*cmp<mode>_ccz_1): New insn with $k alternative.
473 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
474 RTX_COMM_COMPARE as commutative as well.
475 (ix86_binary_operator_ok): Formatting fix.
476 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
477 *<code><mode>3<mask_name><round_saeonly_name>,
478 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
479 *<s>mul<mode>3_highpart<mask_name>,
480 *vec_widen_umult_even_v16si<mask_name>,
481 *vec_widen_umult_even_v8si<mask_name>,
482 *vec_widen_umult_even_v4si<mask_name>,
483 *vec_widen_smult_even_v16si<mask_name>,
484 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
485 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
486 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
487 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
488 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
489 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
490 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
491 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
492 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
493 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
494 ix86_binary_operator_ok. Formatting fixes.
495 (*<plusminus_insn><mode>3<mask_name><round_name>,
496 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
499 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
501 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
502 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
503 eq and ne if TARGET_ISEL.
504 (cmp): New code_iterator.
505 (UNS, UNSU_, UNSIK): New code_attrs.
506 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
507 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
509 ("eq<mode>3"): ... this.
510 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
512 ("ne<mode>3"): ... this.
514 2017-11-07 Julia Koval <julia.koval@intel.com>
517 * common/config/i386/i386-common.c
518 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
519 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
520 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
521 * config/i386/i386.opt: Ditto.
522 * config/i386/i386.c (ix86_target_string): Ditto.
523 (ix86_option_override_internal): Ditto.
524 (ix86_init_mpx_builtins): Move MPX to args2.
525 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
526 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
527 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
528 __builtin_ia32_vgf2p8affineinvqb_v32qi,
529 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
530 __builtin_ia32_vgf2p8affineinvqb_v16qi,
531 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
533 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
536 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
537 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
538 (zero-extendsidi peephole2): Remove peephole.
540 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
543 * stor-layout.c: Include attribs.h.
544 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
545 explicit lookup of "aligned" attribute.
547 2017-11-07 Andrew Waterman <andrew@sifive.com>
549 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
550 (riscv_expand_block_move): Likewise.
551 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
553 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
554 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
555 * config/riscv/riscv.c (riscv_block_move_straight): New function.
556 (riscv_adjust_block_mem): Likewise.
557 (riscv_block_move_loop): Likewise.
558 (riscv_expand_block_move): Likewise.
559 gcc/config/riscv/riscv.md (movmemsi): New pattern.
561 2017-11-07 Michael Clark <michaeljclark@mac.com>
563 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
564 (MUSL_DYNAMIC_LINKER): Likewise.
566 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
568 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
569 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
572 2017-11-07 Richard Biener <rguenther@suse.de>
574 * match.pd: Fix build.
576 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
577 Jackson Woodruff <jackson.woodruff@arm.com>
579 PR tree-optimization/71026
580 * match.pd: Canonicalize negate in division.
582 2017-11-07 Sudakshina Das <sudi.das@arm.com>
585 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
587 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
589 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
590 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
592 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
594 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
595 non-scalar integral types.
596 * match.pd (negate_expr_p): Handle MINUS_EXPR.
597 (-(A-B), -(~A)): New transformations.
599 2017-11-07 Tom de Vries <tom@codesourcery.com>
601 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
602 semicolon after "do {} while (0)".
603 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
604 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
605 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
606 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
607 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
609 2017-11-07 Tom de Vries <tom@codesourcery.com>
611 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
612 after "do {} while (0)".
613 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
614 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
615 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
616 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
617 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
619 2017-11-07 Tom de Vries <tom@codesourcery.com>
622 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
624 (arm_rtx_costs_internal): Add missing semicolon after
625 HANDLE_NARROW_SHIFT_ARITH call.
627 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
629 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
630 disable isel if it was not set explicitly.
632 2017-11-06 James Bowman <james.bowman@ftdichip.com>
634 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
636 (add_type_attribute) likewise.
638 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
640 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
641 of struct ix86_frame.
642 (ix86_initial_elimination_offset): Likewise.
643 (ix86_expand_split_stack_prologue): Likewise.
645 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
647 * tree-vrp.h (enum value_range_type): Update stale comment.
649 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
650 Alan Hayward <alan.hayward@arm.com>
651 David Sherwood <david.sherwood@arm.com>
653 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
654 (aarch64_expand_vec_perm_const): Take the number of units too.
655 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
656 (aarch64_expand_vec_perm_const): Likewise.
657 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
658 (vec_perm<mode>): Update accordingly.
660 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
661 Alan Hayward <alan.hayward@arm.com>
662 David Sherwood <david.sherwood@arm.com>
664 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
665 Take the number of units too.
666 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
667 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
668 but check for a vector mode before rather than after the call.
669 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
670 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
671 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
672 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
673 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
674 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
675 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
676 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
677 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
678 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
679 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
680 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
681 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
682 (aarch64_sqdmull2_n<mode>): Update accordingly.
684 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
685 Alan Hayward <alan.hayward@arm.com>
686 David Sherwood <david.sherwood@arm.com>
688 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
689 the number of units too.
690 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
691 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
692 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
693 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
694 (vec_store_lanesxi<mode>): Update accordingly.
696 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
697 Alan Hayward <alan.hayward@arm.com>
698 David Sherwood <david.sherwood@arm.com>
700 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
701 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
702 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
703 of units rather than the mode.
704 * config/aarch64/iterators.md (nunits): New mode attribute.
705 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
706 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
707 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
708 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
709 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
710 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
711 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
712 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
713 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
714 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
715 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
716 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
717 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
718 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
719 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
720 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
721 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
722 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
723 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
724 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
725 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
726 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
727 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
728 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
729 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
730 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
731 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
732 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
733 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
734 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
735 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
736 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
737 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
738 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
739 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
740 (aarch64_simd_vec_setv2di): Likewise.
742 2017-11-06 Carl Love <cel@us.ibm.com>
744 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
746 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
747 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
749 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
751 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
752 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
753 Add power 8 macro expansions.
754 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
755 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
756 power 8 instructions. (VSX_XXBR): Add iterator.
758 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
760 * config/arm/arm.md (predicable_short_it): Change default to "no",
761 improve documentation, remove uses that are identical to the default.
762 (enabled_for_depr_it): Rename to enabled_for_short_it.
763 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
764 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
765 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
766 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
767 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
769 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
772 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
773 float128 helper macros, which are no longer used after deleting
774 the old 'q' built-in functions, and moving the round to odd
775 built-in functions to being special built-in functions.
776 (BU_FLOAT128_2): Likewise.
777 (BU_FLOAT128_1_HW): Likewise.
778 (BU_FLOAT128_2_HW): Likewise.
779 (BU_FLOAT128_3_HW): Likewise.
780 (FABSQ): Delete old 'q' built-in functions.
781 (COPYSIGNQ): Likewise.
782 (SQRTF128_ODD): Move round to odd built-in functions to be
783 special built-in functions, so that we can handle
784 -mabi=ieeelongdouble.
785 (TRUNCF128_ODD): Likewise.
786 (ADDF128_ODD): Likewise.
787 (SUBF128_ODD): Likewise.
788 (MULF128_ODD): Likewise.
789 (DIVF128_ODD): Likewise.
790 (FMAF128_ODD): Likewise.
791 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
792 built-in names to 'f128'.
793 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
794 old 'q' built-in functions, as the machine independent code for
795 'f128' built-in functions handles this.
796 (rs6000_expand_builtin): Add expansion for float128 round to odd
797 functions, keying off on -mabi=ieeelongdouble of whether to use
798 the KFmode or TFmode variant.
799 (rs6000_init_builtins): Initialize the _Float128 round to odd
801 * doc/extend.texi (PowerPC Built-in Functions): Document the old
802 _Float128 'q' built-in functions are now mapped into the new
803 'f128' built-in functions.
805 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
807 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
808 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
810 2017-11-06 David Malcolm <dmalcolm@redhat.com>
813 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
814 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
815 * toplev.c: Include "ipa-fnsummary.h".
816 (toplev::finalize): Call ipa_fnsummary_c_finalize.
818 2017-11-06 Jakub Jelinek <jakub@redhat.com>
820 PR tree-optimization/82838
821 * gimple-ssa-store-merging.c
822 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
823 on a separate gimple_seq which is then appended to seq.
825 2017-11-06 Jeff Law <law@redhat.com>
828 * config/i386/i386.c (PROBE_INTERVAL): Remove.
829 (get_probe_interval): New functions.
830 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
831 (ix86_adjust_stack_and_probe): Likewise.
832 (output_adjust_stack_and_probe): Likewise.
833 (ix86_emit_probe_stack_range): Likewise.
834 (ix86_expand_prologue): Likewise.
836 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
838 PR tree-optimization/82816
839 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
840 if the modes of the two types are the same.
841 (convert_plusminus_to_widen): Likewise.
843 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
845 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
847 (usadv16qi): New define_expand.
848 (usadv8hi): New define_expand.
850 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
853 * ipa-inline-transform.c (update_noncloned_frequencies): Always
855 (inline_transform): Likewise.
856 * predict.c (counts_to_freqs): Remove useless conditional.
857 * profile-count.h (profile_count::apply_scale): Move sanity check.
858 * tree-inline.c (copy_bb): Always scale.
859 (copy_cfg_body): Likewise.
861 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
864 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
866 (*cmp_ite0): Add enabled_for_depr_it attribute.
867 (*cmp_ite1): Likewise.
869 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
871 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
874 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
876 * tree-vrp.c (vrp_int_const_binop): Return true on success and
877 return the value by pointer.
878 (extract_range_from_multiplicative_op_1): Update accordingly.
879 Return as soon as an operation fails.
881 2017-11-05 Tom de Vries <tom@codesourcery.com>
884 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
885 (DEF_SANITIZER_BUILTIN): ... here.
886 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
887 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
889 2017-11-05 Tom de Vries <tom@codesourcery.com>
892 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
894 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
895 ASM_OUTPUT_BEFORE_CASE_LABEL call.
896 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
898 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
899 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
900 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
902 2017-11-05 Tom de Vries <tom@codesourcery.com>
905 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
908 2017-11-04 Michael Clark <michaeljclark@mac.com>
910 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
911 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
913 (*addsi3_extended): Likewise.
914 (*addsi3_extended2): Likewise.
915 (<optab>si3): Likewise.
916 (<optab>di3): Likewise.
917 (<optab><mode>3): Likewise.
918 (<*optabe>si3_internal): Likewise.
919 (zero_extendqi<SUPERQI:mode>2): Likewise.
920 (*add<mode>hi3): Likewise.
921 (*xor<mode>hi3): Likewise.
922 (<optab>di3): Likewise.
923 (*<optab>si3_extend): Likewise.
924 (*sge<u>_<X:mode><GPR:mode>): Likewise.
925 (*slt<u>_<X:mode><GPR:mode>): Likewise.
926 (*sle<u>_<X:mode><GPR:mode>): Likewise.
928 2017-11-04 Andrew Waterman <andrew@sifive.com>
930 * config/riscv/riscv.c (riscv_option_override): Conditionally set
931 TARGET_STRICT_ALIGN based upon -mtune argument.
933 2017-11-04 Andrew Waterman <andrew@sifive.com>
935 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
937 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
939 config/i386/i386.c (choose_basereg): Use optional scratch
940 register and add assertion.
941 (x86_emit_outlined_ms2sysv_save): Use scratch register when
942 needed, and don't allocate stack.
943 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
944 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
945 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
947 2017-11-03 Jeff Law <law@redhat.com>
949 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
950 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
951 to probe at the start of a noreturn function.
953 2017-11-03 Jakub Jelinek <jakub@redhat.com>
955 PR tree-optimization/78821
956 * gimple-ssa-store-merging.c: Update the file comment.
957 (MAX_STORE_ALIAS_CHECKS): Define.
958 (struct store_operand_info): New type.
959 (store_operand_info::store_operand_info): New constructor.
960 (struct store_immediate_info): Add rhs_code and ops data members.
961 (store_immediate_info::store_immediate_info): Add rhscode, op0r
962 and op1r arguments to the ctor, initialize corresponding data members.
963 (struct merged_store_group): Add load_align_base and load_align
965 (merged_store_group::merged_store_group): Initialize them.
966 (merged_store_group::do_merge): Update them.
967 (merged_store_group::apply_stores): Pick the constant for
968 encode_tree_to_bitpos from one of the two operands, or skip
969 encode_tree_to_bitpos if neither operand is a constant.
970 (class pass_store_merging): Add process_store method decl. Remove
971 bool argument from terminate_all_aliasing_chains method decl.
972 (pass_store_merging::terminate_all_aliasing_chains): Remove
973 var_offset_p argument and corresponding handling.
974 (stmts_may_clobber_ref_p): New function.
975 (compatible_load_p): New function.
976 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
977 if there is overlap and rhs_code is not INTEGER_CST. For
978 non-overlapping stores terminate group if rhs is not mergeable.
979 (get_alias_type_for_stmts): Change first argument from
980 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
981 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
982 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
984 (get_location_for_stmts): Change first argument from
985 auto_vec<gimple *> & to vec<gimple *> &.
986 (struct split_store): Remove orig_stmts data member, add orig_stores.
987 (split_store::split_store): Create orig_stores rather than orig_stmts.
988 (find_constituent_stmts): Renamed to ...
989 (find_constituent_stores): ... this. Change second argument from
990 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
991 to info structures rather than the statements.
992 (split_group): Rename ALLOW_UNALIGNED argument to
993 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
994 it. Adjust find_constituent_stores caller.
995 (imm_store_chain_info::output_merged_store): Handle rhs_code other
996 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
997 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
998 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
999 (mem_valid_for_store_merging): New function.
1000 (handled_load): New function.
1001 (pass_store_merging::process_store): New method.
1002 (pass_store_merging::execute): Use process_store method. Adjust
1003 terminate_all_aliasing_chains caller.
1005 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1007 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1008 Return true for more constants, symbols and label references.
1009 (aarch64_valid_floating_const): Remove unused function.
1011 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1014 * config/aarch64/aarch64.c (aarch64_layout_frame):
1015 Undo forcing of LR at bottom of frame.
1017 2017-11-03 Jeff Law <law@redhat.com>
1020 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1021 for int_registers_saved.
1023 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1024 extracted from tree-ssa-dom.c.
1025 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1026 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1027 (record_equivalences_from_incoming_edge): Add additional argument
1028 to single_pred_edge_ignoring_loop_edges call.
1029 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1030 (uncprop_dom_walker::before_dom_children): Add additional argument
1031 to single_pred_edge_ignoring_loop_edges call.
1032 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1033 single_pred_edge_ignoring_loop_edges rather than open coding.
1034 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1036 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1038 * match.pd (-(-A)): Rewrite.
1040 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1042 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1043 (rs6000_emit_int_cmove): New declaration.
1044 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1045 (rs6000_emit_sISEL): Delete.
1046 (rs6000_emit_int_cmove): Make non-static.
1047 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1048 instead of rs6000_emit_sISEL.
1050 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1052 * asan.c (create_cond_insert_point): Maintain profile.
1053 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1055 * basic-block.h (struct basic_block_def): Remove frequency.
1056 (EDGE_FREQUENCY): Use to_frequency
1057 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1059 (find_traces): Update to use to_frequency.
1060 (find_traces_1_round): Likewise; use only IPA counts.
1061 (bb_to_key): Likewise.
1062 (connect_traces): Use IPA counts only.
1063 (copy_bb_p): Update to use to_frequency.
1064 (fix_up_crossing_landing_pad): Likewise.
1065 (sanitize_hot_paths): Likewise.
1066 * bt-load.c (basic_block_freq): Likewise.
1067 * cfg.c (init_flow): Set count_max to uninitialized.
1068 (check_bb_profile): Remove frequencies; check counts.
1069 (dump_bb_info): Do not dump frequencies.
1070 (update_bb_profile_for_threading): Update counts only.
1071 (scale_bbs_frequencies_int): Likewise.
1072 (MAX_SAFE_MULTIPLIER): Remove.
1073 (scale_bbs_frequencies_gcov_type): Update counts only.
1074 (scale_bbs_frequencies_profile_count): Update counts only.
1075 (scale_bbs_frequencies): Update counts only.
1076 * cfg.h (struct control_flow_graph): Add count-max.
1077 (update_bb_profile_for_threading): Update prototype.
1078 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1079 (find_many_sub_basic_blocks): Likewise.
1080 * cfgcleanup.c (try_forward_edges): Likewise.
1081 (try_crossjump_to_edge): Likewise.
1082 * cfgexpand.c (expand_gimple_cond): Likewise.
1083 (expand_gimple_tailcall): Likewise.
1084 (construct_init_block): Likewise.
1085 (construct_exit_block): Likewise.
1086 * cfghooks.c (verify_flow_info): Check consistency of counts.
1087 (dump_bb_for_graph): Do not dump frequencies.
1088 (split_block_1): Do not update frequencies.
1089 (split_edge): Do not update frequencies.
1090 (make_forwarder_block): Do not update frequencies.
1091 (duplicate_block): Do not update frequencies.
1092 (account_profile_record): Do not update frequencies.
1093 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1094 for global heuristics.
1095 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1096 (expected_loop_iterations_unbounded): Use counts only.
1097 * cfgloopmanip.c (scale_loop_profile): Simplify.
1098 (create_empty_loop_on_edge): Simplify
1100 (duplicate_loop_to_header_edge): Simplify
1101 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1102 (update_br_prob_note): Take care of removing note when profile
1104 (relink_block_chain): Do not dump frequency.
1105 (rtl_account_profile_record): Use to_frequency.
1106 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1107 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1108 (cgraph_update_edges_for_call_stmt_node): Likewise.
1109 (cgraph_edge::verify_count_and_frequency): Update.
1110 (cgraph_node::verify_node): Temporarily disable frequency verification.
1111 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1112 to_cgraph_frequency.
1113 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1114 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1116 (cgraph_node::expand_thunk): Update profile.
1117 * except.c (dw2_build_landing_pads): Do not update frequency.
1118 * final.c (compute_alignments): Use to_frequency.
1119 (dump_basic_block_info): Do not dump frequency.
1120 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1121 (dump_gimple_bb_header): Do not dump frequency.
1122 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1124 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1125 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1126 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1127 (init_before_recovery): Do not update frequency.
1128 (sched_create_recovery_edges): Do not update frequency.
1129 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1130 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1131 (ipa_cp_c_finalize): Set max_count to uninitialized.
1132 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1133 (param_change_prob): Use counts.
1134 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1136 * ipa-split.c (consider_split): Use to_frequency.
1137 (split_function): Use to_frequency.
1138 * ira-build.c (loop_compare_func): Likewise.
1139 (mark_loops_for_removal): Likewise.
1140 (mark_all_loops_for_removal): Likewise.
1141 * loop-doloop.c (doloop_modify): Do not update frequency.
1142 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1144 * lto-streamer-in.c (input_function): Update count_max.
1145 * omp-expand.c (expand_omp_taskreg): Update count_max.
1146 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1147 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1148 (maybe_hot_count_p): Use ipa counts only.
1149 (maybe_hot_bb_p): Simplify.
1150 (maybe_hot_edge_p): Simplify.
1151 (probably_never_executed): Do not take frequency argument.
1152 (probably_never_executed_bb_p): Do not pass frequency.
1153 (probably_never_executed_edge_p): Likewise.
1154 (combine_predictions_for_bb): Check that profile is nonzero.
1155 (propagate_freq): Do not set frequency.
1156 (drop_profile): Simplify.
1157 (counts_to_freqs): Simplify.
1158 (expensive_function_p): Use to_frequency.
1159 (propagate_unlikely_bbs_forward): Simplify.
1160 (determine_unlikely_bbs): Simplify.
1161 (estimate_bb_frequencies): Add hack to silence graphite issues.
1162 (compute_function_frequency): Use ipa counts.
1163 (pass_profile::execute): Update.
1164 (rebuild_frequencies): Use counts only.
1165 (force_edge_cold): Use counts only.
1166 * profile-count.c (profile_count::dump): Dump new count types.
1167 (profile_count::differs_from_p): Check compatiblity.
1168 (profile_count::to_frequency): New function.
1169 (profile_count::to_cgraph_frequency): New function.
1170 * profile-count.h (struct function): Declare.
1171 (enum profile_quality): Add profile_guessed_local and
1172 profile_guessed_global0.
1173 (class profile_proability): Decrease number of bits to 29;
1174 update from_reg_br_prob_note and to_reg_br_prob_note.
1175 (class profile_count: Update comment; decrease number of bits
1176 to 61. Check compatibility.
1177 (profile_count::compatible_p): New private member function.
1178 (profile_count::ipa_p): New member function.
1179 (profile_count::operator<): Handle global zero correctly.
1180 (profile_count::operator>): Handle global zero correctly.
1181 (profile_count::operator<=): Handle global zero correctly.
1182 (profile_count::operator>=): Handle global zero correctly.
1183 (profile_count::nonzero_p): New member function.
1184 (profile_count::force_nonzero): New member function.
1185 (profile_count::max): New member function.
1186 (profile_count::apply_scale): Handle IPA scalling.
1187 (profile_count::guessed_local): New member function.
1188 (profile_count::global0): New member function.
1189 (profile_count::ipa): New member function.
1190 (profile_count::to_frequency): Declare.
1191 (profile_count::to_cgraph_frequency): Declare.
1192 * profile.c (OVERLAP_BASE): Delete.
1193 (compute_frequency_overlap): Delete.
1194 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1195 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1196 * sched-ebb.c (rank): Use counts only.
1197 * shrink-wrap.c (handle_simple_exit): Use counts only.
1198 (try_shrink_wrapping): Use counts only.
1199 (place_prologue_for_one_component): Use counts only.
1200 * tracer.c (find_best_predecessor): Use to_frequency.
1201 (find_trace): Use to_frequency.
1202 (tail_duplicate): Use to_frequency.
1203 * trans-mem.c (expand_transaction): Do not update frequency.
1204 * tree-call-cdce.c: Do not update frequency.
1205 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1206 (gimple_merge_blocks): Likewise.
1207 (gimple_split_edge): Likewise.
1208 (gimple_duplicate_sese_region): Likewise.
1209 (gimple_duplicate_sese_tail): Likewise.
1210 (move_sese_region_to_fn): Likewise.
1211 (gimple_account_profile_record): Likewise.
1212 (insert_cond_bb): Likewise.
1213 * tree-complex.c (expand_complex_div_wide): Likewise.
1214 * tree-eh.c (lower_resx): Update profile.
1215 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1217 (initialize_cfun): Do not initialize frequencies
1218 (freqs_to_counts): Delete.
1219 (copy_cfg_body): Ignore count parameter.
1220 (copy_body): Update.
1221 (expand_call_inline): Update count_max.
1222 (optimize_inline_calls): Update count_max.
1223 (tree_function_versioning): Update count_max.
1224 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1225 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1227 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1228 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1229 (try_peel_loop): Likewise.
1230 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1232 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1233 (tree_transform_and_unroll_loop): Do not use frequencies
1234 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1235 Use reliable prediction only.
1236 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1237 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1238 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1239 probability scaling.
1240 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1241 not update frequency
1242 (any_remaining_duplicated_blocks): Likewise.
1243 (update_profile): Likewise.
1244 (estimated_freqs_path): Delete.
1245 (freqs_to_counts_path): Delete.
1246 (clear_counts_path): Delete.
1247 (ssa_fix_duplicate_block_edges): Likewise.
1248 (duplicate_thread_path): Likewise.
1249 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1250 * tree-tailcall.c (decrease_profile): Do not update frequency.
1251 (eliminate_tail_call): Likewise.
1252 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1253 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1254 (optimize_mask_stores): Likewise.
1255 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1256 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1257 (ubsan_expand_ptr_ifn): Update profile.
1258 * value-prof.c (gimple_ic): Simplify.
1259 * value-prof.h (gimple_ic): Update prototype.
1260 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1261 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1263 (want_inline_self_recursive_call_p): Likewise.
1264 (resolve_noninline_speculation): Only cummulate defined counts.
1265 (inline_small_functions): Use nonzero_p.
1266 (ipa_inline): Do not access freed node.
1268 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1270 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1271 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1273 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1275 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1276 non-legitimate address.
1278 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1280 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1281 (*lt0_<mode>di, *lt0_<mode>si): New.
1283 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1285 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1286 TARGET_PAIRED_FLOAT.
1288 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1289 Jim Wilson <jim.wilson@linaro.org>
1291 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1292 * config/aarch64/aarch64-tune.md: Regenerated.
1293 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1294 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1296 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1298 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1299 (arc_expand_prologue): Restore blink for millicode.
1300 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1302 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1305 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1306 gen_vec_duplicate after forcing the scalar into a register.
1308 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1310 * combine (try_combine): Print the insns input to try_combine to the
1313 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1316 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1317 Remove second argument from aarch64_process_target_attr call.
1318 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1320 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1322 (aarch64_handle_attr_arch): Remove second argument.
1323 (aarch64_handle_attr_cpu): Ditto.
1324 (aarch64_handle_attr_tune): Ditto.
1325 (aarch64_handle_attr_isa_flags): Ditto.
1326 (aarch64_process_one_target_attr): Ditto.
1327 (aarch64_process_target_attr): Ditto.
1328 (aarch64_option_valid_attribute_p): Remove second argument.
1329 on aarch64_process_target_attr call.
1331 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1333 * diagnostic.c: Include "selftest-diagnostic.h".
1334 (selftest::assert_location_text): New function.
1335 (selftest::test_diagnostic_get_location_text): New function.
1336 (selftest::diagnostic_c_tests): Call it.
1338 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1340 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1341 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1342 (class selftest::test_diagnostic_context): Move to...
1343 * selftest-diagnostic.c: New file.
1344 * selftest-diagnostic.h: New file.
1346 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1348 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1349 offset range for FT32B.
1350 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1351 * config/ft32/ft32.md: Add TARGET_NOPM.
1352 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1353 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1355 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1357 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1359 2017-11-02 Jeff Law <law@redhat.com>
1361 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1362 virtual keyword on FINAL OVERRIDE members.
1364 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1365 virtuals together. Add virtual destructor.
1366 (substitute_and_fold_engine): Similarly.
1368 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1370 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1372 2017-11-02 Richard Biener <rguenther@suse.de>
1374 PR tree-optimization/82795
1375 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1377 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1379 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1381 (gcc_GAS_CHECK_FEATURE): Remove.
1382 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1383 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1384 * configure: Regenerate.
1386 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1388 * config/arc/arc.c (hwloop_optimize): Account for empty
1391 2017-11-02 Richard Biener <rguenther@suse.de>
1394 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1395 Truncate ARRAY_REF index and element size.
1397 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1399 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1401 2017-11-01 Jeff Law <law@redhat.com>
1403 * tree-ssa-ccp.c (ccp_folder): New class derived from
1404 substitute_and_fold_engine.
1405 (ccp_folder::get_value): New member function.
1406 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1407 (ccp_fold_stmt): Remove prototype.
1408 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1409 * tree-ssa-copy.c (copy_folder): New class derived from
1410 substitute_and_fold_engine.
1411 (copy_folder::get_value): Renamed from get_value.
1412 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1413 * tree-vrp.c (vrp_folder): New class derived from
1414 substitute_and_fold_engine.
1415 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1416 (vrp_folder::get_value): New member function.
1417 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1418 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1419 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1420 provide a class interface to folder/substitute routines.
1421 (ssa_prop_fold_stmt_fn): Remove typedef.
1422 (ssa_prop_get_value_fn): Likewise.
1423 (subsitute_and_fold): Remove prototype.
1424 (replace_uses_in): Likewise.
1425 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1426 Renamed from replace_uses_in. Call the virtual member function
1427 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1428 (substitute_and_fold_dom_walker): Remove initialization of
1429 data member entries for calbacks. Add substitute_and_fold_engine
1430 member and initialize it.
1431 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1432 member functions for get_value, replace_phi_args_in c
1433 replace_uses_in, and fold_stmt calls.
1434 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1435 substitute_and_fold. Remove assert. Update ctor call.
1437 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1438 (ssa_prop_visit_phi_fn): Likewise.
1439 (class ssa_propagation_engine): New class to provide an interface
1441 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1443 (ssa_prop_visit_phi): Likewise.
1444 (ssa_propagation_engine::simulate_stmt): Moved into class.
1445 Call visit_phi/visit_stmt from the class rather than via
1446 file scoped static variables.
1447 (ssa_propagation_engine::simulate_block): Moved into class.
1448 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1449 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1450 set file scoped statics for the visit_stmt/visit_phi callbacks.
1451 * tree-complex.c (complex_propagate): New class derived from
1452 ssa_propagation_engine.
1453 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1454 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1455 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1457 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1458 ssa_propagation_engine.
1459 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1460 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1461 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1462 * tree-ssa-copy.c (copy_prop): New class derived from
1463 ssa_propagation_engine.
1464 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1465 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1466 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1467 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1468 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1469 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1470 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1472 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1474 PR rtl-optimization/82778
1475 PR rtl-optimization/82597
1476 * compare-elim.c (struct comparison): Add in_a_setter field.
1477 (find_comparison_dom_walker::before_dom_children): Remove killed
1478 bitmap and df_simulate_find_defs call, instead walk the defs.
1479 Compute last_setter and initialize in_a_setter. Merge definitions
1480 with first initialization for a few variables.
1481 (try_validate_parallel): Use insn_invalid_p instead of
1482 recog_memoized. Return insn rather than just the pattern.
1483 (try_merge_compare): Fix up comment. Don't uselessly test if
1484 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1486 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1489 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1490 Alan Hayward <alan.hayward@arm.com>
1491 David Sherwood <david.sherwood@arm.com>
1493 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1494 aarch64_hard_regno_nregs to get the number of registers
1497 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1498 Alan Hayward <alan.hayward@arm.com>
1499 David Sherwood <david.sherwood@arm.com>
1501 * config/aarch64/constraints.md (Upl): Rename to...
1503 * config/aarch64/aarch64.md
1504 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1507 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1508 Alan Hayward <alan.hayward@arm.com>
1509 David Sherwood <david.sherwood@arm.com>
1511 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1512 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1515 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1516 Alan Hayward <alan.hayward@arm.com>
1517 David Sherwood <david.sherwood@arm.com>
1519 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1520 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1521 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1523 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1524 of a single element.
1525 * config/aarch64/iterators.md: Add a comment above the permute
1526 unspecs to say that they are generated directly by
1527 aarch64_expand_vec_perm_const.
1528 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1530 2017-11-01 Nathan Sidwell <nathan@acm.org>
1532 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1534 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1536 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1537 and medany code models, and describe what they do.
1539 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1541 Revert accidental duplicate:
1543 * combine.c (can_change_dest_mode): Reject changes in
1544 REGMODE_NATURAL_SIZE.
1546 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1548 PR rtl-optimization/64682
1549 PR rtl-optimization/69567
1550 PR rtl-optimization/69737
1551 PR rtl-optimization/82683
1552 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1553 register mentioned in the note, drop the note, unless it came from I3,
1554 in which case it should go to I3 again.
1556 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1558 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1559 and return false if not.
1560 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1562 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1564 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1565 range is known to be empty.
1567 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1568 Alan Hayward <alan.hayward@arm.com>
1569 David Sherwood <david.sherwood@arm.com>
1571 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1572 and CONST_VECTOR_NUNITS instead of computing the number of units from
1573 the byte sizes of the vector and element.
1574 (simplify_binary_operation_1): Likewise.
1575 (simplify_const_binary_operation): Likewise.
1576 (simplify_ternary_operation): Likewise.
1578 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1579 Alan Hayward <alan.hayward@arm.com>
1580 David Sherwood <david.sherwood@arm.com>
1582 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1583 (int_mem_offset): ...this new function.
1584 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1585 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1586 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1589 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1590 Alan Hayward <alan.hayward@arm.com>
1591 David Sherwood <david.sherwood@arm.com>
1593 * lower-subreg.c (interesting_mode_p): New function.
1594 (compute_costs, find_decomposable_subregs, decompose_register)
1595 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1596 (resolve_clobber, dump_choices): Use it.
1598 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1599 Alan Hayward <alan.hayward@arm.com>
1600 David Sherwood <david.sherwood@arm.com>
1602 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1604 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1605 Alan Hayward <alan.hayward@arm.com>
1606 David Sherwood <david.sherwood@arm.com>
1608 * alias.c (find_base_value, find_base_term): Only process integer
1609 truncations. Check the precision rather than the size.
1611 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1612 Alan Hayward <alan.hayward@arm.com>
1613 David Sherwood <david.sherwood@arm.com>
1615 * machmode.h (is_narrower_int_mode): New function
1616 * optabs.c (expand_float, expand_fix): Use it.
1617 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1619 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1620 Alan Hayward <alan.hayward@arm.com>
1621 David Sherwood <david.sherwood@arm.com>
1623 * rtl.h (narrower_subreg_mode): New function.
1624 * ira-color.c (update_costs_from_allocno): Use it.
1626 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1627 Alan Hayward <alan.hayward@arm.com>
1628 David Sherwood <david.sherwood@arm.com>
1630 * optabs-query.h (convert_optab_p): New function, split out from...
1631 (convert_optab_handler): ...here.
1632 (widening_optab_handler): Delete.
1633 (find_widening_optab_handler): Remove permit_non_widening parameter.
1634 (find_widening_optab_handler_and_mode): Likewise. Provide an
1635 override that operates on mode class wrappers.
1636 * optabs-query.c (widening_optab_handler): Delete.
1637 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1638 parameter. Assert that the two modes are the same class and that
1639 the "from" mode is narrower than the "to" mode. Use
1640 convert_optab_handler instead of widening_optab_handler.
1641 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1642 instead of widening_optab_handler.
1643 * expr.c (expand_expr_real_2): Update calls to
1644 find_widening_optab_handler.
1645 * optabs.c (expand_widen_pattern_expr): Likewise.
1646 (expand_binop_directly): Take the insn_code as a parameter.
1647 (expand_binop): Only call find_widening_optab_handler for
1648 conversion optabs; use optab_handler otherwise. Update calls
1649 to find_widening_optab_handler and expand_binop_directly.
1650 Use convert_optab_handler instead of widening_optab_handler.
1651 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1652 find_widening_optab_handler and use scalar_mode rather than
1654 (convert_plusminus_to_widen): Likewise.
1656 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1657 Alan Hayward <alan.hayward@arm.com>
1658 David Sherwood <david.sherwood@arm.com>
1660 * machmode.h (fixed_size_mode): New class.
1661 * rtl.h (get_pool_mode): Return fixed_size_mode.
1662 * gengtype.c (main): Add fixed_size_mode.
1663 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1664 (get_raw_arg_mode): Likewise.
1665 * doc/tm.texi: Regenerate.
1666 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1667 * targhooks.c (default_get_reg_raw_mode): Likewise.
1668 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1669 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1670 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1671 (msp430_get_raw_result_mode): Likewise.
1672 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1673 * dbxout.c (dbxout_parms): Require fixed-size modes.
1674 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1675 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1676 * omp-low.c (lower_oacc_reductions): Likewise.
1677 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1678 (simplify_subreg): Update accordingly.
1679 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1680 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1681 that aren't fixed-size.
1682 (get_pool_mode): Return a fixed_size_mode.
1683 (output_constant_pool_2): Take a fixed_size_mode.
1685 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1686 Alan Hayward <alan.hayward@arm.com>
1687 David Sherwood <david.sherwood@arm.com>
1689 * doc/rtl.texi (vec_series): Document.
1690 (const): Say that the operand can be a vec_series.
1691 * rtl.def (VEC_SERIES): New rtx code.
1692 * rtl.h (const_vec_series_p_1): Declare.
1693 (const_vec_series_p): New function.
1694 * emit-rtl.h (gen_const_vec_series): Declare.
1695 (gen_vec_series): Likewise.
1696 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1697 (gen_vec_series): Likewise.
1698 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1699 * simplify-rtx.c (simplify_unary_operation): Handle negations
1701 (simplify_binary_operation_series): New function.
1702 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1703 (test_vector_ops_series): New function.
1704 (test_vector_ops): Call it.
1705 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1706 gen_const_vec_series.
1707 (altivec_lvsr): Likewise.
1708 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1710 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1711 Alan Hayward <alan.hayward@arm.com>
1712 David Sherwood <david.sherwood@arm.com>
1714 * doc/rtl.texi (const): Update description of address constants.
1715 Say that vector constants are allowed too.
1716 * common.md (E, F): Use CONSTANT_P instead of checking for
1718 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1719 checking for CONST_VECTOR.
1720 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1722 * expr.c (expand_expr_real_2): Check for vector modes instead
1723 of checking for CONST_VECTOR.
1724 * rtl.h (const_vec_p): New function.
1725 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1726 (unwrap_const_vec_duplicate): Handle them here too.
1728 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1729 David Malcolm <dmalcolm@redhat.com>
1730 Alan Hayward <alan.hayward@arm.com>
1731 David Sherwood <david.sherwood@arm.com>
1733 * rtl.h (vec_duplicate_p): New function.
1734 * selftest-rtl.c (assert_rtx_eq_at): New function.
1735 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1736 (assert_rtx_eq_at): Declare.
1737 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1738 * selftest-run-tests.c (selftest::run_tests): Call it.
1739 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1740 (simplify_unary_operation_1): Recursively handle vector duplicates.
1741 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1743 (simplify_subreg): Handle subregs of vector duplicates.
1744 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1745 (selftest::simplify_rtx_c_tests): New functions.
1747 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1748 Alan Hayward <alan.hayward@arm.com>
1749 David Sherwood <david.sherwood@arm.com>
1751 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1752 (gen_vec_duplicate): Likewise.
1753 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1755 (gen_const_vector): ...here.
1756 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1757 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1758 whose elements are all equal.
1759 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1760 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1761 (simplify_relational_operation): Likewise.
1762 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1764 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1765 (aarch64_expand_vector_init): Likewise.
1766 * config/arm/arm.c (neon_vdup_constant): Likewise.
1767 (neon_expand_vector_init): Likewise.
1768 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1769 (arm_block_set_unaligned_vect): Likewise.
1770 (arm_block_set_aligned_vect): Likewise.
1771 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1772 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1773 (expand_vec_perm_even_odd_pack): Likewise.
1774 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1775 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1776 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1777 gen_const_vec_duplicate.
1778 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1779 * config/mips/mips.c (mips_gen_const_int_vector): Use
1780 gen_const_vec_duplicate.
1781 (mips_expand_vector_init): Use CONST0_RTX.
1782 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1783 (define_split): Use gen_const_vec_duplicate.
1784 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1785 (define_split): Use gen_const_vec_duplicate.
1786 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1787 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1788 * config/spu/spu.c (spu_const): Likewise.
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 * combine.c (can_change_dest_mode): Reject changes in
1795 REGMODE_NATURAL_SIZE.
1797 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1799 * configure.ac (--enable-libssp): New.
1800 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1801 trying to determine target-specific default. Adjust indentation.
1802 * configure: Regenerated.
1803 * doc/install.texi (Configuration): Expand --disable-libssp
1806 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1808 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1811 2017-10-31 Martin Jambor <mjambor@suse.cz>
1814 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1816 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1818 * auto-profile.c (autofdo_source_profile::read): Use
1819 UNKNOWN_LOCATION rather than 0.
1820 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1821 (warning_at): ...this overload.
1822 (warning_at_rich_loc_n): Rename to...
1823 (warning_n): ...this overload.
1824 (error_at_rich_loc): Rename to...
1825 (error_at): ...this overload.
1826 (pedwarn_at_rich_loc): Rename to...
1827 (pedwarn): ...this overload.
1828 (permerror_at_rich_loc): Rename to...
1829 (permerror): ...this overload.
1830 (inform_at_rich_loc): Rename to...
1831 (inform): ...this overload.
1832 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1833 (diagnostic_n_impl_richloc): Rename to...
1834 (diagnostic_n_impl): ...this rich_location *-based decl.
1835 (inform_at_rich_loc): Rename to...
1836 (inform): ...this, and add an assertion.
1837 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1838 (warning_at_rich_loc): Rename to...
1839 (warning_at): ...this, and add an assertion.
1840 (warning_at_rich_loc_n): Rename to...
1841 (warning_n): ...this, and add an assertion.
1842 (warning_n): Update location_t-based implementation for removal of
1843 location_t-based diagnostic_n_impl.
1844 (pedwarn_at_rich_loc): Rename to...
1845 (pedwarn): ...this, and add an assertion.
1846 (permerror_at_rich_loc): Rename to...
1847 (permerror): ...this, and add an assertion.
1848 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1849 (error_at_rich_loc): Rename to...
1850 (error_at): ...this, and add an assertion.
1851 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1852 (driver::do_spec_on_infiles): Likewise.
1853 * substring-locations.c (format_warning_va): Update for renaming
1854 of inform_at_rich_loc.
1856 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1858 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1859 _Float<N>X built-in functions so that the variant without the
1860 "__builtin_" prefix is only enabled for the GNU C and Objective C
1861 languages when they are in non-strict ANSI/ISO mode.
1862 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1863 * target.def (floatn_builtin_p): Add a target hook to control
1864 whether _Float<N> and _Float<N>X built-in functions without the
1865 "__builtin_" prefix are enabled, and return true for C and
1866 Objective C in the default hook. Include langhooks.h in
1868 * targhooks.h (default_floatn_builtin_p): Likewise.
1869 * targhooks.c (default_floatn_builtin_p): Likewise.
1870 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1871 floatn_builtin_p target hook.
1872 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1874 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1875 Eric Botcazou <ebotcazou@adacore.com>
1877 PR rtl-optimization/81803
1878 * lra-constraints.c (curr_insn_transform): Also reload the whole
1879 register for a strict subreg no wider than a word if this is for
1880 a WORD_REGISTER_OPERATIONS target.
1882 2017-10-31 Jason Merrill <jason@redhat.com>
1884 * gdbinit.in: Skip over inlines from timevar.h.
1886 2017-10-31 Martin Liska <mliska@suse.cz>
1888 * doc/gcov.texi: Document new option.
1889 * gcov.c (print_usage): Likewise print it.
1890 (process_args): Support the argument.
1891 (format_count): New function.
1892 (format_gcov): Use the function.
1894 2017-10-31 Martin Liska <mliska@suse.cz>
1896 * gcov.c (struct name_map): do not use typedef.
1897 Define operator== and operator<.
1898 (name_search): Remove.
1899 (name_sort): Remove.
1900 (main): Do not allocate names.
1901 (process_file): Add vertical space.
1902 (generate_results): Use std::find.
1903 (release_structures): Do not release memory.
1904 (find_source): Use std::find.
1906 2017-10-31 Martin Liska <mliska@suse.cz>
1908 * gcov.c (struct line_info): Remove it's typedef.
1909 (line_info::line_info): Add proper ctor.
1910 (line_info::has_block): Do not use a typedef.
1911 (struct source_info): Do not use typedef.
1912 (circuit): Likewise.
1913 (get_cycles_count): Likewise.
1914 (output_intermediate_file): Iterate via vector iterator.
1915 (add_line_counts): Use std::vector methods.
1916 (accumulate_line_counts): Likewise.
1917 (output_lines): Likewise.
1919 2017-10-31 Martin Liska <mliska@suse.cz>
1921 * gcov.c (struct source_info): Remove typedef.
1922 (source_info::source_info): Add proper ctor.
1923 (accumulate_line_counts): Use struct, not it's typedef.
1924 (output_gcov_file): Likewise.
1925 (output_lines): Likewise.
1926 (main): Do not allocate an array.
1927 (output_intermediate_file): Use size of vector container.
1928 (process_file): Resize the vector.
1929 (generate_results): Do not preallocate, use newly added vector
1931 (release_structures): Do not release sources.
1932 (find_source): Use vector methods.
1933 (add_line_counts): Do not use typedef.
1935 2017-10-31 Martin Liska <mliska@suse.cz>
1937 * doc/gcov.texi: Document that.
1938 * gcov.c (add_line_counts): Mark lines with a non-executed
1940 (output_line_beginning): Handle such lines.
1941 (output_lines): Pass new argument.
1942 (output_intermediate_file): Print it in intermediate format.
1944 2017-10-31 Martin Liska <mliska@suse.cz>
1946 * color-macros.h: New file.
1947 * diagnostic-color.c: Factor out color related to macros to
1949 * doc/gcov.texi: Document -k option.
1950 * gcov.c (INCLUDE_STRING): Include string.h.
1951 (print_usage): Add -k option.
1952 (process_args): Parse it.
1953 (pad_count_string): New function.
1954 (output_line_beginning): Likewise.
1955 (DEFAULT_LINE_START): New macro.
1956 (output_lines): Support color output.
1958 2017-10-31 Martin Liska <mliska@suse.cz>
1960 PR gcov-profile/82633
1961 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1962 their interaction with GCOV infrastructure.
1963 * configure.ac: Add -fkeep-{inline,static}-functions to
1965 * configure: Regenerate.
1967 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1970 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1972 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1975 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1976 into a register if it does not fit into an immediate offset field.
1978 2017-10-31 Olivier Hainque <hainque@adacore.com>
1980 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1982 2017-10-31 Julia Koval <julia.koval@intel.com>
1984 * config.gcc: Add gfniintrin.h.
1985 * config/i386/gfniintrin.h: New.
1986 * config/i386/i386-builtin-types.def
1987 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1988 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1989 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1990 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1991 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1992 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1993 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1994 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1995 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1996 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1997 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1998 * config/i386/immintrin.h: Include gfniintrin.h.
1999 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2001 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2003 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2005 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2007 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2008 (arm_ashldi3_1bit): Remove pattern.
2009 (ashrdi3): Remove shift by 1 expansion.
2010 (arm_ashrdi3_1bit): Remove pattern.
2011 (lshrdi3): Remove shift by 1 expansion.
2012 (arm_lshrdi3_1bit): Remove pattern.
2013 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2014 cost of ashldi3 by 1.
2015 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2016 (<shift>di3_neon): Likewise.
2018 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2020 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2021 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2022 and (*aarch64_simd_mov<VQ:mode>).
2023 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2024 pattern alternative.
2025 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2026 attributes to match pattern alternative.
2028 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2030 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2031 * config/rs6000/emmintrin.h: New file.
2032 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2034 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2036 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2037 * (movdi_vfp_cortexa8): Remove pattern.
2039 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2041 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2043 (Specific, alpha*-dec-osf5.1): Remove.
2044 (Specific, mips-sgi-irix5): Remove.
2045 (Specific, mips-sgi-irix6): Remove.
2047 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2050 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2051 arguments to clear_bit_region_be.
2053 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2055 * gimplify.c: Include memmodel.h.
2057 2017-10-30 Martin Jambor <mjambor@suse.cz>
2059 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2060 condition whether loop should be executed at all.
2062 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2064 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2065 gimple folding of vec_madd() intrinsics.
2066 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2067 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2068 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2070 2017-10-30 Richard Biener <rguenther@suse.de>
2072 PR tree-optimization/82762
2074 2017-10-23 Richard Biener <rguenther@suse.de>
2076 PR tree-optimization/82129
2078 2017-08-01 Richard Biener <rguenther@suse.de>
2080 PR tree-optimization/81181
2081 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2082 (compute_antic): ... end of iteration here.
2084 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2086 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2087 -std=iso9899:2017 and -std=gnu17.
2088 * doc/standards.texi (C Language): Document C17 support.
2089 * doc/cpp.texi (Overview): Mention -std=c17.
2090 (Standard Predefined Macros): Document C11 and C17 values of
2091 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2092 * doc/extend.texi (Inline): Do not list individual options for
2093 standards newer than C99.
2094 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2096 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2099 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2101 * asan.c (asan_finish_file): Align asan globals array by shadow
2104 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2107 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2108 (struct store_immediate_info): Add bitregion_start and bitregion_end
2110 (store_immediate_info::store_immediate_info): Add brs and bre
2111 arguments and initialize bitregion_{start,end} from those.
2112 (struct merged_store_group): Add bitregion_start, bitregion_end,
2113 align_base and mask fields. Drop unnecessary struct keyword from
2114 struct store_immediate_info. Add do_merge method.
2115 (clear_bit_region_be): Use memset instead of loop storing zeros.
2116 (merged_store_group::do_merge): New method.
2117 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2118 stores as long as the surrounding bitregions have no gaps.
2119 (merged_store_group::merge_overlapping): Use do_merge.
2120 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2121 is byte aligned, rather than requiring that start and width are
2122 byte aligned. Drop unnecessary struct keyword from
2123 struct store_immediate_info. Allocate and populate also mask array.
2124 Make start of the arrays relative to bitregion_start rather than
2125 start and size them according to bitregion_{end,start} difference.
2126 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2127 struct store_immediate_info.
2128 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2129 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2130 struct keyword from struct store_immediate_info.
2131 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2132 between stores as long as the surrounding bitregions have no gaps.
2134 (struct split_store): Add orig non-static data member.
2135 (split_store::split_store): Initialize orig to false.
2136 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2137 if there is exactly a single original stmt. Change stmts argument
2138 to pointer from reference, if NULL, don't push anything to it. Add
2139 first argument, use it to optimize skipping over orig stmts that
2140 are known to be before bitpos already. Simplify.
2141 (split_group): Return unsigned int count how many stores are or
2142 would be needed rather than a bool. Add allow_unaligned argument.
2143 Change split_stores argument from reference to pointer, if NULL,
2144 only do a dry run computing how many stores would be produced.
2145 Rewritten algorithm to use both alignment and misalign if
2146 !allow_unaligned and handle bitfield stores with gaps.
2147 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2148 from bitregion_start instead of start. Compute allow_unaligned
2149 here, if true, do 2 split_group dry runs to compute which one
2150 produces fewer stores and prefer aligned if equal. Punt if
2151 new count is bigger or equal than original before emitting any
2152 statements, rather than during that. Remove no longer needed
2153 new_ssa_names tracking. Replace num_stmts with
2154 split_stores.length (). Use 32-bit stack allocated entries
2155 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2156 if possible. Handle bitfields with gaps.
2157 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2158 Compute bitregion_{start,end} for the stores and construct
2159 store_immediate_info with that. Formatting fixes.
2161 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2164 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2165 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2167 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2169 * gimplify.c: Include tm_p.h.
2171 * common.opt (gcoff): Re-add as ignored option.
2172 (gcoff1, gcoff2, gcoff3): Likewise.
2174 * Makefile.in (OBJS): Delete sdbout.o.
2175 (GTFILES): Delete $(srcdir)/sdbout.c.
2176 * debug.h: Delete sdb_debug_hooks.
2177 * final.c: Delete sdbout.h include.
2178 (final_scan_insn): Delete SDB_DEBUG check.
2179 (rest_of_clean_state): Likewise.
2180 * output.h: Delete sdb_begin_function_line.
2183 * toplev.c: Delete sdbout.h include.
2184 (process_options): Delete SDB_DEBUG check.
2185 * tree-core.h (tree_type_common): Delete pointer field of
2187 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2188 TYPE_SYMTAB_POINTER.
2189 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2190 (TYPE_SYMTAB_IS_POINTER): Delete.
2191 (TYPE_SYMTAB_IS_DIE): Renumber.
2192 * xcoffout.c: Refer to former sdbout.c file.
2193 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2195 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2196 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2197 (Debugging Options): Delete -gcoff.
2198 (-gstabs): Delete SDB reference.
2200 (-gcoff@var{level}): Delete.
2201 * doc/passes.texi (Debugging information output): Delete SDB and
2202 sdbout.c references.
2203 * doc/tm.texi: Regenerate.
2204 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2205 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2207 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2208 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2209 (SDB_DEBUGGING_INFO): Delete.
2210 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2211 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2212 * target.def (output_source_filename): Delete COFF reference.
2214 * common.opt (gcoff): Delete.
2215 (gxcoff+): Update Negative chain.
2216 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2218 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2219 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2220 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2222 (expand_function_start): Change sdb reference to past tense.
2223 (expand_function_end): Change sdb reference to past tense.
2224 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2225 * opts.c (debug_type_names): Delete coff entry.
2226 (common_handle_option): Delete OPT_gcoff case.
2227 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2229 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2230 * config/cris/cris.h: Delete SDB reference in comment.
2231 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2232 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2233 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2234 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2236 (ix86_expand_prologue): Likewise.
2237 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2238 * config/ia64/ia64.h: Likewise.
2239 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2240 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2242 * config/mmix/mmix.h: Likewise.
2243 * config/nds32/nds32.c: Likewise.
2244 * config/stormy/storym16.h: Likewise.
2245 * config/visium/visium.h: Likewise.
2246 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2248 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2250 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2251 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2252 FRAME_POINTER_REGNUM point at high end of local var area.
2254 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2256 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2257 Move comment around. Do not reset best_edge for a copiable
2258 destination if the copy would cause a partition change.
2259 (better_edge_p): Remove redundant check.
2261 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2263 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2265 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2267 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2268 for math functions that have _Float<N> and _Float<N>X variants.
2269 (mathfn_built_in_2): Add support for math functions that have
2270 _Float<N> and _Float<N>X variants.
2271 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2272 (expand_builtin_mathfn_ternary): Add support for fma with
2273 _Float<N> and _Float<N>X variants.
2274 (expand_builtin): Likewise.
2275 (fold_builtin_3): Likewise.
2276 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2277 create math function _Float<N> and _Float<N>X variants as external
2279 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2280 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2281 the __builtin_ prefix and if not strict ansi, without the prefix.
2282 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2283 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2284 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2285 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2286 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2287 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2288 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2289 function signatures for fma _Float<N> and _Float<N>X variants.
2290 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2291 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2292 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2293 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2294 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2295 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2296 * gencfn-macros.c (print_case_cfn): Add support for math functions
2297 that have _Float<N> and _Float<N>X variants.
2298 (print_define_operator_list): Likewise.
2299 (fltfn_suffixes): Likewise.
2301 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2302 for math functions that have _Float<N> and _Float<N>X variants.
2303 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2304 and _Float<N>X variants.
2305 (COPYSIGN): Likewise.
2308 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2309 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2311 (integer_valued_read_call_p): Likewise.
2312 * fold-const-call.c (fold_const_call_ss): Likewise.
2313 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2314 _Float<N> and _Float<N>X variants.
2315 (fold_const_call_ssss): Add support for fma _Float<N> and
2316 _Float<N>X variants.
2317 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2318 support for copysign and fma _Float<N> and _Float<N>X variants.
2319 (backprop::process_builtin_call_use): Likewise.
2320 * tree-call-cdce.c (can_test_argument_range); Add support for
2321 sqrt _Float<N> and _Float<N>X variants.
2322 (edom_only_function): Likewise.
2323 (get_no_error_domain): Likewise.
2324 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2325 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2326 copysign _Float<N> and _Float<N>X variants.
2327 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2328 handled by machine independent code.
2329 (FMAF128): Likewise.
2330 * doc/cpp.texi (Common Predefined Macros): Document defining
2331 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2332 fma _Float<N> and _Float<N>X variants.
2334 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2337 * config/i386/i386-modes.def (CCFPU): Remove definition.
2338 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2339 (ix86_cc_modes_compatible): Ditto.
2340 (ix86_expand_carry_flag_compare): Ditto.
2341 (ix86_expand_int_movcc): Ditto.
2342 (ix86_expand_int_addcc): Ditto.
2343 (ix86_reverse_condition): Ditto.
2344 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2345 Return true/false for unordered/ordered fp comparisons.
2346 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2347 (ix86_prepare_fp_compare_args): Update for rename.
2348 (ix86_expand_fp_compare): Update for rename. Generate unordered
2349 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2350 (ix86_expand_sse_compare_and_jump): Ditto.
2351 * config/i386/predicates.md (fcmov_comparison_operator):
2352 Remove CCFPU mode handling.
2353 (ix86_comparison_operator): Ditto.
2354 (ix86_carry_flag_operator): Ditto.
2355 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2356 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2357 (*cmpu<mode>_cc_i387): Ditto.
2358 (FPCMP): Remove mode iterator.
2359 (unord): Remove mode attribute.
2360 (unord_subst): New define_subst transformation
2361 (unord): New define_subst attribute.
2363 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2364 (*cmpi<unord>xf_i387): Ditto.
2365 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2366 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2367 using unord_subst transformation.
2368 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2369 (round_saeonly): Also handle CCFP mode.
2370 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2371 Remove UNSPEC_SAHF unspec handling.
2373 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2375 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2377 2017-10-27 Jeff Law <law@redhat.com>
2379 * gimple-ssa-sprintf.c: Include domwalk.h.
2380 (class sprintf_dom_walker): New class, derived from dom_walker.
2381 (sprintf_dom_walker::before_dom_children): New function.
2382 (struct call_info): Moved into sprintf_dom_walker class
2383 (compute_formath_length, handle_gimple_call): Likewise.
2384 (sprintf_length::execute): Call the dominator walker rather
2385 than walking the statements.
2387 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2388 gimple statement locations.
2389 (check_array_bounds): Corresponding changes. Get the statement's
2390 location directly from wi->stmt.
2392 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2395 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2397 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2399 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2400 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2402 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2405 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2406 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2407 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2408 maybe_get_pool_constant.
2409 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2412 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2414 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2415 2.26 caveat. Update gas and gld versions.
2416 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2419 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2421 * cgraph.h (set_malloc_flag): Declare.
2422 * cgraph.c (set_malloc_flag_1): New function.
2423 (set_malloc_flag): Likewise.
2424 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2425 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2427 (read_ipa_call_summary): Add support for reading is_return_callee.
2428 (write_ipa_call_summary): Stream is_return_callee.
2429 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2430 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2431 ipa-prop.h, ipa-fnsummary.h.
2432 (pure_const_names): Change to static.
2433 (malloc_state_e): Define.
2434 (malloc_state_names): Define.
2435 (funct_state_d): Add field malloc_state.
2436 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2437 (check_retval_uses): New function.
2438 (malloc_candidate_p): Likewise.
2439 (analyze_function): Add support for malloc attribute.
2440 (pure_const_write_summary): Stream malloc_state.
2441 (pure_const_read_summary): Add support for reading malloc_state.
2442 (dump_malloc_lattice): New function.
2443 (propagate_malloc): New function.
2444 (warn_function_malloc): New function.
2445 (ipa_pure_const::execute): Call propagate_malloc and
2446 ipa_free_fn_summary.
2447 (pass_local_pure_const::execute): Add support for malloc attribute.
2448 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2449 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2451 2017-10-27 Martin Liska <mliska@suse.cz>
2453 PR gcov-profile/82457
2454 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2455 for fork-like functions to be properly instrumented.
2457 2017-10-27 Richard Biener <rguenther@suse.de>
2460 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2461 info when we redirected EH.
2463 2017-10-26 Michael Collison <michael.collison@arm.com>
2465 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2467 (<optab>_trunchf<GPI:mode>2: New pattern.
2468 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2469 * config/aarch64/iterators.md (wv): New mode attribute.
2470 (vf, VF): New mode attributes.
2471 (vgp, VGP): New mode attributes.
2472 (s): Update attribute with SImode and DImode prefixes.
2474 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2476 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2477 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2478 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2479 (nios2_option_overide): Initialize it. Don't allow R0-relative
2480 addressing with PIC.
2481 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2482 (nios2_symbolic_constant_p): Likewise.
2483 (nios2_legitimate_address_p): Likewise.
2484 (nios2_r0rel_section_name_p): New.
2485 (nios2_symbol_ref_in_r0rel_data_p): New.
2486 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2487 (r0rel_constant_p): New.
2488 (nios2_print_operand_address): Handle r0rel_constant_p.
2489 (nios2_cdx_narrow_form_p): Likewise.
2490 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2491 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2492 (Nios II Options): Document -mr0rel-sec.
2494 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2496 * config/nios2/nios2.c: Include xregex.h.
2497 (nios2_gprel_sec_regex): New.
2498 (nios2_option_overide): Initialize it. Don't allow GP-relative
2499 addressing with PIC.
2500 (nios2_small_section_name_p): Check for regex match.
2501 * config/nios2/nios2.opt (mgprel-sec=): New option.
2502 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2503 (Nios II Options): Document -mgprel-sec.
2505 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2507 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2509 2017-10-26 Tom de Vries <tom@codesourcery.com>
2511 PR tree-optimization/82707
2512 * gimple.c (gimple_copy): Fix unsharing of
2513 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2515 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2517 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2518 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2519 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2520 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2521 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2522 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2523 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2524 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2525 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2526 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2527 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2528 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2529 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2530 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2531 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2532 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2533 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2535 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2537 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2539 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2540 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2541 warning to rs6000.c. Remove the Undocumented flag, since it has
2543 (-mabi=ibmlongdouble): Likewise.
2544 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2545 already set, set the default format for long double.
2546 (rs6000_debug_reg_global): Print whether long double is IBM or
2548 (rs6000_option_override_internal): Rework setting long double
2549 format. Only warn if the user is changing the long double default
2550 and they did not use -Wno-psabi.
2551 * doc/invoke.texi (PowerPC options): Update the documentation for
2552 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2554 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2555 Alan Hayward <alan.hayward@arm.com>
2556 David Sherwood <david.sherwood@arm.com>
2558 * rtl.h (wider_subreg_mode): New function.
2559 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2560 rather than an unsigned int *.
2561 * ira-color.c (regno_max_ref_width): Replace with...
2562 (regno_max_ref_mode): ...this new variable.
2563 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2564 Use wider_subreg_mode.
2565 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2566 rather than an unsigned int *.
2567 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2568 (process_alt_operands): Likewise.
2569 (invariant_p): Likewise.
2570 * lra-spills.c (assign_mem_slot): Likewise.
2571 (add_pseudo_to_slot): Likewise.
2572 * lra.c (collect_non_operand_hard_regs): Likewise.
2573 (add_regs_to_insn_regno_info): Likewise.
2574 * reload1.c (regno_max_ref_width): Replace with...
2575 (regno_max_ref_mode): ...this new variable.
2576 (reload): Update accordingly. Update call to
2577 ira_sort_regnos_for_alter_reg.
2578 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2579 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2580 (scan_paradoxical_subregs): Likewise.
2582 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2584 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2585 (aarch64_frame): Add emit_frame_chain boolean.
2586 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2587 Move eh_return case to aarch64_layout_frame.
2588 (aarch64_layout_frame): Initialize emit_frame_chain.
2589 (aarch64_expand_prologue): Use emit_frame_chain.
2591 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2593 * config/aarch64/aarch64.c (aarch64_layout_frame):
2594 Ensure LR is always stored at the bottom of the callee-saves.
2595 Remove rarely used frame layout which saves callee-saves at top of
2596 frame, so the store of LR can be used as a valid probe in all cases.
2598 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2600 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2601 Improve unaligned TImode/TFmode base/offset split.
2603 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2604 Alan Hayward <alan.hayward@arm.com>
2605 David Sherwood <david.sherwood@arm.com>
2607 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2608 * combine.c (find_single_use_1): Likewise.
2609 (expand_field_assignment): Likewise.
2610 (move_deaths): Likewise.
2611 * lra-constraints.c (simplify_operand_subreg): Likewise.
2612 (curr_insn_transform): Likewise.
2613 * lra.c (collect_non_operand_hard_regs): Likewise.
2614 (add_regs_to_insn_regno_info): Likewise.
2615 * rtlanal.c (reg_referenced_p): Likewise.
2616 (covers_regno_no_parallel_p): Likewise.
2618 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2620 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2621 Don't print any bits outside the precision of the value.
2622 * wide-int.cc (test_printing): Add some new tests.
2624 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2626 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2627 supports -xbrace_comment option.
2628 * configure: Regenerate.
2629 * config.in: Regenerate.
2630 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2631 (ASM_CPU_SPEC): Use it.
2633 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2635 * target.def (static_rtx_alignment): New hook.
2636 * targhooks.h (default_static_rtx_alignment): Declare.
2637 * targhooks.c (default_static_rtx_alignment): New function.
2638 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2639 * doc/tm.texi: Regenerate.
2640 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2641 instead of targetm.constant_alignment. Remove call to
2643 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2644 (cris_preferred_mininum_alignment): New function, split out from...
2645 (cris_constant_alignment): ...here.
2646 (cris_static_rtx_alignment): New function.
2647 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2649 (ix86_constant_alignment): ...here.
2650 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2651 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2652 (mmix_static_rtx_alignment): New function.
2653 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2654 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2656 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2659 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2660 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2662 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2664 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2665 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2666 mark as a sign-extending load.
2667 (local_pic_load_u): Define.
2669 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2672 * fold-const.c (operand_equal_for_comparison_p): Also return true
2673 if ARG0 is a simple variant of ARG1 with narrower precision.
2674 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2676 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2678 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2680 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2681 scatter_static, scatter_per_elt.
2682 * x86-tune-costs.h: Add new cost entries.
2684 2017-10-25 Richard Biener <rguenther@suse.de>
2686 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2687 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2688 class pass_fre): Move to ...
2689 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2690 class pass_fre): ... here and adjust for statistics.
2692 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2695 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2696 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2697 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2698 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2701 2017-10-25 Richard Biener <rguenther@suse.de>
2703 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2704 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2705 eliminate_push_avail, eliminate_insert): Move inside...
2706 (class eliminate_dom_walker): ... this class in preparation
2708 (fini_eliminate): Remove by merging with ...
2709 (eliminate): ... this function. Adjust for class changes.
2710 (pass_pre::execute): Remove fini_eliminate call.
2711 (pass_fre::execute): Likewise.
2713 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2716 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2717 (VPERMI2, VPERMI2I): New mode iterators.
2718 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2719 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2721 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2722 mode iterator. Remove 3 old define_insn patterns.
2723 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2724 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2725 VPERMI2 mode iterator, remove the other two expanders.
2726 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2727 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2728 instructions, remove the other two patterns.
2729 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2730 mode iterator, remove the other two patterns.
2731 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2732 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2733 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2734 and adjust argument order accordingly.
2735 (ix86_expand_vec_perm): Adjust caller.
2736 (expand_vec_perm_1): Likewise.
2737 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2738 (expand_vec_perm_vpermt2_vpshub2): ... this.
2739 (ix86_expand_vec_perm_const_1): Adjust caller.
2740 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2743 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2744 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2745 (vec_shl_<mode>): Remove unused expander.
2746 (avx512bw_<shift_insn><mode>3): New define_insn.
2747 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2748 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2750 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2753 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2756 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2758 PR rtl-optimization/82396
2759 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2760 (autopref_multipass_init): Simplify initialization.
2761 (autopref_rank_data): Simplify sort order.
2762 * gcc/sched-int.h (autopref_multipass_data_): Remove
2763 multi_mem_insn_p, min_offset and max_offset.
2765 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2768 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2769 Check special value of flag_omit_frame_pointer.
2770 (aarch64_can_eliminate): Likewise.
2771 (aarch64_override_options_after_change_1): Simplify handling of
2772 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2774 2017-10-24 Richard Biener <rguenther@suse.de>
2776 PR tree-optimization/82697
2777 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2778 zero for conditional load and unconditional store.
2780 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2782 * doc/install.texi: Document bootstrap-cet.
2784 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2787 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2788 ENDBR instruction at function entrance if function is only
2791 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2794 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2795 patterns to better describe from which operation the CF is computed.
2796 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2797 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2798 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2799 is 0, use _0 suffixed expanders instead of emitting a comparison
2802 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2804 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2805 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2807 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2810 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2811 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2812 * loop-iv.c (iv_get_reaching_def): Likewise.
2813 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2814 variable is promoted and the partition contains undefined values.
2816 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2818 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2820 (nios2_address_cost): Define.
2821 (nios2_legitimize_address): Recognize (exp + constant) directly.
2822 (TARGET_ADDRESS_COST): Define.
2824 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2826 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2827 (nios2_symbolic_memory_operand_p): Declare.
2828 (nios2_split_large_constant): Declare.
2829 (nios2_split_symbolic_memory_operand): Declare.
2830 * config/nios2/nios2.c: Adjust includes.
2831 (nios2_symbolic_constant_allowed): New.
2832 (nios2_symbolic_constant_p): New.
2833 (nios2_plus_symbolic_constant_p): New.
2834 (nios2_valid_addr_expr_p): Recognize addresses involving
2836 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2837 (nios2_symbolic_memory_operand_p): New.
2838 (nios2_large_constant_p): New.
2839 (nios2_split_large_constant): New.
2840 (nios2_split_plus_large_constant): New.
2841 (nios2_split_symbolic_memory_operand): New.
2842 (nios2_legitimize_address): Code refactoring. Handle addresses
2843 involving symbolic constants.
2844 (nios2_emit_move_sequence): Likewise.
2845 (nios2_print_operand): Improve error output.
2846 (nios2_print_operand_address): Handle LO_SUM.
2847 (nios2_cdx_narrow_form_p): Likewise.
2848 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2849 operands involving symbolic constants.
2850 (movhi_internal, movsi_internal): Likewise.
2851 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2852 (extendhisi2, extendqi<mode>2): Likewise.
2854 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2856 * tree-pass.h (PROP_rtl_split_insns): Define.
2857 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2859 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2861 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2863 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2866 * target.def (const_not_ok_for_debug_p): Default to
2867 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2868 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2869 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2870 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2871 which targetm.const_not_ok_for_debug_p returned true.
2872 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2874 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2876 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2877 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2878 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2880 (ix86_const_not_ok_for_debug_p): New function.
2881 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2882 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2884 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2887 * system.h: Conditionally include "unique-ptr.h" if
2888 INCLUDE_UNIQUE_PTR is defined.
2889 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2890 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2892 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2894 * config/rl78/rl78.md: New define_expand "subdi3".
2896 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2899 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2900 DF_REF_INSN if DF_REF_INSN_INFO is false.
2902 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2904 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2905 xmm_move instead of sse_move.
2906 (sse_store_index): New function.
2907 (ix86_register_move_cost): Be more sensible about mismatch stall;
2908 model AVX moves correctly; make difference between sse->integer and
2910 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2911 moves; make difference between SSE and AVX.
2912 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2913 and zmm_move. Increase size of sse load and store tables;
2914 add unaligned load and store tables; add ssemmx_to_integer.
2915 * x86-tune-costs.h: Update all entries according to real
2916 move latencies from Agner Fog's manual and chip documentation.
2918 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2921 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2922 * config/i386/constraints.md (Wf): New constraint.
2923 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2924 (cmp<dwi>_doubleword): Removed.
2925 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2926 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2927 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2928 expand with cmp<dwi>_doubleword. For LTU and GEU use
2929 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2931 * common.opt (gcolumn-info): Enable by default.
2932 * doc/invoke.texi (gcolumn-info): Document new default.
2934 2017-10-23 Richard Biener <rguenther@suse.de>
2936 PR tree-optimization/82672
2937 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2938 Fold the stmt if we propagated into it.
2940 2017-10-23 Richard Biener <rguenther@suse.de>
2942 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2943 (bitmap_remove_expr_from_set): ... this. All callers call this
2944 for non-constant values.
2945 (bitmap_set_subtract): Rename to...
2946 (bitmap_set_subtract_expressions): ... this. Adjust and
2948 (bitmap_set_contains_value): Remove superfluous check.
2949 (bitmap_set_replace_value): Inline into single caller ...
2950 (bitmap_value_replace_in_set): ... here and simplify.
2951 (dependent_clean): Merge into ...
2952 (clean): ... this using an overload. Adjust.
2953 (prune_clobbered_mems): Adjust.
2954 (compute_antic_aux): Likewise.
2955 (compute_partial_antic_aux): Likewise.
2957 2017-10-23 Richard Biener <rguenther@suse.de>
2959 PR tree-optimization/82129
2961 2017-08-01 Richard Biener <rguenther@suse.de>
2963 PR tree-optimization/81181
2964 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2965 (compute_antic): ... end of iteration here.
2967 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2969 * target.def (starting_frame_offset): New hook.
2970 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2971 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2972 * doc/tm.texi.in: Regenerate.
2973 * hooks.h (hook_hwi_void_0): Declare.
2974 * hooks.c (hook_hwi_void_0): New function.
2975 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2976 STARTING_FRAME_OFFSET.
2977 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2978 * reload1.c (reload): Likewise.
2979 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2980 instead of STARTING_FRAME_OFFSET.
2981 * function.c (try_fit_stack_local): Likewise.
2982 (assign_stack_local_1): Likewise
2983 (instantiate_virtual_regs): Likewise.
2984 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2985 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2986 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2987 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2988 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2989 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2990 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2991 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2992 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2993 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2994 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2995 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2996 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2997 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2998 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2999 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3000 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3001 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3002 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3003 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3004 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3005 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3006 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3007 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3008 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3009 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3010 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3011 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3012 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3013 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3014 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3015 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3016 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3017 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3018 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3019 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3020 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3021 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3022 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3023 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3024 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3025 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3026 return a HOST_WIDE_INT.
3027 (avr_builtin_setjmp_frame_value): Use it instead of
3028 STARTING_FRAME_OFFSET.
3029 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3030 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3031 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3033 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3034 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3035 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3036 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3037 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3038 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3039 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3040 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3041 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3042 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3043 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3044 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3046 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3047 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3048 * config/mips/mips.c (mips_compute_frame_info): Refer to
3049 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3050 (mips_starting_frame_offset): New function.
3051 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3052 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3053 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3054 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3055 and return a HOST_WIDE_INT.
3056 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3057 (mmix_initial_elimination_offset): Refer to
3058 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3059 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3060 * config/pa/pa.c (pa_starting_frame_offset): New function.
3061 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3062 (pa_expand_prologue): Likewise.
3063 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3064 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3065 !FRAME_GROWS_DOWNWARD handling to...
3066 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3067 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3068 !FRAME_GROWS_DOWNWARD handling to...
3069 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3070 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3071 !FRAME_GROWS_DOWNWARD handling to...
3072 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3073 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3075 (rs6000_starting_frame_offset): New function.
3076 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3077 !FRAME_GROWS_DOWNWARD handling to...
3078 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3079 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3080 !FRAME_GROWS_DOWNWARD handling to...
3081 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3082 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3083 !FRAME_GROWS_DOWNWARD handling to...
3084 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3085 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3086 (rs6000_starting_frame_offset): New function.
3087 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3088 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3089 * config/vax/vax.c (vax_starting_frame_offset): New function.
3090 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3091 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3092 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3093 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3094 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3095 * system.h (STARTING_FRAME_OFFSET): Poison.
3097 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3099 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3100 SCALAR_TYPE_MODE instead of TYPE_MODE.
3102 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3103 Alan Hayward <alan.hayward@arm.com>
3104 David Sherwood <david.sherwood@arm.com>
3106 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3108 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3109 Alan Hayward <alan.hayward@arm.com>
3110 David Sherwood <david.sherwood@arm.com>
3112 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3114 2017-10-23 Richard Biener <rguenther@suse.de>
3116 PR tree-optimization/82129
3117 * tree-ssa-pre.c (bitmap_set_and): Remove.
3118 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3119 canonicalizing expressions in the set to those with lowest
3120 ID rather than taking that from the first edge.
3122 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3124 * combine.c (rtx_equal_for_field_assignment_p): Use
3125 byte_lowpart_offset.
3127 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3128 Alan Hayward <alan.hayward@arm.com>
3129 David Sherwood <david.sherwood@arm.com>
3131 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3132 to a SUBREG_PROMOTED_VAR.
3134 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3135 Alan Hayward <alan.hayward@arm.com>
3136 David Sherwood <david.sherwood@arm.com>
3138 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3139 (expand_debug_source_expr): Likewise.
3140 * combine.c (combine_simplify_rtx): Likewise.
3141 * cse.c (fold_rtx): Likewise.
3142 * optabs.c (expand_float): Likewise.
3143 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3144 (simplify_binary_operation_1): Likewise.
3146 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3147 Alan Hayward <alan.hayward@arm.com>
3148 David Sherwood <david.sherwood@arm.com>
3150 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3151 (record_promoted_value): Likewise.
3152 * expr.c (expand_expr_real_2): Likewise.
3153 * ree.c (update_reg_equal_equiv_notes): Likewise.
3154 (combine_set_extension): Likewise.
3155 * rtlanal.c (low_bitmask_len): Likewise.
3156 * simplify-rtx.c (neg_const_int): Likewise.
3157 (simplify_binary_operation_1): Likewise.
3159 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3160 Alan Hayward <alan.hayward@arm.com>
3161 David Sherwood <david.sherwood@arm.com>
3163 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3164 * regcprop.c (maybe_mode_change): Likewise.
3165 * reload1.c (alter_reg): Likewise.
3167 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3169 * inchash.h (inchash::hash::add_wide_int): New function.
3170 * lto-streamer-out.c (hash_tree): Use it.
3172 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3174 * inchash.h (inchash::hash::add_wide_int): Rename to...
3175 (inchash::hash::add_hwi): ...this.
3176 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3177 (polymorphic_call_target_hasher::hash): Likewise.
3178 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3179 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3180 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3181 * lto-streamer-out.c (hash_tree): Likewise.
3182 * optc-save-gen.awk: Likewise.
3183 * tree.c (add_expr): Likewise.
3185 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3188 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3189 for ordered inequality comparisons even with TARGET_IEEE_FP.
3191 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3194 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3195 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3196 Expand with cmp<dwi>_doubleword.
3198 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3200 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3201 List CET intrinsics.
3202 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3203 specific to -fcf-protection option.
3205 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3207 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3208 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3209 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3210 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3211 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3212 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3213 (extra_objs): Add cet.o for Linux/x86 targets.
3214 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3215 * config/i386/cet.c: New file.
3216 * config/i386/cetintrin.h: Likewise.
3217 * config/i386/t-cet: Likewise.
3218 * config/i386/cpuid.h (bit_SHSTK): New.
3219 (bit_IBT): Likewise.
3220 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3221 pass IBT and SHSTK bits.
3222 * config/i386/i386-builtin-types.def
3223 (VOID_FTYPE_UNSIGNED_PVOID): New.
3224 (VOID_FTYPE_UINT64_PVOID): Likewise.
3225 * config/i386/i386-builtin.def: Add CET intrinsics.
3226 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3227 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3228 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3229 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3231 * config/i386/i386.c (rest_of_insert_endbranch): New.
3232 (pass_data_insert_endbranch): Likewise.
3233 (pass_insert_endbranch): Likewise.
3234 (make_pass_insert_endbranch): Likewise.
3235 (ix86_notrack_prefixed_insn_p): Likewise.
3236 (ix86_target_string): Add -mibt, -mshstk flags.
3237 (ix86_option_override_internal): Add flag_cf_protection
3239 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3240 (ix86_print_operand): Add 'notrack' prefix output.
3241 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3242 (ix86_expand_builtin): Expand CET intrinsics.
3243 (x86_output_mi_thunk): Add 'endbranch' instruction.
3244 * config/i386/i386.h (TARGET_IBT): New.
3245 (TARGET_IBT_P): Likewise.
3246 (TARGET_SHSTK): Likewise.
3247 (TARGET_SHSTK_P): Likewise.
3248 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3249 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3250 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3251 (builtin_setjmp_setup): New pattern.
3252 (builtin_longjmp): Likewise.
3253 (rdssp<mode>): Likewise.
3254 (incssp<mode>): Likewise.
3255 (saveprevssp): Likewise.
3256 (rstorssp): Likewise.
3257 (wrss<mode>): Likewise.
3258 (wruss<mode>): Likewise.
3259 (setssbsy): Likewise.
3260 (clrssbsy): Likewise.
3261 (nop_endbr): Likewise.
3262 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3264 * config/i386/immintrin.h: Include <cetintrin.h>.
3265 * config/i386/linux-common.h
3266 (file_end_indicate_exec_stack_and_cet): New prototype.
3267 (TARGET_ASM_FILE_END): New.
3269 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3271 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3272 latencies instead of having separate table; make difference between
3273 integer and float costs.
3274 * i386.h (processor_costs): Remove scalar_stmt_cost,
3275 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3276 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3278 * x86-tune-costs.h: Remove entries which has been removed in
3279 procesor_costs from all tables; make cond_taken_branch_cost
3280 and cond_not_taken_branch_cost COST_N_INSNS based.
3282 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3284 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3286 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3288 * config/i386/i386.md (isa): Remove fma_avx512f.
3289 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3290 <avx512>_fmadd_<mode>_mask3<round_name>,
3291 <avx512>_fmsub_<mode>_mask<round_name>,
3292 <avx512>_fmsub_<mode>_mask3<round_name>,
3293 <avx512>_fnmadd_<mode>_mask<round_name>,
3294 <avx512>_fnmadd_<mode>_mask3<round_name>,
3295 <avx512>_fnmsub_<mode>_mask<round_name>,
3296 <avx512>_fnmsub_<mode>_mask3<round_name>,
3297 <avx512>_fmaddsub_<mode>_mask<round_name>,
3298 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3299 <avx512>_fmsubadd_<mode>_mask<round_name>,
3300 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3301 (*vec_widen_umult_even_v16si<mask_name>,
3302 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3303 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3305 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3307 * extend.texi: Add 'nocf_check' documentation.
3308 * gimple.texi: Add second parameter to
3309 gimple_build_call_from_tree.
3310 * invoke.texi: Add -fcf-protection documentation.
3311 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3313 2017-10-20 Richard Biener <rguenther@suse.de>
3315 PR tree-optimization/82473
3316 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3317 the largest input type.
3319 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3321 * c-attribs.c (handle_nocf_check_attribute): New function.
3322 (c_common_attribute_table): Add 'nocf_check' handling.
3323 * gimple-parser.c: Add second argument NULL to
3324 gimple_build_call_from_tree.
3325 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3326 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3328 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3329 * common.opt: Add fcf-protection flag.
3330 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3331 * flag-types.h: Add enum cf_protection_level.
3332 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3333 Add 'nocf_check' attribute propagation to gimple call.
3334 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3335 (gimple_build_call_from_tree): Update prototype.
3336 (gimple_call_nocf_check_p): New function.
3337 (gimple_call_set_nocf_check): Likewise.
3338 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3339 * ipa-icf.c: Add nocf_check attribute in statement hash.
3340 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3341 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3342 * toplev.c (process_options): Add flag_cf_protection handling.
3344 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3346 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3348 2017-10-20 Richard Biener <rguenther@suse.de>
3350 PR tree-optimization/82603
3351 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3352 remove false predicated stores.
3354 2017-10-20 Richard Biener <rguenther@suse.de>
3356 * graphite-isl-ast-to-gimple.c
3357 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3358 Remove return value and simplify, dump copied stmt after lhs
3360 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3361 Reduce dump verbosity.
3362 (gsi_insert_earliest): Likewise.
3363 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3364 * graphite.c (print_global_statistics): Adjust dumping.
3365 (print_graphite_scop_statistics): Likewise.
3366 (print_graphite_statistics): Do not dump loops here.
3367 (graphite_transform_loops): But here.
3369 2017-10-20 Nicolas Roche <roche@adacore.com>
3371 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3372 * configure: Regenerate.
3374 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3377 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3378 functions when optimizing replace GIMPLE_RETURN stmts with
3379 calls to __builtin_unreachable ().
3382 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3383 for -fsanitize=thread link of executables.
3384 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3385 link of executables.
3388 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3390 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3391 define_insns for logical vector shifts to use VI248_AVX512BW
3392 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3393 condition, useless isa and prefix attributes. Change the first
3394 2 of these define_insns to ...
3395 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3396 define_insn for avx512vl.
3397 (<shift_insn><mode>3): ... and this, new define_insn without
3398 masking for non-avx512vl.
3401 * config/i386/sse.md (*andnot<mode>3,
3402 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3403 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3404 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3405 not applied use empty suffix even for TARGET_AVX512VL.
3406 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3407 is applied, supply evex,evex or evex,evex,evex instead of just
3410 2017-10-20 Julia Koval <julia.koval@intel.com>
3412 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3413 (OPTION_MASK_ISA_GFNI_UNSET): New.
3414 (ix86_handle_option): Handle OPT_mgfni.
3415 * config/i386/cpuid.h (bit_GFNI): New.
3416 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3417 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3418 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3419 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3420 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3421 * config/i386/i386.opt: Add mgfni.
3423 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3425 * config/msp430/msp430.c (msp430_option_override): Disable
3426 -fdelete-null-pointer-checks.
3427 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3429 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3431 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3432 of x87 and SSE instructions.
3434 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3436 * asan.c (create_cond_insert_point): Do not update edge count.
3437 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3438 (afdo_propagate_circuit): Likewise.
3439 (afdo_calculate_branch_prob): Likewise.
3440 (afdo_annotate_cfg): Likewise.
3441 * basic-block.h (struct edge_def): Remove count.
3442 (edge_def::count): New accessor.
3443 * bb-reorder.c (rotate_loop): Update.
3444 (find_traces_1_round): Update.
3445 (connect_traces): Update.
3446 (sanitize_hot_paths): Update.
3447 * cfg.c (unchecked_make_edge): Update.
3448 (make_single_succ_edge): Update.
3449 (check_bb_profile): Update.
3450 (dump_edge_info): Update.
3451 (update_bb_profile_for_threading): Update.
3452 (scale_bbs_frequencies_int): Update.
3453 (scale_bbs_frequencies_gcov_type): Update.
3454 (scale_bbs_frequencies_profile_count): Update.
3455 (scale_bbs_frequencies): Update.
3456 * cfganal.c (connect_infinite_loops_to_exit): Update.
3457 * cfgbuild.c (compute_outgoing_frequencies): Update.
3458 (find_many_sub_basic_blocks): Update.
3459 * cfgcleanup.c (try_forward_edges): Update.
3460 (try_crossjump_to_edge): Update
3461 * cfgexpand.c (expand_gimple_cond): Update
3462 (expand_gimple_tailcall): Update
3463 (construct_exit_block): Update
3464 * cfghooks.c (verify_flow_info): Update
3465 (redirect_edge_succ_nodup): Update
3466 (split_edge): Update
3467 (make_forwarder_block): Update
3468 (duplicate_block): Update
3469 (account_profile_record): Update
3470 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3471 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3472 * cfgloopmanip.c (scale_loop_profile): Update.
3474 (lv_adjust_loop_entry_edge): Update.
3475 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3476 (force_nonfallthru_and_redirect): Update.
3477 (purge_dead_edges): Update.
3478 (rtl_flow_call_edges_add): Update.
3479 * cgraphunit.c (init_lowered_empty_function): Update.
3480 (cgraph_node::expand_thunk): Update.
3481 * gimple-pretty-print.c (dump_probability): Update.
3482 (dump_edge_probability): Update.
3483 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3484 * haifa-sched.c (sched_create_recovery_edges): Update.
3485 * hsa-gen.c (convert_switch_statements): Update.
3486 * ifcvt.c (dead_or_predicable): Update.
3487 * ipa-inline-transform.c (inline_transform): Update.
3488 * ipa-split.c (split_function): Update.
3489 * ipa-utils.c (ipa_merge_profiles): Update.
3490 * loop-doloop.c (add_test): Update.
3491 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3492 * lto-streamer-in.c (input_cfg): Update.
3493 (input_function): Update.
3494 * lto-streamer-out.c (output_cfg): Update.
3495 * modulo-sched.c (sms_schedule): Update.
3496 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3497 * predict.c (maybe_hot_edge_p): Update.
3498 (unlikely_executed_edge_p): Update.
3499 (probably_never_executed_edge_p): Update.
3500 (dump_prediction): Update.
3501 (drop_profile): Update.
3502 (propagate_unlikely_bbs_forward): Update.
3503 (determine_unlikely_bbs): Update.
3504 (force_edge_cold): Update.
3505 * profile.c (compute_branch_probabilities): Update.
3506 * reg-stack.c (better_edge): Update.
3507 * shrink-wrap.c (handle_simple_exit): Update.
3508 * tracer.c (better_p): Update.
3509 * trans-mem.c (expand_transaction): Update.
3510 (split_bb_make_tm_edge): Update.
3511 * tree-call-cdce.c: Update.
3512 * tree-cfg.c (gimple_find_sub_bbs): Update.
3513 (gimple_split_edge): Update.
3514 (gimple_duplicate_sese_region): Update.
3515 (gimple_duplicate_sese_tail): Update.
3516 (gimple_flow_call_edges_add): Update.
3517 (insert_cond_bb): Update.
3518 (execute_fixup_cfg): Update.
3519 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3520 * tree-complex.c (expand_complex_div_wide): Update.
3521 * tree-eh.c (lower_resx): Update.
3522 (unsplit_eh): Update.
3523 (cleanup_empty_eh_move_lp): Update.
3524 * tree-inline.c (copy_edges_for_bb): Update.
3525 (freqs_to_counts): Update.
3526 (copy_cfg_body): Update.
3527 * tree-ssa-dce.c (remove_dead_stmt): Update.
3528 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3529 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3530 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3531 (unloop_loops): Update.
3532 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3533 * tree-ssa-loop-split.c (connect_loops): Update.
3534 (split_loop): Update.
3535 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3536 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3537 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3538 * tree-ssa-reassoc.c (branch_fixup): Update.
3539 * tree-ssa-tail-merge.c (replace_block_by): Update.
3540 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3541 (compute_path_counts): Update.
3542 (update_profile): Update.
3543 (recompute_probabilities): Update.
3544 (update_joiner_offpath_counts): Update.
3545 (estimated_freqs_path): Update.
3546 (freqs_to_counts_path): Update.
3547 (clear_counts_path): Update.
3548 (ssa_fix_duplicate_block_edges): Update.
3549 (duplicate_thread_path): Update.
3550 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3551 (case_bit_test_cmp): Update.
3552 (collect_switch_conv_info): Update.
3553 (gen_inbound_check): Update.
3554 (do_jump_if_equal): Update.
3555 (emit_cmp_and_jump_insns): Update.
3556 * tree-tailcall.c (decrease_profile): Update.
3557 (eliminate_tail_call): Update.
3558 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3559 (vect_do_peeling): Update.
3560 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3561 * ubsan.c (ubsan_expand_null_ifn): Update.
3562 (ubsan_expand_ptr_ifn): Update.
3563 * value-prof.c (gimple_divmod_fixed_value): Update.
3564 (gimple_mod_pow2): Update.
3565 (gimple_mod_subtract): Update.
3566 (gimple_ic): Update.
3567 (gimple_stringop_fixed_value): Update.
3569 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3572 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3574 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3576 PR rtl-optimization/82395
3577 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3578 based on non_spilled_static_chain_regno_p.
3580 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3582 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3583 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3584 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3586 2017-10-19 Martin Sebor <msebor@redhat.com>
3588 PR tree-optimization/82596
3589 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3591 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3593 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3594 (get_mem_refs_of_builtin_call): Likewise.
3595 * builtins.c (expand_builtin_apply): Adjust call to
3596 allocate_dynamic_stack_space.
3597 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3598 the third argument to allocate_dynamic_stack_space, otherwise -1.
3599 (expand_builtin): Deal with all alloca variants.
3600 (is_inexpensive_builtin): Likewise.
3601 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3602 * calls.c (special_function_p): Deal with all alloca variants.
3603 (initialize_argument_information): Adjust call to
3604 allocate_dynamic_stack_space.
3605 (expand_call): Likewise.
3606 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3607 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3608 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3609 use it for the stack usage computation.
3610 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3611 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3612 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3613 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3614 (in_loop_p): Remove first argument and useless check.
3615 (pass_walloca::execute): Remove useless test and adjust call to above.
3616 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3617 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3618 (gimplify_call_expr): Deal with all alloca variants.
3619 * hsa-gen.c (gen_hsa_alloca): Likewise.
3620 (gen_hsa_insns_for_call): Likewise.
3621 * ipa-pure-const.c (special_builtin_state): Likewise.
3622 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3623 * tree-object-size.c (alloc_object_size): Likewise.
3624 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3625 (call_may_clobber_ref_p_1): Likewise.
3626 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3627 (ccp_fold_stmt): Likewise.
3628 (optimize_stack_restore): Likewise.
3629 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3630 (mark_all_reaching_defs_necessary_1): Likewise.
3631 (propagate_necessity): Likewise.
3632 (eliminate_unnecessary_stmts): Likewise.
3633 * tree.c (build_common_builtin_nodes): Build
3634 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3635 (build_alloca_call_expr): New function.
3636 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3637 (CASE_BUILT_IN_ALLOCA): Likewise.
3638 (build_alloca_call_expr): Declare.
3639 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3641 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3644 * dwarf2out.c (new_die_raw): New static inline function.
3645 (new_die): Use it to create the DIE.
3646 (add_AT_external_die_ref): Likewise.
3647 (clone_die): Likewise.
3648 (clone_as_declaration): Likewise.
3649 (dwarf2out_vms_debug_main_pointer): Likewise.
3650 (base_type_die): Likewise. Remove early return for corner cases.
3651 Do not call add_pubtype on the DIE here.
3652 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3653 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3654 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3655 native order exists for base types, attach the DIE manually and call
3656 add_pubtype on it. Do not equate a reverse order DIE to the type.
3658 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3660 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3661 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3663 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3665 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3667 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3668 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3669 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3670 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3671 * opts.c (sanitizer_opts): Add builtin.
3672 * ubsan.c (instrument_builtin): New function.
3673 (pass_ubsan::execute): Call it.
3674 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3675 * doc/invoke.texi: Document -fsanitize=builtin.
3677 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3678 builtins, store max (log2 (align), 0) into uchar field instead of
3679 align into uptr field.
3680 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3681 store uchar 0 field instead of uptr 0 field.
3682 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3683 instead of passing one address of struct with 2 locations pass
3684 two addresses of structs with 1 location each.
3685 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3686 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3687 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3688 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3689 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3690 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3691 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3692 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3694 2017-10-19 Martin Liska <mliska@suse.cz>
3697 * file-find.c (remove_prefix): Remove.
3698 * file-find.h (remove_prefix): Likewise.
3699 * gcc-ar.c: Remove smartness of lookup.
3701 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3703 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3704 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3705 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3707 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3710 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3711 (setcc + and to xor + setcc): New peephole2.
3713 2017-10-19 Tom de Vries <tom@codesourcery.com>
3715 * doc/sourcebuild.texi (Test Directives, Variants of
3716 dg-require-support): Add dg-require-stack-size.
3718 2017-10-19 Martin Liska <mliska@suse.cz>
3721 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3722 that have a large alignment.
3723 (gimplify_target_expr): Likewise.
3725 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3727 PR rtl-optimization/82602
3728 * ira.c (rtx_moveable_p): Return false for volatile asm.
3730 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3733 * config/i386/i386-modes.def (CCGZ): New CC mode.
3734 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3735 * config/i386/predicates.md (ix86_comparison_operator):
3737 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3738 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3739 with double-word subtraction.
3740 (put_condition_code): Handle CCGZmode.
3742 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3744 * wide-int.cc (debug (const wide_int &)): New.
3745 (debug (const wide_int *)): New.
3746 (debug (const widest_int &)): New.
3747 (debug (const widest_int *)): New.
3749 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3752 * lra-constraints.c (curr_insn_transform): Use non-input operand
3753 instead of output one for matched reload.
3755 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3757 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3758 (tree-ssa-loop-ivopts.h): New header file.
3759 (struct builtin_info): New fields.
3760 (classify_builtin_1): Compute and record base and offset parts for
3761 memset builtin partition by calling strip_offset.
3762 (offset_cmp, fuse_memset_builtins): New functions.
3763 (finalize_partitions): Fuse adjacent memset partitions by calling
3765 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3766 Expose the interface.
3767 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3769 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3771 PR tree-optimization/82574
3772 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3773 that data reference must be executed exactly once per iteration
3774 against the outermost loop in nest.
3775 (classify_partition): Update call to above function.
3777 2017-10-18 Richard Biener <rguenther@suse.de>
3779 PR tree-optimization/82591
3780 * graphite.c (graphite_transform_loops): Move code gen message
3782 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3783 Here. Handle scop_to_isl_ast failing.
3784 (scop_to_isl_ast): Limit the number of ISL operations.
3786 2017-10-18 Richard Biener <rguenther@suse.de>
3788 * graphite-isl-ast-to-gimple.c
3789 (translate_isl_ast_to_gimple::set_rename): Simplify.
3790 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3791 (graphite_copy_stmts_from_block): ... here.
3792 (copy_bb_and_scalar_dependences): Simplify.
3793 (add_parameters_to_ivs_params): Canonicalize.
3794 (generate_entry_out_of_ssa_copies): Simplify.
3795 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3796 by passing in ISL dimension.
3797 (parameter_index_in_region_1): Rename to ...
3798 (parameter_index_in_region): ... this.
3799 (extract_affine): Adjust assert, pass down parameter index.
3800 (add_param_constraints): Use range-info when available.
3801 (build_scop_context): Adjust.
3802 * sese.c (new_sese_info): Adjust.
3803 (free_sese_info): Likewise.
3804 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3805 Remove unused typedefs.
3806 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3808 2017-10-18 Martin Liska <mliska@suse.cz>
3810 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3812 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3814 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3815 (s390_sched_init): Do not reset s390_sched_state if we entered the
3816 current basic block via a fallthru edge and all others are unlikely.
3818 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3820 * config/s390/s390.c (NUM_SIDES): New variable.
3821 (LONGRUNNING_THRESHOLD): New variable.
3822 (LATENCY_FACTOR): New variable.
3823 (s390_sched_score): Decrease score for long-running instructions on
3825 (s390_sched_variable_issue): Perform bookkeeping for long-running
3828 2017-10-18 Richard Biener <rguenther@suse.de>
3830 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3831 Simplify with removal of the parameter rename map.
3832 (set_rename): Likewise.
3833 (should_copy_to_new_region): Likewise.
3834 (graphite_copy_stmts_from_block): Likewise.
3835 (copy_bb_and_scalar_dependences): Remove initialization of
3836 unused copied_bb_map.
3838 (copy_internal_parameters): Likewise.
3839 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3840 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3841 Use INTEGRAL_TYPE_P.
3842 (parameter_index_in_region_1): Rename to ...
3843 (assign_parameter_index_in_region): ... this. Assert we have
3844 a parameter we handle.
3845 (scan_tree_for_params): Adjust.
3846 * sese.h (parameter_rename_map_t): Remove.
3847 (struct sese_info_t): Remove unused parameter_rename_map and
3848 copied_bb_map members.
3849 * sese.c (new_sese_info): Adjust.
3850 (free_sese_info): Likewise.
3852 2017-10-18 Martin Liska <mliska@suse.cz>
3855 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3856 on an abnormal edge.
3858 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3860 * doc/invoke.texi (ffunction-sections and fdata-sections):
3863 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3865 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3866 the use statement can throw internally.
3868 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3870 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3871 any RTX present on the RHS of a SET.
3872 * compare-elim.c (try_eliminate_compare): Restore comment.
3874 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3876 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3877 may be also called on tcc_type nodes.
3878 * langhooks.c (lhd_tree_size): Likewise.
3880 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3882 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3883 format_warning_at_substring.
3884 (maybe_warn): Convert source_range * param to a location_t. Pass
3885 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3886 (format_directive): Remove code to extract source_ranges and
3887 source_range * in favor of just a location_t.
3888 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3890 * substring-locations.c (format_warning_va): Convert
3891 source_range * param to a location_t.
3892 (format_warning_at_substring): Likewise.
3893 * substring-locations.h (format_warning_va): Likewise.
3894 (format_warning_at_substring): Likewise.
3896 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3898 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3900 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3901 and scatter/gather ops.
3903 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3904 vec_gather_load and vec_scatter_store.
3905 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3906 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3907 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3908 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3909 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3910 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3912 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3914 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3915 Detect FTST instruction and handle its register pops. Only pop
3916 second operand if can_pop_second_op is true.
3917 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3918 set can_pop_second_op to false in the compare_for_stack_reg call.
3920 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3921 output_fp_compare for stack register operands.
3922 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3923 instructions here. Do not emit stack register pops here. Assert
3924 that FCOMPP pops next to top stack register. Rewrite function.
3926 2017-10-17 Nathan Sidwell <nathan@acm.org>
3929 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3930 use DECL_ASSEMBLER_NAME_RAW.
3933 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3936 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3937 Wilco Dijkstra <wilco.dijkstra@arm.com>
3939 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3940 converstion to Pmode from the buf_addr.
3942 2017-10-17 Richard Biener <rguenther@suse.de>
3944 * graphite-dependences.c (scop_get_reads_and_writes): Change
3945 output parameters to references.
3947 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3949 PR 71026/tree-optimization
3950 * fold-const.c (distribute_real_division): Removed.
3951 (fold_binary_loc): Remove calls to distribute_real_divison.
3953 2017-10-17 Richard Biener <rguenther@suse.de>
3955 * graphite-scop-detection.c
3956 (scop_detection::stmt_has_simple_data_refs_p): Always use
3957 the full nest as region.
3958 (try_generate_gimple_bb): Likewise.
3959 * sese.c (scalar_evolution_in_region): Simplify now that
3960 SCEV can handle instantiation in regions.
3961 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3962 in the non-loop part of a function if requested.
3964 2017-10-17 Richard Biener <rguenther@suse.de>
3966 PR tree-optimization/82563
3967 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3969 (graphite_regenerate_ast_isl): Call it.
3970 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3972 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3974 PR tree-optimization/82549
3975 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3976 Formatting fixes. Instead of calling make_bit_field_ref with negative
3979 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3981 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3982 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3983 _mm_maskz_reduce_ss): New.
3984 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3985 __builtin_ia32_reducess_mask): Ditto..
3986 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3987 * config/i386/sse.md (reduces<mode>): Renamed to ...
3988 (reduces<mode><mask_scalar_name>): ... this.
3989 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3991 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3992 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3994 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3996 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3997 * selftest-run-tests.c (selftest::run_tests): Call
3998 selftest::unique_ptr_tests_cc_tests.
3999 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4000 * unique-ptr-tests.cc: New file.
4002 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4005 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4007 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4008 (make_hard_regno_born, make_hard_regno_dead): Update
4009 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4011 2017-10-16 Jeff Law <law@redhat.com>
4013 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4015 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4017 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4019 2017-10-16 Olivier Hainque <hainque@adacore.com>
4021 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4022 with_cpu if we were configured for an e500v2 target cpu name.
4024 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4026 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4027 * doc/invoke.texi: Document +nodsp as a valid extension for
4030 2017-10-16 Martin Liska <mliska@suse.cz>
4032 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4033 (test_set_range): Likewise.
4034 (test_range_functions): Rename to ...
4035 (test_bit_in_range): ... this.
4036 (sbitmap_c_tests): Add new test.
4038 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4040 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4042 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4043 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4045 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4047 * config/aarch64/aarch64-builtins.c
4048 (aarch64_types_quadopu_lane_qualifiers): New.
4049 (TYPES_QUADOPU_LANE): New.
4050 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4051 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4052 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4053 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4054 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4055 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4056 (Vdottype, DOTPROD): New.
4057 (sur): Add SDOT and UDOT.
4059 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4061 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4062 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4063 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4065 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4066 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4067 Enable TARGET_DOTPROD.
4068 (cortex-a75.cortex-a55): Likewise.
4069 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4071 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4073 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4074 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4076 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4078 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4079 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4080 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4081 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4082 * config/arm/types.md (neon_dot, neon_dot_q): New.
4083 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4085 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4087 * config/arm/arm.h (TARGET_DOTPROD): New.
4088 * config/arm/arm.c (arm_arch_dotprod): New.
4089 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4090 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4091 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4092 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4093 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4094 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4095 * doc/invoke.texi (armv8.2-a): Document dotprod
4097 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4099 * i386.c (ix86_vec_cost): New function.
4100 (ix86_rtx_costs): Handle vector operations better.
4101 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4102 * x86-tune-costs.h: Add new costs to all tables.
4104 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4106 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4108 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4109 divsd, sqrtss and sqrtsd
4110 * x86-tune-costs.h: Add new entries to all costs.
4111 (znver1_cost): Fix to match real instruction latencies.
4113 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4114 Michael Collison <michael.collison@arm.com>
4116 * compare-elim.c: Include emit-rtl.h.
4117 (can_merge_compare_into_arith): New function.
4118 (try_validate_parallel): Likewise.
4119 (try_merge_compare): Likewise.
4120 (try_eliminate_compare): Call the above when no previous clobber
4122 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4125 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4129 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4130 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4134 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4135 to be any operand_equal_p operands. For & (B - 1) require
4136 B to be power of 2. Recognize
4137 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4139 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4142 * optabs.c (expand_memory_blockage): Fix call of
4143 targetm.have_memory_blockage.
4145 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4148 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4149 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4150 objects to extra_objs instead of overwriting it.
4152 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4154 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4155 Use any_fp_register_operand as operand[3] predicate. Simplify
4156 equality test for operands[2] and operands[4] memory location.
4157 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4158 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4159 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4160 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4161 any_fp_register_operand as operand[1] predicate. Simplify
4162 equality test for operands[0] and operands[3] memory location.
4163 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4164 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4165 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4167 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4169 * target-insns.def: Add memory_blockage.
4170 * optabs.c (expand_memory_blockage): New function.
4171 (expand_asm_memory_barrier): Rename ...
4172 (expand_asm_memory_blockage): ... to this.
4173 (expand_mem_thread_fence): Call expand_memory_blockage
4174 instead of expand_asm_memory_barrier.
4175 (expand_mem_singnal_fence): Ditto.
4176 (expand_atomic_load): Ditto.
4177 (expand_atomic_store): Ditto.
4178 * doc/md.texi (Standard Pattern Names For Generation):
4179 Document memory_blockage instruction pattern.
4181 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4183 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4184 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4185 * config/rl78/rl78.md: New define_expand "adddi3".
4187 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4189 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4192 2017-10-13 Jeff Law <law@redhat.com>
4194 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4196 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4199 * internal-fn.c (expand_mul_overflow): If both operands have
4200 the same highpart of -1 or 0 and the topmost bit of lowpart
4201 is different, overflow is if res <= 0 rather than res < 0.
4203 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4205 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4206 TARGET_P9_VECTOR code for unaligned_load case.
4208 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4210 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4212 2017-10-13 Nathan Sidwell <nathan@acm.org>
4214 * tree-core.h (tree_contains_struct): Make bool.
4215 * tree.c (tree_contains_struct): Likewise.
4216 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4217 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4218 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4219 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4220 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4222 2017-10-13 Richard Biener <rguenther@suse.de>
4224 * graphite-isl-ast-to-gimple.c
4225 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4226 parameters and dominance check.
4227 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4228 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4229 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4230 Do not update SSA form here or do intermediate IL verification.
4231 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4232 (graphite_initialize): Remove check on the number of loops in
4233 the function and inline into graphite_transform_loops.
4234 (graphite_finalize): Inline into graphite_transform_loops.
4235 (graphite_transform_loops): Perform SSA update and IL verification
4237 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4239 2017-10-13 Richard Biener <rguenther@suse.de>
4241 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4242 graphite_expression_type_precision): Avoid global constructor
4244 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4245 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4246 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4247 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4248 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4249 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4251 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4254 * config/i386/i386.h (ix86_red_zone_size): New.
4255 * config/i386/i386.md (push peephole2s): Replace
4256 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4258 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4259 Alan Hayward <alan.hayward@arm.com>
4260 David Sherwood <david.sherwood@arm.com>
4262 * combine.c (can_change_dest_mode): Reject changes in
4263 REGMODE_NATURAL_SIZE.
4265 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4266 Alan Hayward <alan.hayward@arm.com>
4267 David Sherwood <david.sherwood@arm.com>
4269 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4270 (expand_debug_source_expr): Likewise.
4271 * combine.c (combine_simplify_rtx): Likewise.
4272 * cse.c (fold_rtx): Likewise.
4273 * fwprop.c (canonicalize_address): Likewise.
4274 * targhooks.c (default_shift_truncation_mask): Likewise.
4276 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4277 Alan Hayward <alan.hayward@arm.com>
4278 David Sherwood <david.sherwood@arm.com>
4280 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4281 (widened_mode): Likewise.
4282 (expand_unop): Likewise.
4283 * ree.c (transform_ifelse): Likewise.
4284 (merge_def_and_ext): Likewise.
4285 (combine_reaching_defs): Likewise.
4286 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4288 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4289 Alan Hayward <alan.hayward@arm.com>
4290 David Sherwood <david.sherwood@arm.com>
4292 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4293 * combine.c (gen_lowpart_for_combine): Likewise.
4294 * dwarf2out.c (rtl_for_decl_location): Likewise.
4295 * final.c (alter_subreg): Likewise.
4296 * rtlhooks.c (gen_lowpart_general): Likewise.
4297 (gen_lowpart_if_possible): Likewise.
4299 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4300 Alan Hayward <alan.hayward@arm.com>
4301 David Sherwood <david.sherwood@arm.com>
4303 * calls.c (expand_call): Use subreg_lowpart_offset.
4304 * cse.c (cse_insn): Likewise.
4305 * regcprop.c (copy_value): Likewise.
4306 (copyprop_hardreg_forward_1): Likewise.
4308 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4311 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4312 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4313 =Q constraints to +Q and into insn condition add check
4314 that operands[0] and operands[1] are equal.
4315 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4316 =Q constraints to +Q and into insn condition add check
4317 that operands[0] is equal to either operands[1] or operands[2].
4320 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4321 instead of handling MINUS_EXPR twice (once for each argument),
4322 canonicalize operand order and handle just once, use rtype where
4323 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4326 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4327 any values of __C while still being pattern recognizable as a simple
4330 2017-10-13 Richard Biener <rguenther@suse.de>
4332 PR tree-optimization/82451
4334 2017-10-02 Richard Biener <rguenther@suse.de>
4336 PR tree-optimization/82355
4337 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4338 a mapping for the enclosing loop but avoid generating one for
4340 (copy_bb_and_scalar_dependences): Remove premature codegen
4341 error on PHIs in blocks duplicated into multiple places.
4342 * graphite-scop-detection.c
4343 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4344 in the region use it as loop and nest to analyze the DR in.
4345 (try_generate_gimple_bb): Likewise.
4346 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4347 (add_loop_constraints): For blocks in a loop not in the region
4348 create a dimension with a single iteration.
4349 * sese.h (gbb_loop_at_index): Remove assert.
4351 * cfgloop.c (loop_preheader_edge): For the loop tree root
4352 return the single successor of the entry block.
4353 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4354 Reset the SCEV hashtable and niters.
4355 * graphite-scop-detection.c
4356 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4357 assert that we only have POLYNOMIAL_CHREC that vary in loops
4358 contained in the region.
4359 (scop_detection::graphite_can_represent_expr): Adjust.
4360 (scop_detection::stmt_has_simple_data_refs_p): For loops
4361 not in the region set loop to NULL. The nest is now the
4362 entry edge to the region.
4363 (try_generate_gimple_bb): Likewise.
4364 * sese.c (scalar_evolution_in_region): Adjust for
4365 instantiate_scev change.
4366 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4367 Make nest parameter the edge into the region.
4368 (create_data_ref): Likewise.
4369 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4370 entry edge into a region and adjust instantiate_scev calls.
4371 (create_data_ref): Likewise.
4372 (graphite_find_data_references_in_stmt): Likewise.
4373 (find_data_references_in_stmt): Pass the loop preheader edge
4374 from the nest argument.
4375 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4376 parameter the edge into the region.
4377 (instantiate_parameters): Use the loop preheader edge as entry.
4378 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4380 (get_instantiated_value_entry): Make instantiate_below parameter
4381 the edge into the region.
4382 (instantiate_scev_name): Likewise. Adjust dominance checks,
4383 when we cannot use loop-based instantiation instantiate by
4384 walking use-def chains.
4385 (instantiate_scev_poly): Adjust.
4386 (instantiate_scev_binary): Likewise.
4387 (instantiate_scev_convert): Likewise.
4388 (instantiate_scev_not): Likewise.
4389 (instantiate_array_ref): Remove.
4390 (instantiate_scev_3): Likewise.
4391 (instantiate_scev_2): Likewise.
4392 (instantiate_scev_1): Likewise.
4393 (instantiate_scev_r): Do not blindly handle N-operand trees.
4394 Do not instantiate array-refs. Handle all constants and invariants.
4395 (instantiate_scev): Make instantiate_below parameter
4396 the edge into the region.
4397 (resolve_mixers): Use the loop preheader edge for the region
4398 parameter to instantiate_scev_r.
4399 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4401 2017-10-13 Richard Biener <rguenther@suse.de>
4403 PR tree-optimization/82525
4404 * graphite-isl-ast-to-gimple.c
4405 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4407 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4408 Fail code generation when we cannot represent the isl integer.
4409 (binary_op_to_tree): Elide modulo operations that are no-ops
4410 in the type we code generate. Remove now superfluous code
4413 2017-10-13 Richard Biener <rguenther@suse.de>
4415 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4416 (scop_detection::harmful_loop_in_region): Remove premature
4417 IV type restriction.
4418 (scop_detection::graphite_can_represent_scev): We can handle
4419 pointer IVs just fine.
4421 2017-10-13 Alan Modra <amodra@gmail.com>
4423 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4424 "Clobbers and Scratch Registers". Add paragraph on
4425 alternative to clobbers for scratch registers and OpenBLAS
4428 2017-10-13 Alan Modra <amodra@gmail.com>
4430 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4431 example of a memory input for a string of known length. Move
4432 commentary out of table. Add a number of new examples
4433 covering array memory inputs.
4435 2017-10-12 Martin Liska <mliska@suse.cz>
4437 PR tree-optimization/82493
4438 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4439 (test_range_functions): New function.
4440 (sbitmap_c_tests): Likewise.
4441 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4442 * selftest.h (sbitmap_c_tests): New function.
4444 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4446 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4448 * config/rs6000/amo.h: Fix spacing issue.
4450 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4453 * config/i386/i386.md (*ashl<mode>3_mask_1,
4454 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4455 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4458 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4460 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4461 (profile_probability): Set max_probability
4462 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4464 * profile-count.c (profile_probability::differs_from_p): Do not
4465 rely on max_probaiblity == 10000
4467 2017-10-12 Jeff Law <law@redhat.com>
4469 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4472 2017-10-12 Martin Sebor <msebor@redhat.com>
4476 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4477 (handle_alias_pairs): Call it.
4478 * common.opt (-Wattribute-alias): New option.
4479 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4480 * doc/invoke.texi (-Wattribute-alias): Document.
4482 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4485 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4487 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4489 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4490 (make_hard_regno_born, make_hard_regno_dead): Update
4491 bb_killed_pseudos and bb_gen_pseudos.
4493 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4495 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4497 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4499 * config/alpha/alpha.c (alpha_split_conditional_move):
4500 Use std::swap instead of manually swapping.
4501 (alpha_stdarg_optimize_hook): Ditto.
4502 (alpha_canonicalize_comparison): Ditto.
4504 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4506 * tree-loop-distribution.c (struct builtin_info): New struct.
4507 (struct partition): Refactor fields into struct builtin_info.
4508 (partition_free): Free struct builtin_info.
4509 (build_size_arg_loc, build_addr_arg_loc): Delete.
4510 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4511 information from struct builtin_info.
4512 (find_single_drs): New function refactored from classify_partition.
4513 Also moved builtin validity checks to this function.
4514 (compute_access_range, alloc_builtin): New functions.
4515 (classify_builtin_st, classify_builtin_ldst): New functions.
4516 (classify_partition): Refactor code into functions find_single_drs,
4517 classify_builtin_st and classify_builtin_ldst.
4518 (distribute_loop): Don't do runtime alias check when distributing
4520 (find_seed_stmts_for_distribution): New function.
4521 (pass_loop_distribution::execute): Refactor code finding seed
4522 stmts into above function. Support distribution for the innermost
4523 two-level loop nest. Adjust dump information.
4525 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4527 * tree-loop-distribution.c: Adjust the general comment.
4528 (NUM_PARTITION_THRESHOLD): New macro.
4529 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4530 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4531 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4532 in call to build_partition_graph.
4533 (finalize_partitions): New parameter. Make loop distribution more
4534 conservative by fusing more partitions.
4535 (distribute_loop): Don't do runtime alias check in case of loop nest
4537 (find_seed_stmts_for_distribution): New function.
4538 (prepare_perfect_loop_nest): New function.
4539 (pass_loop_distribution::execute): Refactor code finding seed stmts
4540 and loop nest into above functions. Support loop nest distribution.
4541 Adjust dump information accordingly.
4543 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4545 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4546 and set PTYPE_SEQUENTIAL for merged partition.
4548 2017-10-12 Richard Biener <rguenther@suse.de>
4550 PR tree-optimization/69728
4552 2017-09-19 Richard Biener <rguenther@suse.de>
4554 PR tree-optimization/69728
4555 * graphite-sese-to-poly.c (schedule_error): New global.
4556 (add_loop_schedule): Handle empty domain by failing the
4558 (build_original_schedule): Handle schedule_error.
4560 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4561 domain by returning an unchanged schedule.
4563 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4565 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4566 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4568 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4570 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4573 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4576 * expr.c (store_field): Don't optimize away bitsize == 0 store
4577 from CALL_EXPR with addressable return type.
4579 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4581 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4582 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4583 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4584 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4585 TARGET_ISEL instead of TARGET_ISEL<sel>.
4587 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4589 * config/rs6000/rs6000.c
4590 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4592 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4594 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4596 (reg_or_cint_operand): Fix comment.
4597 (reg_or_zero_operand): New predicate.
4598 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4599 * config/rs6000/rs6000.c (output_isel): Delete.
4600 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4601 instead of reg_or_cint_operand. Output instruction directly (not via
4603 (isel_unsigned_<mode>): Ditto.
4604 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4605 gpc_reg_operand. Add an instruction alternative for this. Output
4606 instruction directly.
4607 (*isel_reversed_unsigned_<mode>): Ditto.
4609 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4611 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4612 (TARGET_CANONICALIZE_COMPARISON): Define.
4614 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4617 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4618 Check whether the dest is REG before adding REG_EQUIV note.
4620 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4623 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4625 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4626 (make_hard_regno_born, make_hard_regno_dead): Update
4627 bb_killed_pseudos and bb_gen_pseudos.
4629 2017-10-11 Nathan Sidwell <nathan@acm.org>
4631 * incpath.h (enum incpath_kind): Name enum, prefix values.
4632 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4633 * incpath.c (heads, tails): Use INC_MAX.
4634 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4635 (merge_include_chains, split_quote_chain,
4636 register_include_chains): Update incpath_kind names.
4637 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4638 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4640 (add_framework_path, darwin_register_objc_includes): Likewise.
4641 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4643 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4645 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4646 Do not use float_operator operator predicate.
4647 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4648 * config/i386/predicates.md (float_operator): Remove predicate.
4650 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4652 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4653 (*jccxf_i387): Ditto.
4654 (*jcc<mode>_i387): Ditto.
4655 (*jccu<mode>_i387): Ditto.
4656 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4657 (*jcc_*_i387 splitters): Remove.
4658 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4659 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4660 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4663 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4665 * profile-count.h (slow_safe_scale_64bit): New function.
4666 (safe_scale_64bit): New inline.
4667 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4668 * profile-count.c: Include wide-int.h
4669 (slow_safe_scale_64bit): New.
4671 2017-10-11 Nathan Sidwell <nathan@acm.org>
4673 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4674 HAS_DECL_ASSEMBLER_NAME_P.
4675 * gimple-expr.c (gimple_decl_printable_name: Check
4676 HAS_DECL_ASSEMBLER_NAME_P too.
4677 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4678 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4679 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4680 * passes.c (rest_of_decl_compilation): Check
4681 HAS_DECL_ASSEMBLER_NAME_P too.
4682 * recog.c (verify_changes): Likewise.
4683 * tree-pretty-print.c (dump_decl_name): Likewise.
4684 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4686 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4687 (SET_DECL_ASSEMBLER_NAME): Use it.
4688 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4689 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4690 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4692 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4694 * config.gcc (i386, x86_64): Add extra objects.
4695 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4696 (ix86_min_insn_size): Declare.
4697 (ix86_issue_rate): Declare.
4698 (ix86_adjust_cost): Declare.
4699 (ia32_multipass_dfa_lookahead): Declare.
4700 (ix86_macro_fusion_p): Declare.
4701 (ix86_macro_fusion_pair_p): Declare.
4702 (ix86_bd_has_dispatch): Declare.
4703 (ix86_bd_do_dispatch): Declare.
4704 (ix86_core2i7_init_hooks): Declare.
4705 (ix86_atom_sched_reorder): Declare.
4706 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4707 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4708 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4709 (rip_relative_addr_p): Rename to ...
4710 (ix86_rip_relative_addr_p): ... this one; export.
4711 (memory_address_length): Update.
4712 (ix86_issue_rate): Move to x86-tune-sched.c.
4713 (ix86_flags_dependent): Move to x86-tune-sched.c.
4714 (ix86_agi_dependent): Move to x86-tune-sched.c.
4715 (exact_dependency_1): Move to x86-tune-sched.c.
4716 (exact_store_load_dependency): Move to x86-tune-sched.c.
4717 (ix86_adjust_cost): Move to x86-tune-sched.c.
4718 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4719 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4720 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4721 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4722 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4723 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4724 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4725 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4726 (min_insn_size): Rename to ...
4727 (ix86_min_insn_size): ... this one; export.
4728 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4729 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4730 (core2i7_first_cycle_multipass_backtrack): Move to
4731 x86-tune-sched-core.c.
4732 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4733 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4734 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4735 (ix86_avoid_jump_mispredicts): Update.
4736 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4737 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4738 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4739 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4740 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4741 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4742 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4743 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4744 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4745 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4746 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4747 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4748 (BIG): Move to ix86-tune-sched-bd.c.
4749 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4750 (enum insn_path): Move to ix86-tune-sched-bd.c.
4751 (get_mem_group): Move to ix86-tune-sched-bd.c.
4752 (is_cmp): Move to ix86-tune-sched-bd.c.
4753 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4754 (is_branch): Move to ix86-tune-sched-bd.c.
4755 (is_prefetch): Move to ix86-tune-sched-bd.c.
4756 (init_window): Move to ix86-tune-sched-bd.c.
4757 (allocate_window): Move to ix86-tune-sched-bd.c.
4758 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4759 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4760 (process_end_window): Move to ix86-tune-sched-bd.c.
4761 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4762 (find_constant): Move to ix86-tune-sched-bd.c.
4763 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4764 (has_immediate): Move to ix86-tune-sched-bd.c.
4765 (get_insn_path): Move to ix86-tune-sched-bd.c.
4766 (get_insn_group): Move to ix86-tune-sched-bd.c.
4767 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4768 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4769 (add_insn_window): Move to ix86-tune-sched-bd.c.
4770 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4771 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4772 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4773 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4774 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4775 (do_dispatch): Move to ix86-tune-sched-bd.c.
4776 (has_dispatch): Move to ix86-tune-sched-bd.c.
4777 * config/i386/t-i386: Add new object files.
4778 * config/i386/x86-tune-costs.h: New file.
4779 * config/i386/x86-tune-sched-atom.c: New file.
4780 * config/i386/x86-tune-sched-bd.c: New file.
4781 * config/i386/x86-tune-sched-core.c: New file.
4782 * config/i386/x86-tune-sched.c: New file.
4784 2017-10-11 Liu Hao <lh_mouse@126.com>
4786 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4787 the generic version below instead.
4788 (should_colorize): Recognize Windows consoles as terminals
4790 * pretty-print.c [__MINGW32__] (write_all): New function.
4791 [__MINGW32__] (find_esc_head): Likewise.
4792 [__MINGW32__] (find_esc_terminator): Likewise.
4793 [__MINGW32__] (eat_esc_sequence): Likewise.
4794 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4796 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4799 2017-10-11 Richard Biener <rguenther@suse.de>
4801 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4802 Properly call analyze_scalar_evolution with the loop of the stmt.
4804 2017-10-11 Richard Biener <rguenther@suse.de>
4806 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4807 * tree-core.h (tree_base): Add chrec_var union member.
4808 * tree.h (CHREC_VAR): Remove.
4809 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4810 * tree-chrec.h (build_polynomial_chrec): Adjust.
4811 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4812 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4814 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4816 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4817 * match.pd: ... here.
4818 ((T) X == (T) Y): Relax condition.
4820 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4822 PR tree-optimization/82472
4823 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4825 (break_alias_scc_partitions): Update postorder number.
4827 2017-10-11 Martin Liska <mliska@suse.cz>
4830 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4832 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4834 2017-10-11 Martin Liska <mliska@suse.cz>
4836 * print-rtl.c (print_insn): Move declaration of idbuf
4837 to same scope as name.
4839 2017-10-11 Martin Liska <mliska@suse.cz>
4844 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4847 2017-10-11 Martin Liska <mliska@suse.cz>
4850 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4853 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4855 * config.gcc (default_gnu_indirect_function): Default to yes for
4856 arm*-*-linux* with glibc.
4858 2017-10-11 Richard Biener <rguenther@suse.de>
4860 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4861 default-defs and types we do not want to analyze.
4862 (interpret_loop_phi): Replace unreachable code with an assert.
4863 (compute_scalar_evolution_in_loop): Remove and inline ...
4864 (analyze_scalar_evolution_1): ... here, replacing condition with
4865 what makes the intent clearer. Remove handling of cases
4866 get_scalar_evolution now handles.
4868 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4870 PR rtl-optimization/81434
4871 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4872 comment for main loop. In sched_group_found if, also add checks for
4873 pass and min_cost_group.
4875 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4877 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4878 (rs6000_insn_cost): New function.
4879 * config/rs6000/rs6000.md (cost): New attribute.
4881 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4882 H.J. Lu <hongjiu.lu@intel.com>
4886 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4887 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4888 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4889 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4890 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4891 OPTION_MASK_ISA_MMX and some other bit set require both
4892 mmx and the ISAs without the mmx bit.
4893 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4894 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4895 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4896 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4897 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4898 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4899 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4900 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4901 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4902 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4903 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4904 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4905 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4906 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4907 Add OPTION_MASK_ISA_MMX.
4909 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4911 * config.gcc (armv7*-*-freebsd*): New target.
4912 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4914 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4916 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4917 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4920 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4922 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4923 a template parameter.
4924 (WIDE_INT_REF_FOR): Update accordingly.
4925 * tree.h (wi::int_traits <const_tree>): Delete.
4926 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4927 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4928 (wi::tree_to_wide_ref): New typedef.
4929 (wi::to_wide): New function.
4930 * calls.c (get_size_range): Use wi::to_wide when operating on
4932 * cgraph.c (cgraph_node::create_thunk): Likewise.
4933 * config/i386/i386.c (ix86_data_alignment): Likewise.
4934 (ix86_local_alignment): Likewise.
4935 * dbxout.c (stabstr_O): Likewise.
4936 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4937 * expr.c (const_vector_from_tree): Likewise.
4938 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4939 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4940 (fold_negate_expr_1, int_const_binop_1, const_binop)
4941 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4942 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4943 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4944 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4945 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4946 (fold_not_const, round_up_loc): Likewise.
4947 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4948 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4949 (alloca_call_type): Likewise.
4950 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4951 * godump.c (go_output_typedef): Likewise.
4952 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4953 * internal-fn.c (get_min_precision): Likewise.
4954 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4955 * ipa-polymorphic-call.c
4956 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4957 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4958 (ipa_modify_call_arguments): Likewise.
4959 * match.pd: Likewise.
4960 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4961 * print-tree.c (print_node_brief, print_node): Likewise.
4962 * stmt.c (expand_case): Likewise.
4963 * stor-layout.c (layout_type): Likewise.
4964 * tree-affine.c (tree_to_aff_combination): Likewise.
4965 * tree-cfg.c (group_case_labels_stmt): Likewise.
4966 * tree-data-ref.c (dr_analyze_indices): Likewise.
4967 (prune_runtime_alias_test_list): Likewise.
4968 * tree-dump.c (dequeue_and_dump): Likewise.
4969 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4970 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4971 * tree-pretty-print.c (dump_generic_node): Likewise.
4972 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4973 (simple_iv_with_niters): Likewise.
4974 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4975 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4976 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4977 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4978 (refine_value_range_using_guard, number_of_iterations_ne_max)
4979 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4980 (get_cst_init_from_scev, record_nonwrapping_iv)
4981 (scev_var_range_cant_overflow): Likewise.
4982 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4983 * tree-ssa-pre.c (compute_avail): Likewise.
4984 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4985 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4986 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4987 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4988 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4989 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4990 (dump_case_nodes, try_switch_expansion): Likewise.
4991 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4992 (vect_do_peeling): Likewise.
4993 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4994 * tree-vect-stmts.c (vectorizable_load): Likewise.
4995 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4996 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4997 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4998 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4999 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5000 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5001 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5002 (evrp_dom_walker::before_dom_children): Likewise.
5003 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5004 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5005 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5006 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5007 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5008 (get_range_pos_neg): Likewise.
5009 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5010 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5011 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5012 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5013 * config/avr/avr.c (avr_fold_builtin): Likewise.
5014 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5015 * config/msp430/msp430.c (msp430_attr): Likewise.
5016 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5017 * config/powerpcspe/powerpcspe-c.c
5018 (altivec_resolve_overloaded_builtin): Likewise.
5019 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5020 (rs6000_expand_ternop_builtin): Likewise.
5021 * config/rs6000/rs6000-c.c
5022 (altivec_resolve_overloaded_builtin): Likewise.
5023 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5024 (rs6000_expand_ternop_builtin): Likewise.
5025 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5027 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5029 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5030 when copying loop nest with only one inner loop.
5032 2017-10-10 Richard Biener <rguenther@suse.de>
5034 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5035 blocks if SCEV is active.
5036 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5038 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5039 (scev_initialize): Assert we are not yet initialized.
5041 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5043 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5044 inner loop's exit stmt by making it always exit the loop, otherwise
5045 we would generate an infinite empty loop.
5047 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5049 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5050 renaming variables in new preheader if it's deleted.
5052 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5054 * tree-loop-distribution.c (struct partition): Remove unused field
5055 loops of the structure.
5056 (partition_alloc, partition_free): Ditto.
5057 (build_rdg_partition_for_vertex): Ditto.
5059 2017-10-09 Jeff Law <law@redhat.com>
5061 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5062 return type to match prototype and documentation.
5064 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5066 * config/rs6000/rs6000.c (processor_costs): Move to ...
5067 * config/rs6000/rs6000.h: ... here.
5068 (rs6000_cost): Declare.
5070 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5072 * except.c (setjmp_fn): New global variable.
5073 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5074 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5075 if DONT_USE_BUILTIN_SETJMP is defined.
5077 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5079 * target.def (insn_cost): New hook.
5080 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5081 * doc/tm.texi: Regenerate.
5082 * rtlanal.c (insn_cost): Use the new hook.
5084 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5086 * combine.c (combine_validate_cost): Compute the new insn_cost,
5087 not just pattern_cost.
5088 (try_combine): Adjust comment.
5090 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5092 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5094 * combine.c (uid_insn_cost): Adjust comment.
5095 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5097 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5098 * dse.c (find_shift_sequence): Ditto.
5099 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5100 (bb_valid_for_noce_process_p): Use pattern_cost.
5101 * rtl.h (insn_rtx_cost): Delete.
5102 (pattern_cost): New prototype.
5103 (insn_cost): New prototype.
5104 * rtlanal.c (insn_rtx_cost): Rename to...
5105 (pattern_cost): ... this.
5108 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5110 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5111 (*jcc<mode>_0_r_i387): Ditto.
5112 (*jccxf_r_i387): Ditto.
5113 (*jcc<mode>_r_i387): Ditto.
5114 (*jccu<mode>_r_i387): Ditto.
5115 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5116 (*jcc): Rename from *jcc_1.
5118 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5120 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5121 deferred rescans after the lvx/stvx recombination pre-pass.
5123 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5125 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5126 memory operation instruction support.
5127 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5128 (rs6000-ibm-aix[789]*): Likewise.
5129 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5130 Document new functions.
5132 2017-10-09 Richard Biener <rguenther@suse.de>
5134 PR tree-optimization/82397
5135 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5136 equality only for semantically equal trees.
5138 2017-10-09 Richard Biener <rguenther@suse.de>
5140 PR tree-optimization/82449
5141 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5142 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5143 allow constant addresses.
5144 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5147 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5149 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5152 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5155 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5158 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5161 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5163 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5166 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5167 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5169 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5171 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5172 (WI_BINARY_PREDICATE_RESULT): Likewise.
5173 (wi::binary_traits::operator_result): New type.
5174 (wi::binary_traits::predicate_result): Likewise.
5175 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5176 (generic_wide_int::operator==, generic_wide_int::operator!=)
5177 (generic_wide_int::operator&, generic_wide_int::and_not)
5178 (generic_wide_int::operator|, generic_wide_int::or_not)
5179 (generic_wide_int::operator^, generic_wide_int::operator+
5180 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5182 (operator~, unary operator-, operator==, operator!=, operator&)
5183 (operator|, operator^, operator+, binary operator-, operator*): New
5185 * expr.c (get_inner_reference): Use wi::bit_and_not.
5186 * fold-const.c (fold_binary_loc): Likewise.
5187 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5188 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5189 (bit_value_binop): Likewise.
5190 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5191 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5192 (extract_range_from_binary_expr_1): Likewise.
5193 (masked_increment): Likewise.
5194 (simplify_bit_ops_using_ranges): Likewise.
5196 2017-10-09 Martin Jambor <mjambor@suse.cz>
5199 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5200 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5201 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5203 (hsa_fixup_mov_insn_type): New function.
5204 (hsa_op_with_type::get_in_type): Use it.
5205 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5206 immediates in an assert.
5207 (hsa_op_with_type::extend_int_to_32bit): New method.
5208 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5209 types. Convert to dest type if necessary.
5210 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5211 (reg_for_gimple_ssa): Pass false as min32int to
5212 hsa_type_for_scalar_tree_type.
5213 (gen_hsa_addr): Fixup type when creating addresable temporary.
5214 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5215 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5216 necessary. Call hsa_fixup_mov_insn_type.
5217 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5218 extend operands and convert to dest type if necessary.
5219 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5220 to dest type if necessary.
5221 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5222 if conversion nt necessary and size matches.
5223 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5224 to dest type if necessary.
5225 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5226 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5228 (gen_hsa_clrsb): Likewise.
5229 (gen_hsa_ffs): Likewise.
5230 (gen_hsa_divmod): Extend operands and convert to dest type if
5232 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5234 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5236 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5237 Remove empty default arguments. Use a brace block as output
5239 (conditional return): Ditto.
5241 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5242 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5244 (group_ending_nop): Ditto.
5245 (doloop_end): Ditto.
5246 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5247 (splitters for those): Ditto.
5249 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5251 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5252 a conditional jump (and the compare for it) so that pc_rtx is the
5254 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5255 for the deleted and renamed ctr<mode>_internal[234] patterns.
5256 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5257 Delete second conditional return pattern.
5258 (ctr<mode>_internal2): Delete this second bdnz pattern.
5259 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5260 (ctr<mode>_internal4): Delete this second bdz pattern.
5262 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5264 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5265 (always_initialized_rtx_for_ssa_name_p): New predicate.
5266 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5267 (finish_out_of_ssa): Free new field of SA.
5268 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5269 * tree-ssa-coalesce.c: Include tree-ssa.h.
5270 (get_parm_default_def_partitions): Remove extern keyword.
5271 (get_undefined_value_partitions): New function.
5272 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5273 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5275 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5277 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5279 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5281 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5283 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5284 for targets that preffer 128bit.
5286 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5288 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5290 2017-10-08 Olivier Hainque <hainque@adacore.com>
5292 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5293 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5295 (thumb_set_return_address): Likewise.
5297 2017-10-08 Olivier Hainque <hainque@adacore.com>
5299 * common/config/arm/arm-common.c (arm_except_unwind_info):
5300 Handle DWARF2_UNWIND_INFO.
5302 2017-10-07 Michael Collison <michael.collison@arm.com>
5304 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5307 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5309 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5310 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5311 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5312 defined, force the creation of a new block for a dispatch label.
5314 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5316 * invoke.texi (Wsuggest-attribute=cold): Document.
5317 * common.opt (Wsuggest-attribute=cold): New
5318 * ipa-pure-const.c (warn_function_cold): New function.
5319 * predict.c (compute_function_frequency): Use it.
5320 * predict.h (warn_function_cold): Declare.
5322 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5324 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5327 2017-10-06 Martin Liska <mliska@suse.cz>
5329 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5330 keyword for member functions.
5331 (struct sanopt_tree_couple): New struct.
5332 (struct sanopt_tree_couple_hash): New function.
5333 (struct sanopt_ctx): Add new hash_map.
5334 (has_dominating_ubsan_ptr_check): New function.
5335 (record_ubsan_ptr_check_stmt): Likewise.
5336 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5337 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5338 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5340 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5343 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5344 aarch64_simd_valid_immediate on CONST_VECTORs.
5345 (aarch64_reg_or_bic_imm): Likewise.
5347 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5349 PR rtl-optimization/82396
5350 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5352 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5354 * graphite-dependences.c (scop_get_reads): Move code to...
5355 (scop_get_must_writes): Move code to...
5356 (scop_get_may_writes): Move code to...
5357 (scop_get_reads_and_writes): ... here.
5358 (scop_get_dependences): Call scop_get_reads_and_writes.
5360 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5362 PR tree-optimization/82434
5363 * fold-const.h (can_native_encode_type_p,
5364 can_native_encode_string_p): Remove.
5365 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5366 don't encode anything, just return what would be otherwise returned.
5367 (native_encode_fixed, native_encode_complex, native_encode_vector):
5369 (native_encode_string): Likewise. Inline by hand
5370 can_native_encode_string_p.
5371 (can_native_encode_type_p): Remove.
5372 (can_native_encode_string_p): Remove.
5373 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5374 STRING_CSTs using can_native_encode_string_p, test all
5375 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5376 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5377 argument from native_encode_expr.
5378 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5379 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5382 2017-10-06 Richard Biener <rguenther@suse.de>
5384 PR tree-optimization/82397
5385 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5386 operand_equal_p but rely on data_ref_compare_tree for detecting
5388 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5389 to match up with dr_group_sort_cmp.
5391 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5394 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5396 * config/s390/s390-builtin-types.def: Regenerate.
5398 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5401 * config/s390/s390-builtin-types.def: Regenerate.
5402 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5403 Change flag from B_VXE to B_VX.
5404 (s390_vec_min_dbl): Remove B_VXE flag.
5406 2017-10-06 Richard Biener <rguenther@suse.de>
5408 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5409 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5410 translate_isl_ast_to_gimple::is_valid_rename,
5411 translate_isl_ast_to_gimple::get_rename,
5412 translate_isl_ast_to_gimple::get_def_bb_for_const,
5413 translate_isl_ast_to_gimple::get_new_name,
5414 translate_isl_ast_to_gimple::collect_all_ssa_names,
5415 translate_isl_ast_to_gimple::copy_loop_phi_args,
5416 translate_isl_ast_to_gimple::collect_all_ssa_names,
5417 translate_isl_ast_to_gimple::copy_loop_phi_args,
5418 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5419 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5420 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5421 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5422 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5423 translate_isl_ast_to_gimple::copy_cond_phi_args,
5424 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5425 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5426 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5427 translate_isl_ast_to_gimple::rename_uses,
5428 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5429 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5430 (set_rename_for_each_def): Likewise.
5431 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5432 here. Handle rewriting SCEV analyzable uses here.
5433 (copy_bb_and_scalar_dependences): Generate code for PHI
5435 (graphite_regenerate_ast_isl): Adjust.
5436 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5437 (add_write, add_read): New functions.
5438 (build_cross_bb_scalars_def): Use it and simplify.
5439 (build_cross_bb_scalars_use): Likewise.
5440 (graphite_find_cross_bb_scalar_vars): Inline into...
5441 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5442 simulating out-of-SSA. Compute liveout and add dependencies.
5443 (build_scops): Force an empty entry block.
5444 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5446 (sese_build_liveouts): Declare.
5447 (sese_trivially_empty_bb_p): Likewise.
5448 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5449 compute liveout and debug_liveout.
5450 (sese_bad_liveouts_use): Remove.
5451 (sese_reset_debug_liveouts_bb): Likewise.
5452 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5453 (sese_build_liveouts): Build liveout and debug_liveout and store
5455 (new_sese_info): Adjust.
5456 (free_sese_info): Likewise.
5457 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5458 do not build liveout here.
5459 (move_sese_in_condition): Adjust region entry.
5460 (scev_analyzable_p): Match up with chrec_apply requirements.
5461 (sese_trivially_empty_bb_p): New.
5462 * tree-into-ssa.c (get_reaching_def): Properly support generating
5463 default-defs for incremental rewrite of anonymous names.
5465 2017-10-06 Richard Biener <rguenther@suse.de>
5467 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5468 precision do not perform modulo reduction.
5470 2017-10-06 Richard Biener <rguenther@suse.de>
5472 PR tree-optimization/82436
5473 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5474 conservatively choose the vectorization factor when checking
5475 whether we can perform the required load permutation.
5476 (vect_transform_slp_perm_load): Assert when we may not fail.
5478 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5480 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5481 message for incompatible -msdata=* and -mcall-* options.
5483 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5485 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5486 rate for post-reload scheduling.
5488 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5490 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5492 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5494 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5495 to improve monte carlo in scimark.
5497 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5499 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5500 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5501 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5502 pentium4_cost, nocona_cost): Set reassociation width to 1.
5503 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5504 width to 2 for fp operations and 1 otherwise.
5505 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5507 (atom_cost): Set reassociation width to 2.
5508 (slm_cost, generic_cost): Set fp reassociation width
5509 to 2 and 1 otherwise.
5510 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5511 (core_cost): Set fp reassociation width to 4 and vector to 2.
5512 (ix86_reassociation_width): Rewrite using cost table; special case
5513 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5514 and TARGET_AVX128_OPTIMAL.
5515 * config/i386/i386.h (processor_costs): Add
5516 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5517 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5518 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5519 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5520 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5521 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5523 2017-10-05 Nathan Sidwell <nathan@acm.org>
5525 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5527 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5529 * config/arm/arm.c (arm_test_fpu_data): New.
5530 (arm_run_selftests): Call arm_test_fpu_data.
5532 2017-10-04 Nathan Sidwell <nathan@acm.org>
5534 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5536 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5538 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5540 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5541 check type for aarch64_simd_valid_immediate.
5542 (aarch64_output_simd_mov_immediate): Update prototype.
5543 (aarch64_simd_valid_immediate): Update prototype.
5544 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5545 support for ORR-immediate.
5546 (and<mode>3): modified pattern to add support for BIC-immediate.
5547 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5548 now checks for valid immediate for BIC and ORR based on new enum
5550 (aarch64_output_simd_mov_immediate): Function now used to output
5551 BIC/ORR imm as well based on new enum argument.
5552 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5554 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5555 (aarch64_reg_or_bic_imm): Likewise.
5557 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5559 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5560 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5562 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5566 PR rtl-optimization/82396
5567 * haifa-sched.c (autopref_multipass_init): Simplify
5569 (autopref_rank_data): Simplify sort order.
5570 * sched-int.h (autopref_multipass_data_): Remove
5571 multi_mem_insn_p, min_offset and max_offset.
5573 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5575 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5577 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5579 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5582 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5584 * doc/sourcebuild.texi: Document vect_long_mult.
5586 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5588 PR tree-optimization/82413
5589 * fold-const.c (build_range_check): Use widest_int when comparing
5590 the maximum ETYPE value with HIGH.
5592 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5594 PR rtl-optimization/82396
5595 * haifa-sched.c (autopref_multipass_init): Simplify
5597 (autopref_rank_data): Simplify sort order.
5598 * sched-int.h (autopref_multipass_data_): Remove
5599 multi_mem_insn_p, min_offset and max_offset.
5601 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5603 PR tree-optimization/82381
5604 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5605 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5608 PR tree-optimization/82374
5609 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5610 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5611 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5612 current_function_decl to the new decl.
5614 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5616 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5617 helper macro for IEEE float128 hardware built-in functions.
5618 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5620 (TRUNCF128_ODD): Likewise.
5621 (ADDF128_ODD): Likewise.
5622 (SUBF128_ODD): Likewise.
5623 (MULF128_ODD): Likewise.
5624 (DIVF128_ODD): Likewise.
5625 (FMAF128_ODD): Likewise.
5626 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5627 UNSPEC_TRUNC_ROUND_TO_ODD.
5628 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5629 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5630 floating point round to odd instructions.
5631 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5632 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5633 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5634 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5635 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5636 (trunc<mode>sf2_hw): Change the truncate with round to odd
5637 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5638 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5639 to odd hardware instructions.
5640 (sub<mode>3_odd): Likewise.
5641 (mul<mode>3_odd): Likewise.
5642 (div<mode>3_odd): Likewise.
5643 (sqrt<mode>2_odd): Likewise.
5644 (fma<mode>4_odd): Likewise.
5645 (fms<mode>4_odd): Likewise.
5646 (nfma<mode>4_odd): Likewise.
5647 (nfms<mode>4_odd): Likewise.
5648 (trunc<mode>df2_odd): Change the truncate with round to odd
5649 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5651 * doc/extend.texi (PowerPC built-in functions): Update documentation
5652 for existing IEEE float128-bit built-in functions. Add built-in
5653 functions that generate the IEEE 128-bit floating point round to
5656 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5658 PR rtl-optimization/77729
5659 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5660 to (X&(C1&~C2))|C2 transformations.
5662 2017-10-03 Martin Jambor <mjambor@suse.cz>
5664 PR tree-optimization/82363
5665 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5666 mismatch, mark lacc written regardless of racc.
5668 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5670 PR tree-optimization/82381
5671 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5672 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5673 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5674 fallthrough into reassoc_stmt_dominates_stmt_p.
5677 * combine.c (combine_instructions): Don't combine in unreachable
5680 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5683 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5684 function to not use the have_cpu variable. Do not set cpu_index,
5685 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5687 (rs6000_valid_attribute_p): Remove duplicate initializations of
5688 old_optimize and func_optimize.
5689 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5690 (rs6000_activate_target_options): Make global.
5691 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5694 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5696 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5697 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5698 Set *poffset to 0 and *psize and *pmax_size to -1 if
5699 *poffset + *psize overflows in HOST_WIDE_INT.
5701 PR tree-optimization/82387
5702 PR tree-optimization/82388
5703 PR tree-optimization/82389
5704 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5705 instead of live_bytes non-NULL.
5707 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5710 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5713 2017-10-02 Richard Biener <rguenther@suse.de>
5715 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5716 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5717 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5719 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5721 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5722 requested precision matches the type's.
5723 * calls.c (alloc_max_size): Calculate the new candidate size as
5724 a widest_int and use wi::to_widest when comparing it with the
5725 current candidate size.
5726 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5727 zero rather than integer_zero_node.
5728 * match.pd: Check for a no-op conversion before using wi::add
5729 rather than after. Use tree_to_uhwi when summing small shift
5730 counts into an unsigned int.
5732 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5733 Alan Hayward <alan.hayward@arm.com>
5734 David Sherwood <david.sherwood@arm.com>
5737 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5738 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5739 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5740 POINTER_AND_FP_REGS.
5742 2017-10-02 Richard Biener <rguenther@suse.de>
5744 PR tree-optimization/82355
5745 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5746 a mapping for the enclosing loop but avoid generating one for
5748 (copy_bb_and_scalar_dependences): Remove premature codegen
5749 error on PHIs in blocks duplicated into multiple places.
5750 * graphite-scop-detection.c
5751 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5752 in the region use it as loop and nest to analyze the DR in.
5753 (try_generate_gimple_bb): Likewise.
5754 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5755 (add_loop_constraints): For blocks in a loop not in the region
5756 create a dimension with a single iteration.
5757 * sese.h (gbb_loop_at_index): Remove assert.
5759 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5761 * omp-expand.c (adjust_context_scope): New function.
5762 (expand_parallel_call): Call adjust_context_scope.
5764 2017-10-01 Jeff Law <law@redhat.com>
5766 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5767 dom_opt_dom_walker class with direct access to private members.
5768 Add comments. Call test_for_singularity.
5769 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5770 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5771 m_dummy_cond anymore.
5772 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5774 (pass_dominator:execute): Build the dummy_cond here and pass it
5775 to the dom_opt_dom_walker ctor.
5776 (test_for_singularity): New function.
5778 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5779 Maya Rashish <coypu@sdf.org>
5781 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5782 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5783 (alpha*-*-netbsd*) Use nbsd_tm_file.
5784 (arm*-*-netbsdelf*) Likewise.
5785 (i[34567]86-*-netbsdelf*) Likewise.
5786 (x86_64-*-netbsd*) Likewise.
5787 (mips*-*-netbsd*) Likewise.
5788 (powerpc-*-netbsd*) Likewise.
5789 (sh*-*-netbsd*) Likewise.
5790 (sparc-*-netbsdelf*) Likewise.
5791 (sparc64-*-netbsd*) Likewise.
5792 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5794 (vax-*-netbsdelf*) Likewise.
5795 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5796 (UINT_FAST8_TYPE) Likewise.
5797 (INT_FAST16_TYPE) Check CHAR_FAST16.
5798 (UINT_FAST16_TYPE) Likewise.
5800 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5803 * config/i386/i386.md
5804 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5805 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5806 *divmodsi4_zext_2): New define_insn_and_split.
5807 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5808 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5809 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5810 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5811 New define_insn_and_split.
5812 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5813 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5814 operands[1] having DImode when mode is SImode.
5816 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5817 always SImode for DIV and MOD in REG_EQUAL notes.
5819 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5822 * match.pd: Fix handling of NaNs in pattern.
5824 2017-09-29 Jeff Law <law@redhat.com>
5826 * sbitmap.c (bitmap_bit_in_range_p): New function.
5827 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5828 * tree-ssa-dse.c (live_bytes_read): New function.
5829 (dse_classify_store): Ignore reads of dead bytes.
5831 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5832 typos and whitespace errors.
5833 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5834 * config/s390/s390.c (s390_emit_prologue): Likewise.
5836 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5839 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5840 with a symbol for LRA.
5842 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5844 PR rtl-optimization/82338
5845 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5847 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5849 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5850 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5851 (qsort_chk): Declare.
5852 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5853 (qsort_chk): New function.
5855 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5857 PR tree-optimization/82337
5858 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5859 phi definition if the PHI result appears in an abnormal PHI.
5860 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5861 basis appears in an abnormal PHI.
5863 2017-09-29 Richard Biener <rguenther@suse.de>
5865 * graphite-isl-ast-to-gimple.c
5866 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5867 (binary_op_to_tree): Use it.
5868 (get_rename_from_scev): Likewise.
5869 (copy_loop_phi_nodes): Likewise.
5870 (copy_bb_and_scalar_dependences): Likewise.
5871 (translate_pending_phi_nodes): Likewise.
5873 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5876 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5877 for movabsq $(i32 << shift), r64.
5879 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5881 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5882 index when encoding %esp as %rsp to avoid 0x67 prefix.
5884 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5886 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5887 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5889 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5891 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5892 Extensions with more than 16 double VFP registers.
5893 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5894 to_clear_mask and all code related to it. Replace the remaining
5895 entry by a sbitmap and adapt code accordingly.
5897 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5899 * brig-builtins.def: Change pure attributes to const.
5901 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5903 * config.gcc (default_gnu_indirect_function): Default to yes for
5904 sparc*-*-linux* with glibc.
5906 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5908 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5909 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5910 when creating .init_array and .fini_array sections with priority
5913 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5916 * config/aarch64/aarch64.c
5917 (aarch64_builtin_support_vector_misalignment): Always return false
5918 when misalignment is unknown.
5920 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5922 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5923 this function to return false if the definition used by the swap
5924 instruction is artificial, or if the memory address from which the
5925 constant value is loaded is not represented by a base address held
5926 in a register or if the base address register is a frame or stack
5927 pointer. Additionally, return false if the base address of the
5928 loaded constant is a SYMBOL_REF but is not considered to be a
5930 (replace_swapped_load_constant): New function.
5931 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5932 loaded constant vector with a load of a swapped constant vector.
5934 2017-09-27 Carl Love <cel@us.ibm.com>
5936 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5937 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5938 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5941 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5943 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5944 first, always call autopref_rank_data otherwise.
5946 2017-09-27 Richard Biener <rguenther@suse.de>
5948 * graphite-scop-detection.c (find_scop_parameters): Move
5949 loop bound handling ...
5950 (gather_bbs::before_dom_children): ... here, avoiding the need
5951 to build scop_info->loop_nest.
5952 (record_loop_in_sese): Remove.
5953 * sese.h (sese_info_t::loop_nest): Remove.
5954 * sese.c (new_sese_info): Do not allocate loop_nest.
5955 (free_sese_info): Do not free loop_nest.
5957 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5960 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5961 lhs from calls if the lhs has addressable type.
5963 2017-09-27 Richard Biener <rguenther@suse.de>
5965 * graphite.h (scop::max_alias_set): New member.
5966 * graphite-scop-detection.c: Remove references to non-existing
5967 --param in comments.
5968 (build_alias_sets): Record the maximum alias set used for drs.
5969 (build_scops): Support zero as unlimited for
5970 --param graphite-max-arrays-per-scop.
5971 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5973 (build_poly_sr_1): ... here. Compute alias set based on the
5974 maximum alias set used for drs rather than
5975 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5977 2017-09-27 Richard Biener <rguenther@suse.de>
5979 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5980 --param loop-block-tile-size=0 to disable tiling.
5982 2017-09-27 Richard Biener <rguenther@suse.de>
5984 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5985 (graphite-max-nb-scop-params): Document special value zero.
5986 * domwalk.h (dom_walker::STOP): New symbolical constant.
5987 (dom_walker::dom_walker): Add optional parameter for bb to
5989 (dom_walker::~dom_walker): Declare.
5990 (dom_walker::before_dom_children): Document STOP return value.
5991 (dom_walker::m_user_bb_to_rpo): New member.
5992 (dom_walker::m_bb_to_rpo): Likewise.
5993 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5994 mapping here if not provided by the user.
5995 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5996 provided by the user.
5997 (dom_walker::STOP): Define.
5998 (dom_walker::walk): Do not compute bb to RPO mapping here.
5999 Support STOP return value from before_dom_children to stop
6001 * graphite-optimize-isl.c (optimize_isl): If the schedule
6002 is the same still generate code if -fgraphite-identity
6003 or -floop-parallelize-all are given.
6004 * graphite-scop-detection.c: Include cfganal.h.
6005 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6007 (gather_bbs::before_dom_children): Return STOP for BBs
6009 (build_scops): Compute bb to RPO mapping and pass it to
6010 the domwalk. Treat --param graphite-max-nb-scop-params=0
6011 as not limiting the number of params.
6012 * graphite.c (graphite_initialize): Remove limit on the
6013 number of basic-blocks in a function.
6014 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6015 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6016 default value of 10.
6018 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6020 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6021 Adjust code to eliminate needing to do the shift right 32-bits
6022 operation after XSCVDPSPN.
6024 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6026 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6027 ((X / Y) != 0 -> X >= Y): Likewise.
6029 2017-09-26 Carl Love <cel@us.ibm.com>
6031 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6032 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6033 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6034 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6035 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6036 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6037 definitions and overloading.
6038 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6039 statement for P9V_BUILTIN_XST_LEN_R.
6040 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6041 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6042 define_expand and define_insn for the instructions and builtins.
6043 * doc/extend.texi: Update the built-in documentation file for the new
6045 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6046 define_insn for the instructions
6048 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6051 * gcc/config/netbsd-protos.h: New file.
6052 * gcc/config/netbsd.c: New file.
6053 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6054 * gcc/config/t-netbsd: New file.
6055 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6056 (tmake_file) Add t-netbsd.
6057 (extra_objs) Add netbsd.o.
6059 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6063 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6065 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6067 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6068 sign extension from a vector register to a GPR by doing a 32-bit
6069 direct move and then an EXTSW.
6070 (extendsi<mode>2 splitter): Likewise.
6071 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6072 right or vector extract after doing XSCVDPSPN. Use
6073 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6075 (movdi_from_sf_zero_ext): Likewise.
6076 (reload_gpr_from_vsxsf): Likewise.
6077 (p8_mfvsrd_4_disf): Delete, no longer used.
6078 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6079 then needing to move the SFmode to a GPR to use the XSCVDPSP
6080 instruction instead of FRSP and XSCVDPSPN.
6081 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6082 it is adjacent to the other XSCVSPDP insns.
6083 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6084 SFmode to be in traditional Altivec registers.
6085 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6086 (vsx_xscvspdpn): Likewise.
6087 (vsx_xscvspdpn_scalar): Likewise.
6089 2017-09-26 Martin Jambor <mjambor@suse.cz>
6091 * tree-sra.c (compare_access_positions): Put integral types first,
6092 stabilize sorting of integral types, remove conditions putting
6093 non-full-precision integers last.
6094 (sort_and_splice_var_accesses): Disable scalarization if a
6095 non-integert would be represented by a non-full-precision integer.
6097 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6099 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6100 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6101 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6102 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6103 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6104 conditionals inside the function instead of around it. Call
6105 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6106 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6108 2017-09-26 Richard Biener <rguenther@suse.de>
6110 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6112 (scop_detection::build_scop_breadth): ... this. Removed.
6113 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6114 (scop_detection::harmful_stmt_in_bb): Likewise.
6115 (scop_detection::graphite_can_represent_stmt): Likewise.
6116 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6117 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6118 (scop_detection::can_represent_loop_1): ... this. Removed.
6119 (scop_detection::harmful_loop_in_region): Simplify after inlining
6120 the above and remove more quadraticness.
6121 (build_scops): Adjust.
6122 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6125 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6128 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6129 REGNO (base) == SP_REG if base is a REG.
6132 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6133 if it is different SSA_NAME.
6134 (optimize_range_tests_cmp_bitwise): New function.
6135 (optimize_range_tests): Call it.
6137 2017-09-26 Richard Biener <rguenther@suse.de>
6139 PR tree-optimization/82321
6140 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6141 for the def being inside the loop.
6143 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6145 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6147 * config/s390/s390-builtins.def: Fix constraint on op4.
6149 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6151 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6152 independent expanders.
6153 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6154 ("vec_ordered", "vec_unordered"): New expanders.
6156 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6158 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6161 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6163 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6164 vec_unpacks_lo_v16qi.
6165 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6167 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6169 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6170 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6171 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6173 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6175 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6177 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6179 ("*vec_slb<mode>"): New insn pattern.
6180 ("vec_shr_<mode>"): New expander.
6181 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6182 and force the shift count operand to V16QImode.
6183 ("vec_srb<mode>"): Set shift count mode to V16QI.
6185 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6187 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6188 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6189 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6191 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6194 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6196 2017-09-26 Richard Biener <rguenther@suse.de>
6198 PR tree-optimization/82320
6199 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6202 2017-09-25 Jeff Law <law@redhat.com>
6204 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6205 prototype for new argument.
6206 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6207 mostly extracted from rs6000_emit_allocate_stack.
6208 (rs6000_emit_probe_stack_range_stack_clash): New function.
6209 (rs6000_emit_allocate_stack): Call
6210 rs6000_emit_probe_stack_range_stack_clash as needed.
6211 (rs6000_emit_probe_stack_range): Add additional argument
6212 to call to gen_probe_stack_range{si,di}.
6213 (output_probe_stack_range): New.
6214 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6215 (output_probe_stack_range_stack_clash): New.
6216 (rs6000_emit_prologue): Emit notes into dump file as requested.
6217 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6218 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6219 Add additional operand and pass it to output_probe_stack_range.
6221 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6223 PR tree-optimization/82163
6224 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6225 (checking_verify_loop_closed_ssa): New parameter.
6226 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6227 (check_loop_closed_ssa_stmt): Delete.
6228 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6229 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6230 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6233 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6235 * brig-builtins.def: Treat HSAIL barrier builtins as
6236 setjmp/longjump style functions.
6238 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6240 * target.def (constant_alignment): New hook.
6241 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6242 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6243 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6244 * doc/tm.texi: Regenerate.
6245 * targhooks.h (default_constant_alignment): Declare.
6246 (constant_alignment_word_strings): Likewise.
6247 * targhooks.c (default_constant_alignment): New function.
6248 (constant_alignment_word_strings): Likewise.
6249 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6250 instead of CONSTANT_ALIGNMENT.
6251 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6252 (force_const_mem): Likewise.
6253 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6254 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6255 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6256 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6257 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6259 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6260 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6261 constant_alignment_word_strings.
6262 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6263 (CONSTANT_ALIGNMENT): Likewise.
6264 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6265 (arm_constant_alignment): New function.
6266 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6267 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6268 constant_alignment_word_strings.
6269 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6270 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6271 constant_alignment_word_strings.
6272 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6273 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6274 (cris_constant_alignment): New function.
6275 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6276 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6277 (epiphany_constant_alignment): New function.
6278 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6279 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6280 constant_alignment_word_strings.
6281 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6282 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6283 constant_alignment_word_strings.
6284 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6285 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6286 constant_alignment_word_strings.
6287 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6288 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6289 * config/i386/i386.c (ix86_constant_alignment): Make static.
6290 Use the same interface as the target hook.
6291 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6292 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6293 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6294 constant_alignment_word_strings.
6295 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6296 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6297 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6298 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6299 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6300 constant_alignment_word_strings.
6301 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6302 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6303 constant_alignment_word_strings.
6304 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6305 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6306 constant_alignment_word_strings.
6307 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6308 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6310 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6311 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6312 * config/mips/mips.c (mips_constant_alignment): New function.
6313 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6314 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6315 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6316 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6317 (mmix_constant_alignment): Make static. Use the same interface
6319 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6320 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6321 constant_alignment_word_strings.
6322 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6323 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6324 constant_alignment_word_strings.
6325 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6326 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6327 constant_alignment_word_strings.
6328 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6329 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6330 (rs6000_constant_alignment): New function.
6331 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6332 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6333 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6334 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6335 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6336 (rs6000_constant_alignment): New function.
6337 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6338 * config/s390/s390.c (s390_constant_alignment): New function.
6339 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6340 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6341 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6342 constant_alignment_word_strings.
6343 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6344 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6345 (sparc_constant_alignment): New function.
6346 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6347 * config/spu/spu.c (spu_constant_alignment): New function.
6348 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6349 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6350 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6351 constant_alignment_word_strings.
6352 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6353 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6354 constant_alignment_word_strings.
6355 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6356 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6357 constant_alignment_word_strings.
6358 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6359 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6360 (visium_constant_alignment): New function.
6361 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6362 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6363 (xtensa_constant_alignment): New function.
6364 * system.h (CONSTANT_ALIGNMENT): Poison.
6366 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6368 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6369 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6370 (rs6000_builtin_valid_without_lhs): New helper function.
6371 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6372 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6374 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6376 * target.h (vec_perm_indices): Use unsigned short rather than
6378 (auto_vec_perm_indices): Likewise.
6379 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6380 Use unsigned int rather than unsigned char.
6381 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6383 2017-09-25 Richard Biener <rguenther@suse.de>
6385 * cfgloop.h (sort_sibling_loops): Declare.
6386 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6387 (sort_sibling_loops): New function sorting the sibling loop list
6389 * graphite.c (graphite_transform_loops): Sort sibling loops.
6391 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6393 * target.def (vec_perm_const_ok): Change sel parameter to
6395 * optabs-query.c (can_vec_perm_p): Update accordingly.
6396 * doc/tm.texi: Regenerate.
6397 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6398 auto_vec_perm_indices and remove separate nelt field.
6399 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6400 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6401 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6402 (aarch64_expand_vec_perm_const): Update accordingly.
6403 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6404 to vec_perm_indices.
6405 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6406 auto_vec_perm_indices and remove separate nelt field.
6407 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6408 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6409 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6411 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6412 to vec_perm_indices.
6413 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6414 sel to vec_perm_indices.
6415 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6416 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6417 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6419 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6422 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6425 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6426 on the FUNCTION_DECL function context if it has a DIE that is a
6429 2017-09-25 Richard Biener <rguenther@suse.de>
6431 PR tree-optimization/82285
6432 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6435 2017-09-25 Tom de Vries <tom@codesourcery.com>
6439 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6442 2017-09-25 Richard Biener <rguenther@suse.de>
6444 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6445 ISL errors other than isl_error_quota happen. Dump if the
6446 schedule is the same.
6447 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6448 errors instead of aborting inside ISL.
6450 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6453 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6454 of libgcc_eh for m64.
6455 * config/i386/darwin64.h: Likewise.
6457 2017-09-25 Richard Biener <rguenther@suse.de>
6460 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6461 attribute for incomplete types nor twice for complete ones.
6463 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6466 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6467 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6470 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6473 * config/i386/i386.opt (mprefer-avx256): Use
6474 ix86_target_flags variable.
6475 * config/i386/i386.c (ix86_target_string): Move
6476 -mprefer-avx256 to flag2_opts.
6478 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6481 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6482 and x != -1 | y != -1 into (x & y) != -1.
6484 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6486 * config.gcc: Add new case statement to set
6487 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6488 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6489 s390x-*-linux* case statements. Added aarch64 to the list of
6490 supported architectures.
6492 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6494 PR tree-optimization/82289
6495 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6496 STMT_VINFO_RELEVANT_P.
6498 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6499 Alan Hayward <alan.hayward@arm.com>
6500 David Sherwood <david.sherwood@arm.com>
6502 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6503 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6504 (extract_range_from_binary_expr_1): Don't call
6505 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6507 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6508 Alan Hayward <alan.hayward@arm.com>
6509 David Sherwood <david.sherwood@arm.com>
6511 * target.def (preferred_vector_alignment): New hook.
6512 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6514 * doc/tm.texi: Regenerate.
6515 * targhooks.h (default_preferred_vector_alignment): Declare.
6516 * targhooks.c (default_preferred_vector_alignment): New function.
6517 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6519 (DR_TARGET_ALIGNMENT): New macro.
6520 (aligned_access_p): Update commentary.
6521 (vect_known_alignment_in_bytes): New function.
6522 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6524 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6525 Calculate the misalignment based on the target alignment rather than
6527 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6528 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6529 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6530 the target alignment, rather than masking the element misalignment
6531 with the number of elements in a vector. Also use the target
6532 alignment when calculating the maximum number of peels.
6533 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6534 instead of TYPE_ALIGN_UNIT.
6535 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6536 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6537 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6538 (vect_create_data_ref_ptr): Likewise.
6539 (vect_setup_realignment): Realign by ANDing with
6540 -DR_TARGET_MISALIGNMENT.
6541 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6542 the number of peels based on DR_TARGET_ALIGNMENT.
6543 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6544 with the guaranteed alignment boundary when deciding whether
6546 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6547 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6548 (ensure_base_align): Remove stmt_info parameter. Get the
6549 target base alignment from DR_TARGET_ALIGNMENT.
6550 (vectorizable_store): Update call accordingly. Interpret
6551 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6553 (vectorizable_load): Likewise.
6555 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6556 Alan Hayward <alan.hayward@arm.com>
6557 David Sherwood <david.sherwood@arm.com>
6559 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6560 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6561 (vect_enhance_data_refs_alignment): Likewise.
6563 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6565 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6566 error. Only quit immediately if parsing is complete.
6567 (BEGIN): Initialize fatal_err and parse_done.
6568 (begin fpu, end fpu): Check number of arguments.
6569 (begin arch, end arch): Likewise.
6570 (begin cpu, end cpu): Likewise.
6571 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6572 (optalias): Likewise.
6574 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6576 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6577 * config/arm/arm-isa.h: Delete. Move definitions to ...
6578 * arm-cpus.in: ... here. Use new feature and fgroup values.
6579 * config/arm/arm.c (arm_option_override): Use lower case for feature
6581 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6582 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6583 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6585 (print_isa_bits_for): New function.
6586 (gen_isa): New function.
6587 (gen_comm_data): Use print_isa_bits_for.
6588 (define feature): New keyword.
6589 (define fgroup): New keyword.
6590 * config/arm/t-arm (TM_H): Remove.
6591 (GTM_H): Add arm-isa.h.
6592 (arm-isa.h): Add rule to generate file.
6593 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6594 case for feature bit names.
6596 2017-09-22 Richard Biener <rguenther@suse.de>
6598 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6600 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6601 print of no dependency loops ...
6602 * graphite.c (graphite_transform_loops): ... here.
6603 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6605 (same_close_phi_node, remove_duplicate_close_phi,
6606 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6607 (canonicalize_loop_closed_ssa): ... here and simplify.
6608 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6609 (optimize_isl): Use dump_printf_loc to tell when we stopped
6610 optimizing because of an ISL timeout.
6612 2017-09-22 Richard Biener <rguenther@suse.de>
6614 PR tree-optimization/82291
6615 * tree-if-conv.c (predicate_mem_writes): Make sure to
6616 remove writes in blocks predicated with false.
6618 2017-09-22 Richard Biener <rguenther@suse.de>
6620 * sese.c: Include cfganal.h.
6621 (if_region_set_false_region): Remove.
6622 (create_if_region_on_edge): Likewise.
6623 (move_sese_in_condition): Re-implement without destroying
6626 2017-09-22 Richard Biener <rguenther@suse.de>
6628 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6629 Verify both BBs contain loop PHI nodes before dispatching to
6631 (graphite_regenerate_ast_isl): Do not recompute dominators,
6632 do not verify three times. Restructure for clarity.
6633 * graphite-scop-detection.c (same_close_phi_node,
6634 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6635 defined_in_loop_p, canonicalize_loop_closed_ssa,
6636 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6637 checking and SSA rewrite, move to ...
6638 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6639 (graphite_initialize): Do not pass in ctx, do not reset the
6640 SCEV cache, compute only dominators.
6641 (graphite_transform_loops): Allocate ISL ctx after
6642 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6643 Maintain post-dominators only around build_scops.
6644 * sese.c (if_region_set_false_region): Make static. Free
6645 and recompute dominators.
6646 (move_sese_in_condition): Assert we don't get called with
6647 post-dominators computed.
6648 * sese.h (if_region_set_false_region): Remove.
6650 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6652 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6653 mode attribute for TARGET_AVX512VL.
6655 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6657 * config/i386/i386.opt (mprefer-avx256): New option.
6658 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6660 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6661 for TARGET_PREFER_AVX256.
6662 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6664 2017-09-21 Jeff Law <law@redhat.com>
6666 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6667 Fix dump output if the only stack space is for pushed registers.
6669 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6671 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6674 2017-09-21 Martin Sebor <msebor@redhat.com>
6677 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6678 code (in C++) or code that triggers warnings.
6680 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6682 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6684 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6686 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6687 * sched-rgn.c: Ditto.
6688 * sel-sched-ir.c: Ditto.
6690 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6692 * toplev.h (set_random_seed): Adjust return type.
6693 * toplev.c (init_local_tick): Move eager initialization of random_seed
6694 to get_random_seed. Adjust comment.
6695 (init_random_seed): Inline to get_random_seed, delete.
6696 (get_random_seed): Initialize random_seed lazily.
6697 (set_random_seed): Do not return previous value.
6698 (print_switch_value): Do not call get_random_seed.
6700 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6702 * cgraph.c (delete_function_version): New, broken out from...
6703 (cgraph_node::delete_function_version): ...here. Rename to
6704 cgraph_node::delete_function_version_by_decl. Update all uses.
6705 (cgraph_node::remove): Call delete_function_version.
6707 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6710 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6711 VAR_DECLs to which addressable non-volatile parameters are mapped
6712 and for id->retvar after the return value assignment. Clear
6713 id->retval and id->retbnd after inlining.
6715 2017-09-21 Richard Biener <rguenther@suse.de>
6717 PR tree-optimization/82276
6718 PR tree-optimization/82244
6719 * tree-vrp.c (build_assert_expr_for): Set
6720 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6722 (remove_range_assertions): Revert earlier change.
6724 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6727 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6729 2017-09-21 Richard Biener <rguenther@suse.de>
6731 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6732 Restore valid IL after code generation errors.
6733 * graphite.c (graphite_transform_loops): Diagnose code
6734 generation issues as MSG_MISSED_OPTIMIZATION and continue
6735 with processing SCOPs.
6737 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6738 Alan Hayward <alan.hayward@arm.com>
6739 David Sherwood <david.sherwood@arm.com>
6741 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6742 rather than choosing between plus_constant and gen_rtx_<CODE>.
6743 * expr.c (emit_push_insn): Likewise.
6744 (expand_expr_real_2): Likewise.
6746 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6747 Alan Hayward <alan.hayward@arm.com>
6748 David Sherwood <david.sherwood@arm.com>
6750 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6752 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6753 Alan Hayward <alan.hayward@arm.com>
6754 David Sherwood <david.sherwood@arm.com>
6756 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6757 calling tree_to_uhwi.
6759 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6760 Alan Hayward <alan.hayward@arm.com>
6761 David Sherwood <david.sherwood@arm.com>
6763 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6764 INTEGER_CST rather than a negative test for ADDR_EXPR.
6766 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6767 Alan Hayward <alan.hayward@arm.com>
6768 David Sherwood <david.sherwood@arm.com>
6770 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6771 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6773 2017-09-21 Richard Biener <rguenther@suse.de>
6775 PR tree-optimization/71351
6776 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6777 graphite_create_new_loop_guard): Remove, fold remaining parts
6779 (translate_isl_ast_node_for): ... here and simplify.
6781 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6784 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6785 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6786 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6787 alternative always use QI mode, for -Os imov (=R,R) alternative
6788 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6791 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6792 Jeff Law <law@redhat.com>
6794 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6795 (allocate_stack_space): New function, partially extracted from
6797 (s390_emit_prologue): Track offset to most recent stack probe.
6798 Code to allocate space moved into allocate_stack_space.
6799 Dump actions when no stack is allocated.
6800 (s390_prologue_plus_offset): New function.
6801 (s390_emit_stack_probe): Likewise.
6803 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6805 * common.opt (Wa, Wl, Wp, g, gz=): Add
6807 (gno-column-info): Remove.
6808 (gcolumn-info): Drop RejectNegative.
6810 (gno-record-gcc-switches): Remove.
6811 (grecord-gcc-switches): Drop RejectNegative.
6812 (gno-split-dwarf): Remove.
6813 (gsplit-dwarf): Drop RejectNegative.
6814 (gno-strict-dwarf): Remove.
6815 (gstrict-dwarf): Drop RejectNegative.
6816 * config/darwin.opt (gfull, gused): Add RejectNegative.
6817 * dwarf2out.c (gen_producer_string): Drop
6818 gno-record-gcc-switches handler.
6819 * optc-gen.awk: Add g to prefixes with negative forms.
6820 * opts-common.c (remapping_prefix_p): New.
6821 (find_opt): Check it.
6822 (generate_canonical_option): Test g prefix.
6823 (option_map): Add -gno- mapping.
6824 (add_misspelling_candidates): Check remapping_prefix_p.
6826 2017-09-20 Jeff Law <law@redhat.com>
6828 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6829 thinko in stack clash protection support.
6831 * explow.c (compute_stack_clash_protection_loop_data): Use
6832 CONST_INT_P instead of explicit test. Verify object is a
6833 CONST_INT_P before looking at INTVAL.
6834 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6835 instead of explicit test.
6837 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6840 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6841 address instead of to r1 and r11.
6843 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6845 * config.gcc: Support "knm".
6846 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6847 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6849 * config/i386/i386.c (m_KNM): Define.
6850 (processor_target_table): Add "knm".
6852 (ix86_option_override_internal): Add "knm".
6853 (ix86_issue_rate): Add PROCESSOR_KNM.
6854 (ix86_adjust_cost): Ditto.
6855 (ia32_multipass_dfa_lookahead): Ditto.
6856 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6857 (fold_builtin_cpu): Add M_INTEL_KNM.
6858 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6859 (processor_type): Add PROCESSOR_KNM.
6860 * config/i386/x86-tune.def: Add m_KNM.
6861 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6863 2017-09-20 Richard Biener <rguenther@suse.de>
6865 PR tree-optimization/80213
6866 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6867 are allowed in empty BBs as well.
6868 (canonicalize_loop_closed_ssa): Also look for other complex
6870 (scop_detection::get_sese): Include the loop-closed PHI block
6872 (scop_detection::merge_sese): Remove code adding extra blocks.
6873 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6874 (build_scops): Assert the final returned scop is invalid.
6876 2017-09-20 Richard Biener <rguenther@suse.de>
6878 PR tree-optimization/82264
6879 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6881 (vn_phi_lookup): Likewise.
6882 (vn_phi_insert): Likewise.
6884 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6886 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6887 that fits into uhwi or shwi, add DW_AT_const_value regardless
6888 of early_dwarf without going through RTL, using add_AT_unsigned
6891 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6892 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6893 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6894 (DEBUG_MACRO_SECTION): Likewise.
6895 (DEBUG_LTO_MACRO_SECTION): Likewise.
6896 (DEBUG_STR_DWO_SECTION): Likewise.
6897 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6898 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6899 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6900 (DEBUG_LTO_LINE_STR_SECTION): Define.
6901 (init_sections_and_labels): Initialize debug_line_str_section
6902 variable. Initialize debug_loc_section for -gdwarf-5 to
6903 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6905 2017-09-20 Richard Biener <rguenther@suse.de>
6907 * graphite-sese-to-poly.c (extract_affine): Properly handle
6908 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6910 2017-09-20 Richard Biener <rguenther@suse.de>
6912 PR tree-optimization/81373
6913 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6914 Force SESE live-out defs to be handled even if they are
6917 2017-09-19 Jeff Law <law@redhat.com>
6919 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6920 nothing for stack adjustments with REG_STACK_CHECK.
6921 * sched-deps.c (parse_add_or_inc): Reject insns with
6922 REG_STACK_CHECK from dependency breaking.
6923 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6924 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6925 * reg-notes.def (STACK_CHECK): New note.
6927 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6928 (ix86_expand_prologue): Dump stack clash info as needed.
6929 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6931 * function.c (dump_stack_clash_frame_info): New function.
6932 * function.h (dump_stack_clash_frame_info): Prototype.
6933 (enum stack_clash_probes): New enum.
6935 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6936 flag_stack_clash_protection.
6937 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6938 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6939 (arm_frame_pointer_required): Likewise.
6940 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6941 (ia64_expand_prologue): Likewise.
6942 * config/mips/mips.c (mips_expand_prologue): Likewise.
6943 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6944 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6945 (sparc_flat_expand_prologue): Likewise.
6946 * config/spu/spu.c (spu_expand_prologue): Likewise.
6948 * explow.c: Include "params.h".
6949 (anti_adjust_stack_and_probe_stack_clash): New function.
6950 (get_stack_check_protect): Likewise.
6951 (compute_stack_clash_protection_loop_data): Likewise.
6952 (emit_stack_clash_protection_loop_start): Likewise.
6953 (emit_stack_clash_protection_loop_end): Likewise.
6954 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6955 Use anti_adjust_stack_and_probe_stack_clash.
6956 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6957 (emit_stack_clash_protection_loop_start): Likewise.
6958 (emit_stack_clash_protection_loop_end): Likewise.
6959 * rtl.h (get_stack_check_protect): Prototype.
6960 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6961 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6962 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6964 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6966 * doc/tm.texi: Rebuilt.
6967 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6968 get_stack_check_protect.
6969 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6970 * config/arm/arm.c (arm_expand_prologue): Likewise.
6971 (arm_frame_pointer_required): Likewise.
6972 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6973 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6974 * config/mips/mips.c (mips_expand_prologue): Likewise.
6975 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6976 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6977 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6978 (sparc_flat_expand_prologue): Likewise.
6980 * common.opt (-fstack-clash-protection): New option.
6981 * flag-types.h (enum stack_check_type): Note difference between
6982 -fstack-check= and -fstack-clash-protection.
6983 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6984 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6985 * toplev.c (process_options): Issue warnings/errors for cases
6986 not handled with -fstack-clash-protection.
6987 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6988 (-fstack-check): Note additional problem with -fstack-check=generic.
6989 Note that -fstack-check is primarily for Ada and refer users
6990 to -fstack-clash-protection for stack-clash-protection.
6991 Document new params for stack clash protection.
6993 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6995 * config/i386/i386.c (ix86_split_long_move): Do not handle
6996 address used for LEA in a special way.
6998 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7000 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7002 2017-09-19 Martin Sebor <msebor@redhat.com>
7005 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7006 of incompatible types.
7008 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7010 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7011 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7012 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7013 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7015 2017-09-19 Richard Biener <rguenther@suse.de>
7017 PR tree-optimization/82244
7018 * tree-vrp.c (remove_range_assertions): Do not propagate
7019 a constant to abnormals but replace the assert with a copy.
7021 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7023 PR rtl-optimization/57878
7024 PR rtl-optimization/68988
7025 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7026 avoidance test involving non_reload_pseudos. Move frequency test
7027 below the general fragmentation avoidance test.
7029 2017-09-19 Richard Biener <rguenther@suse.de>
7031 PR tree-optimization/69728
7032 * graphite-sese-to-poly.c (schedule_error): New global.
7033 (add_loop_schedule): Handle empty domain by failing the
7035 (build_original_schedule): Handle schedule_error.
7037 2017-09-19 Richard Biener <rguenther@suse.de>
7039 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7040 Do not iterate to sibling loops but only to siblings of inner
7043 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7046 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7047 that the registers are different.
7049 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7051 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7052 to processor_model and "amdfam17h" to arch_names_table.
7053 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7055 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7058 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7060 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7061 Alan Hayward <alan.hayward@arm.com>
7062 David Sherwood <david.sherwood@arm.com>
7064 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7066 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7068 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7069 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7070 vect_schedule_slp_instance.
7071 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7072 Update call to vect_slp_analyze_node_operations. Simplify return
7074 (vect_slp_analyze_bb_1): Update call accordingly.
7075 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7076 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7077 (vect_schedule_slp): Update call accordingly.
7079 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7080 Alan Hayward <alan.hayward@arm.com>
7081 David Sherwood <david.sherwood@arm.com>
7083 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7084 with types that aren't in fact scalar.
7086 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7088 * tree-vect-slp.c (vect_record_max_nunits): New function,
7090 (vect_build_slp_tree_1): ...here.
7091 (vect_build_slp_tree_2): Call it for phis too.
7093 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7095 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7096 to vect_get_vec_def_for_operand when getting the mask operand.
7098 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7099 Alan Hayward <alan.hayward@arm.com>
7100 David Sherwood <david.sherwood@arm.com>
7102 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7105 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7106 Alan Hayward <alan.hayward@arm.com>
7107 David Sherwood <david.sherwood@arm.com>
7109 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7110 calculation for vector booleans.
7112 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7113 Alan Hayward <alan.hayward@arm.com>
7114 David Sherwood <david.sherwood@arm.com>
7116 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7118 (vect_transform_stmt): ...here.
7119 (vect_analyze_stmt): Use it instead of calling
7120 vectorizable_live_operation directly.
7122 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7124 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7125 non-SIMT targets in acc vector loops.
7127 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7129 * configure.ac: Add arc and check if assembler supports gdwarf2.
7130 * configure: Regenerate.
7132 2017-09-18 Richard Biener <rguenther@suse.de>
7134 PR tree-optimization/82220
7135 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7136 epilogue niters from the min_profitable_iters compute.
7138 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7141 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7142 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7143 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7145 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7148 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7149 switching to a new text section.
7151 2017-09-18 Richard Biener <rguenther@suse.de>
7153 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7155 (build_alias_set): Reject aliases with no access function.
7157 2017-09-18 Richard Biener <rguenther@suse.de>
7159 PR tree-optimization/79622
7160 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7162 (build_cross_bb_scalars_use): Likewise.
7164 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7166 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7168 2017-09-18 Alan Modra <amodra@gmail.com>
7171 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7172 stack_pointer_rtx for count 0. Update comments. Break up
7173 large rtl expression.
7175 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7177 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7178 Increase to 20 bytes.
7179 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7180 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7181 or avx version of the stub.
7183 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7186 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7187 compute the minimum stack alignment. Also update preferred stack
7188 boundary for leaf functions.
7190 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7192 PR tree-optimization/82228
7193 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7196 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7198 * common/config/nds32/nds32-common.c
7199 (nds32_option_optimization_table): Refine formatting.
7200 (nds32_option_optimization_table): Use -fsched-pressure and
7201 -fomit-frame-pointer for specific optimization level.
7203 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7205 * config/nds32/nds32.c: Refine formatting and comments.
7206 * config/nds32/nds32.h: Likewise.
7207 * config/nds32/nds32.md: Likewise.
7208 * config/nds32/nds32-cost.c: Likewise.
7209 * config/nds32/nds32-isr.c: Likewise.
7210 * config/nds32/nds32-md-auxiliary.c: Likewise.
7211 * config/nds32/nds32-multiple.md: Likewise.
7212 * config/nds32/nds32-predicates.c: Likewise.
7214 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7215 Jakub Jelinek <jakub@redhat.com>
7217 Add support for -std=c++2a.
7218 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7220 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7222 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7225 * doc/extend.texi (Common Function Attributes): Add
7226 references to ARM, AArch64, and S/390 specific attributes.
7227 (Function Specific Option Pragmas): Add AArch64 and S/390
7228 to list of back ends that support the target pragma.
7230 2017-09-15 Nathan Sidwell <nathan@acm.org>
7232 * doc/standards.texi: Fix C++17 description. Update URLs for
7235 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7237 * common.opt (Wcast-align=strict): New warning option.
7238 * doc/invoke.texi: Document -Wcast-align=strict.
7240 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7242 * cgraph.h (cgraph_thunk_info): Add comments.
7243 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7244 assert for VIRTUAL_* arguments stricter.
7246 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7248 PR tree-optimization/71026
7249 * match.pd: Move RDIV patterns from fold-const.c
7250 * fold-const.c (distribute_real_division): Removed.
7251 (fold_binary_loc): Remove calls to distribute_real_divison.
7253 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7255 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7256 c++1z and gnu++1z as deprecated. Change other references to
7257 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7258 Change -Wc++1z-compat to -Wc++17-compat.
7259 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7260 * dwarf2out.c (highest_c_language): Handle C++17.
7261 (gen_compile_unit_die): Likewise.
7263 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7265 PR rtl-optimization/82192
7266 * combine.c (make_extraction): Don't look through non-paradoxical
7267 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7270 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7271 Alan Hayward <alan.hayward@arm.com>
7272 David Sherwood <david.sherwood@arm.com>
7274 * target.def (function_arg_offset): New hook.
7275 * targhooks.h (default_function_arg_offset): Declare.
7276 * targhooks.c (default_function_arg_offset): New function.
7277 * function.c (locate_and_pad_parm): Use
7278 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7279 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7280 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7281 * doc/tm.texi: Regenerate.
7282 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7283 * config/spu/spu.c (spu_function_arg_offset): New function.
7284 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7285 * system.h (FUNCTION_ARG_OFFSET): Poison.
7287 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7288 Alan Hayard <alan.hayward@arm.com>
7289 David Sherwood <david.sherwood@arm.com>
7291 * target.def (truly_noop_truncation): New hook.
7292 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7293 than TRULY_NOOP_TRUNCATION.
7294 * hooks.h (hook_bool_uint_uint_true): Declare.
7295 * hooks.c (hook_bool_uint_uint_true): New function.
7296 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7297 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7298 * doc/tm.texi: Regenerate.
7299 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7300 rather than TRULY_NOOP_TRUNCATION in comments.
7301 (simplify_comparison): Likewise.
7302 (record_truncated_value): Likewise.
7303 * expmed.c (extract_bit_field_1): Likewise.
7304 (extract_split_bit_field): Likewise.
7305 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7306 instead of TRULY_NOOP_TRUNCATION.
7307 * function.c (assign_parm_setup_block): Likewise.
7308 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7309 * rtlhooks.c: Include target.h.
7310 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7311 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7312 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7313 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7314 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7315 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7316 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7317 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7318 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7319 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7320 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7321 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7322 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7323 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7324 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7325 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7326 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7327 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7328 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7329 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7330 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7331 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7332 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7333 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7334 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7335 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7336 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7337 rather than TRULY_NOOP_TRUNCATION in comments.
7338 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7339 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7340 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7341 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7342 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7343 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7344 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7345 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7346 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7347 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7348 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7349 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7350 rather than TRULY_NOOP_TRUNCATION in comments.
7351 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7352 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7353 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7354 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7355 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7356 TRULY_NOOP_TRUNCATION condition.
7357 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7358 (TRULY_NOOP_TRUNCATION): Delete.
7359 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7360 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7361 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7362 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7363 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7364 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7365 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7366 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7367 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7368 rather than TRULY_NOOP_TRUNCATION in comments.
7369 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7370 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7371 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7372 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7373 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7374 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7376 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7379 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7380 (*cmp_ior): Likewise.
7381 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7382 (*ior_scc_scc_cmp): Likewise.
7383 (*and_scc_scc): Likewise.
7384 (*and_scc_scc_cmp): Likewise.
7386 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7387 Alan Hayard <alan.hayward@arm.com>
7388 David Sherwood <david.sherwood@arm.com>
7390 * target.def (can_change_mode_class): New hook.
7391 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7392 (hard_regno_nregs): Likewise.
7393 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7394 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7395 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7396 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7397 (LOAD_EXTEND_OP): Update accordingly.
7398 * doc/tm.texi: Regenerate.
7399 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7400 CANNOT_CHANGE_MODE_CLASS.
7401 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7402 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7403 * combine.c (simplify_set): Update accordingly.
7404 * emit-rtl.c (validate_subreg): Likewise.
7405 * recog.c (general_operand): Likewise.
7406 * regcprop.c (mode_change_ok): Likewise.
7407 * reload1.c (choose_reload_regs): Likewise.
7408 (inherit_piecemeal_p): Likewise.
7409 * rtlanal.c (simplify_subreg_regno): Likewise.
7410 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7411 instead of CANNOT_CHANGE_MODE_CLASS.
7412 (reload_cse_simplify_operands): Likewise.
7413 * reload.c (push_reload): Use targetm.can_change_mode_class
7414 instead of CANNOT_CHANGE_MODE_CLASS.
7415 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7416 REG_CANNOT_CHANGE_MODE_P.
7417 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7418 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7419 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7420 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7421 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7422 (arm_can_change_mode_class): New function.
7423 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7424 than CANNOT_CHANGE_MODE_CLASS in comments.
7425 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7426 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7427 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7428 (ix86_can_change_mode_class): ...this new function, inverting the
7429 sense of the return value.
7430 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7431 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7432 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7433 (ia64_can_change_mode_class): New function.
7434 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7435 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7436 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7437 (m32c_can_change_mode_class): ...this new function, inverting the
7438 sense of the return value.
7439 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7440 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7441 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7442 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7443 (mips_can_change_mode_class): ...this new function, inverting the
7444 sense of the return value.
7445 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7446 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7447 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7448 (msp430_can_change_mode_class): New function.
7449 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7450 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7451 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7452 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7453 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7454 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7455 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7456 (pa_cannot_change_mode_class): Replace with...
7457 (pa_can_change_mode_class): ...this new function, inverting the
7458 sense of the return value.
7459 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7460 than CANNOT_CHANGE_MODE_CLASS in comments.
7461 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7462 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7463 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7464 (pdp11_cannot_change_mode_class): Replace with...
7465 (pdp11_can_change_mode_class): ...this new function, inverting the
7466 sense of the return value.
7467 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7468 * config/powerpcspe/powerpcspe-protos.h
7469 (rs6000_cannot_change_mode_class_ptr): Delete.
7470 * config/powerpcspe/powerpcspe.c
7471 (rs6000_cannot_change_mode_class_ptr): Delete.
7472 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7473 (rs6000_option_override_internal): Assign to
7474 targetm.can_change_mode_class instead of
7475 rs6000_cannot_change_mode_class_ptr.
7476 (rs6000_cannot_change_mode_class): Replace with...
7477 (rs6000_can_change_mode_class): ...this new function, inverting the
7478 sense of the return value.
7479 (rs6000_debug_cannot_change_mode_class): Replace with...
7480 (rs6000_debug_can_change_mode_class): ...this new function.
7481 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7482 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7483 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7484 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7485 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7487 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7488 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7489 (rs6000_option_override_internal): Assign to
7490 targetm.can_change_mode_class instead of
7491 rs6000_cannot_change_mode_class_ptr.
7492 (rs6000_cannot_change_mode_class): Replace with...
7493 (rs6000_can_change_mode_class): ...this new function, inverting the
7494 sense of the return value.
7495 (rs6000_debug_cannot_change_mode_class): Replace with...
7496 (rs6000_debug_can_change_mode_class): ...this new function.
7497 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7498 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7499 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7500 (s390_can_change_mode_class): ...this new function, inverting the
7501 sense of the return value.
7502 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7503 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7504 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7505 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7506 (sh_cannot_change_mode_class): Replace with...
7507 (sh_can_change_mode_class): ...this new function, inverting the
7508 sense of the return value.
7509 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7510 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7511 (sparc_can_change_mode_class): New function.
7512 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7513 * config/spu/spu.c (spu_can_change_mode_class): New function.
7514 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7515 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7516 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7517 (visium_can_change_mode_class): New function.
7518 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7520 2017-09-15 Richard Biener <rguenther@suse.de>
7522 PR tree-optimization/82217
7523 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7524 but not undefined case.
7526 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7529 * postreload.c (reload_cse_simplify_operands): Skip
7530 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7532 2017-09-15 Richard Biener <rguenther@suse.de>
7534 PR tree-optimization/68823
7535 * graphite-scop-detection.c (build_alias_set): If we have a
7536 possible dependence check whether we can handle them by just
7537 looking at the DRs DR_ACCESS_FNs.
7538 (build_scops): If build_alias_set fails, fail the SCOP.
7540 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7542 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7543 to support float128 built-in functions that require the ISA 3.0
7545 (BU_FLOAT128_3_HW): Likewise.
7546 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7548 (FMAF128): Likewise.
7550 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7551 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7552 floating point instructions.
7553 (rs6000_invalid_builtin): Likewise.
7554 (rs6000_builtin_mask_names): Likewise.
7555 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7556 (RS6000_BTM_FLOAT128_HW): Likewise.
7557 (RS6000_BTM_COMMON): Likewise.
7558 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7560 * doc/extend.texi (RS/6000 built-in functions): Document the
7561 IEEE 128-bit floating point square root and fused multiply-add
7564 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7566 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7567 reg (r2) isn't in the set of registers defined in the prologue.
7569 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7570 Alan Hayward <alan.hayward@arm.com>
7571 David Sherwood <david.sherwood@arm.com>
7573 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7574 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7575 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7576 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7577 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7579 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7580 max_vectorization_factor.
7581 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7583 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7584 Alan Hayward <alan.hayward@arm.com>
7585 David Sherwood <david.sherwood@arm.com>
7587 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7588 (vect_worthwhile_without_simd_p): Declare.
7589 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7590 (vectorizable_reduction): Use it.
7591 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7592 (vectorizable_operation): Likewise.
7594 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7595 Alan Hayward <alan.hayward@arm.com>
7596 David Sherwood <david.sherwood@arm.com>
7598 * tree-vectorizer.h (vect_get_num_copies): New function.
7599 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7600 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7601 (vectorizable_induction): Likewise.
7602 (vectorizable_live_operation): Likewise.
7603 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7604 (vectorizable_bswap): Likewise.
7605 (vectorizable_call): Likewise.
7606 (vectorizable_conversion): Likewise.
7607 (vectorizable_assignment): Likewise.
7608 (vectorizable_shift): Likewise.
7609 (vectorizable_operation): Likewise.
7610 (vectorizable_store): Likewise.
7611 (vectorizable_load): Likewise.
7612 (vectorizable_condition): Likewise.
7613 (vectorizable_comparison): Likewise.
7614 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7616 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7617 Alan Hayward <alan.hayward@arm.com>
7618 David Sherwood <david.sherwood@arm.com>
7620 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7621 of vect_init_vector.
7623 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7624 Alan Hayward <alan.hayward@arm.com>
7625 David Sherwood <david.sherwood@arm.com>
7627 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7628 an inline wrapper that provides a location.
7629 (gimple_build_vector): Likewise.
7630 * gimple-fold.c (gimple_build_vector_from_val): New function.
7631 (gimple_build_vector): Likewise.
7632 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7633 functions to build the initial value. Always return a gimple value.
7634 (get_initial_defs_for_reduction): Likewise. Only compute
7636 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7637 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7638 (vectorizable_induction): Use gimple_build_vector rather than
7641 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7642 Alan Hayward <alan.hayward@arm.com>
7643 David Sherwood <david.sherwood@arm.com>
7645 * target.h (vec_perm_indices): New typedef.
7646 (auto_vec_perm_indices): Likewise.
7647 * optabs-query.h: Include target.h
7648 (can_vec_perm_p): Take a vec_perm_indices *.
7649 * optabs-query.c (can_vec_perm_p): Likewise.
7650 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7651 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7652 * tree-vect-generic.c (lower_vec_perm): Likewise.
7653 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7654 (vect_grouped_load_supported): Likewise.
7655 (vect_shift_permute_load_chain): Likewise.
7656 (vect_permute_store_chain): Use auto_vec_perm_indices.
7657 (vect_permute_load_chain): Likewise.
7658 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7659 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7660 Update uses of can_vec_perm_p.
7661 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7662 mode with a number of elements. Take a vec_perm_indices *.
7663 (vect_create_epilog_for_reduction): Update accordingly.
7664 Use auto_vec_perm_indices.
7665 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7666 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7667 (vect_transform_slp_perm_load): Likewise.
7668 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7669 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7670 (vect_gen_perm_mask_checked): Likewise.
7671 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7672 (vect_gen_perm_mask_checked): Likewise.
7673 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7674 (vectorizable_store): Likewise.
7675 (vectorizable_load): Likewise.
7676 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7677 (vectorizable_bswap): Likewise.
7679 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7680 Alan Hayward <alan.hayward@arm.com>
7681 David Sherwood <david.sherwood@arm.com>
7683 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7684 * tree.c (build_vector): Likewise.
7685 (build_vector_from_ctor): Update accordingly.
7686 (build_vector_from_val): Likewise.
7687 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7688 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7689 * tree-vect-generic.c (add_rshift): Likewise.
7690 (expand_vector_divmod): Likewise.
7691 (optimize_vector_constructor): Likewise.
7692 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7693 (vect_transform_slp_perm_load): Likewise.
7694 (vect_schedule_slp_instance): Likewise.
7695 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7696 (vectorizable_call): Likewise.
7697 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7698 * expmed.c (make_tree): Likewise.
7699 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7700 a vector passed to build_vector.
7701 (fold_convert_const): Likewise.
7702 (exact_inverse): Likewise.
7703 (fold_ternary_loc): Likewise.
7704 (fold_relational_const): Likewise.
7705 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7706 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7707 (const_unop): Likewise. Store the reduction accumulator in a
7708 variable rather than an array.
7709 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7710 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7711 the new vector, rather than constructing it after the input arrays.
7712 (native_interpret_vector): Use auto_vec<tree> when building
7713 a vector passed to build_vector. Add elements in order.
7714 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7715 auto_vec<tree> when building a vector passed to build_vector.
7716 (vect_create_epilog_for_reduction): Likewise.
7717 (vectorizable_induction): Likewise.
7718 (get_initial_def_for_reduction): Likewise. Fix indentation of
7720 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7722 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7723 passed to build_vector.
7725 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7726 Alan Hayward <alan.hayward@arm.com>
7727 David Sherwood <david.sherwood@arm.com>
7729 * tree-core.h (tree_base::u): Add an "nelts" field.
7730 (tree_vector): Use VECTOR_CST_NELTS as the length.
7731 * tree.c (tree_size): Likewise.
7732 (make_vector): Initialize VECTOR_CST_NELTS.
7733 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7734 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7735 TYPE_VECTOR_SUBPARTS.
7736 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7737 as the number of units, setting it from VECTOR_CST_NELTS.
7738 (const_vector_from_tree): Likewise.
7739 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7740 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7741 (fold_negate_expr_1): Likewise.
7742 (fold_convert_const): Likewise.
7743 (const_binop): Likewise. Differentiate the number of output and
7745 (const_unop): Likewise.
7746 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7747 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7748 in cases that did the opposite.
7750 2017-09-14 Richard Biener <rguenther@suse.de>
7752 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7755 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7757 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7759 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7762 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7763 if and where to split a bb, except for splitting before debug insn
7764 sequences followed by non-label real insn. Delete debug insns
7765 in between basic blocks.
7767 * combine.c (make_compound_operation_int): Formatting fixes.
7769 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7770 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7771 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7772 * config/sol2.h (LINK_EH_SPEC): Likewise.
7773 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7774 * config/s390/linux.h (LINK_SPEC): Likewise.
7775 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7776 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7777 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7778 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7779 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7780 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7781 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7782 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7784 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7786 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7788 (ENDFILE_LINUX_SPEC): Likewise.
7789 (LINK_EH_SPEC): Likewise.
7790 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7791 (LINK_OS_LINUX_SPEC32): Likewise.
7792 (LINK_OS_LINUX_SPEC64): Likewise.
7793 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7794 (LINK_OS_LINUX_SPEC): Likewise.
7796 2017-09-13 Martin Liska <mliska@suse.cz>
7799 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7800 CASE_HIGH is NULL_TREE.
7802 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7803 Alan Hayward <alan.hayward@arm.com>
7804 David Sherwood <david.sherwood@arm.com>
7806 * target.def (secondary_memory_needed): New hook.
7807 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7808 instead of SECONDARY_MEMORY_NEEDED.
7809 (secondary_memory_needed_mode): Likewise.
7810 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7811 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7812 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7813 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7814 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7815 * doc/tm.texi: Regenerate.
7816 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7817 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7818 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7819 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7820 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7821 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7822 mode argument first and change the reg_class arguments to reg_class_t.
7823 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7824 Make static. Update the call to inline_secondary_memory_needed.
7825 (ix86_register_move_cost): Update the call to
7826 inline_secondary_memory_needed.
7827 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7828 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7830 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7831 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7833 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7834 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7835 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7836 and match hook interface. Add comment from mips.h.
7837 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7838 * config/mmix/mmix.md (truncdfsf2): Refer to
7839 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7841 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7842 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7843 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7844 (pa_secondary_memory_needed): New function.
7845 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7846 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7847 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7848 (pdp11_secondary_memory_needed): Make static and match hook interface.
7849 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7850 * config/powerpcspe/powerpcspe-protos.h
7851 (rs6000_secondary_memory_needed_ptr): Delete.
7852 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7854 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7855 (rs6000_option_override_internal): Assign to
7856 targetm.secondary_memory_needed rather than
7857 rs6000_secondary_memory_needed_ptr.
7858 (rs6000_secondary_memory_needed): Match hook interface.
7859 (rs6000_debug_secondary_memory_needed): Likewise.
7860 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7861 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7862 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7863 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7864 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7865 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7867 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7868 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7869 (rs6000_option_override_internal): Assign to
7870 targetm.secondary_memory_needed rather than
7871 rs6000_secondary_memory_needed_ptr.
7872 (rs6000_secondary_memory_needed): Match hook interface.
7873 (rs6000_debug_secondary_memory_needed): Likewise.
7874 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7875 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7876 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7877 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7878 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7879 (sparc_secondary_memory_needed): New function.
7880 * lra-constraints.c (check_and_process_move): Refer to
7881 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7883 (curr_insn_transform): Likewise.
7884 (process_alt_operands): Use targetm.secondary_memory_needed
7885 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7886 (check_secondary_memory_needed_p): Likewise.
7887 (choose_split_class): Likewise.
7888 * reload.c: Unconditionally include code that was previously
7889 conditional on SECONDARY_MEMORY_NEEDED.
7890 (push_secondary_reload): Use targetm.secondary_memory_needed
7891 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7892 (push_reload): Likewise.
7893 * reload1.c: Unconditionally include code that was previously
7894 conditional on SECONDARY_MEMORY_NEEDED.
7895 (choose_reload_regs): Use targetm.secondary_memory_needed
7896 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7897 (gen_reload): Likewise.
7898 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7900 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7901 Alan Hayward <alan.hayward@arm.com>
7902 David Sherwood <david.sherwood@arm.com>
7904 * target.def (secondary_memory_needed_mode): New hook:
7905 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7906 * targhooks.h (default_secondary_memory_needed_mode): New function.
7907 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7908 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7909 * doc/tm.texi: Regenerate.
7910 * lra-constraints.c (check_and_process_move): Use
7911 targetm.secondary_memory_needed_mode instead of
7912 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7913 (curr_insn_transform): Likewise.
7914 * reload.c (get_secondary_mem): Likewise.
7915 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7916 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7918 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7919 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7920 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7921 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7922 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7924 * config/powerpcspe/powerpcspe-protos.h
7925 (rs6000_secondary_memory_needed_mode): Delete.
7926 * config/powerpcspe/powerpcspe.c
7927 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7928 (rs6000_secondary_memory_needed_mode): Make static.
7929 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7930 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7932 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7934 (rs6000_secondary_memory_needed_mode): Make static.
7935 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7936 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7937 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7938 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7939 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7941 (sparc_secondary_memory_needed_mode): New function.
7942 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7944 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7946 * config/aarch64/constraints.md (Umq): New constraint.
7947 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7949 (mov<mode>): Update condition.
7951 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7953 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7954 when bitposition is the same.
7956 2017-09-13 Richard Biener <rguenther@suse.de>
7958 * dwarf2out.c (output_die_symbol): Remove.
7959 (output_die): Do not output a DIEs symbol.
7961 2017-09-13 Richard Biener <rguenther@suse.de>
7964 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7965 default-def to avoid breaking iterator update with the weird
7966 interaction with cgraph_update_edges_for_call_stmt_node.
7968 2017-09-13 Richard Biener <rguenther@suse.de>
7970 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7971 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7972 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7973 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7975 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7977 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7978 Disable pc relative literal load irrespective of
7979 TARGET_FIX_ERR_A53_84341 for default.
7981 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7983 * config/sparc/sparc.c (output_return): Output the source location of
7984 the insn in the delay slot, if any.
7985 (output_sibcall): Likewise.
7987 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7990 * common.opt (-static-pie): New alias.
7991 (shared): Negate static-pie.
7992 (-no-pie): Update help text.
7994 (static-pie): New option.
7995 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7996 -static-pie support.
7997 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7998 (LINK_EH_SPEC): Likewise.
7999 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8000 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8001 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8002 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8003 (init_gcc_specs): Likewise.
8004 (init_spec): Likewise.
8005 (display_help): Update help message for -pie.
8006 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8009 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8011 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8012 (movdi_aarch64): Likewise.
8013 (movti_aarch64): Likewise.
8015 2017-09-12 Simon Wright <simon@pushface.org>
8018 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8019 calculation of the minor version, always output as 0.
8021 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8024 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8025 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8026 on it early, rather than manual conversion late. For
8027 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8028 instead of performing manual conversion.
8030 2017-09-12 Carl Love <cel@us.ibm.com>
8032 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8033 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8035 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8036 VMULOSW): Add definitions.
8037 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8038 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8039 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8040 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8041 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8043 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8045 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8047 (movti_aarch64): Likewise.
8048 (movdf_aarch64): Likewise.
8049 (movtf_aarch64): Likewise.
8050 (load_pairdi): Likewise.
8051 (store_pairdi): Likewise.
8052 (load_pairdf): Likewise.
8053 (store_pairdf): Likewise.
8054 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8055 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8056 (ldr_got_small_<mode>): Likewise.
8057 (ldr_got_small_28k_<mode>): Likewise.
8058 (ldr_got_tiny): Likewise.
8059 * config/aarch64/iterators.md (ldst_sz): New.
8060 (ldpstp_sz): Likewise.
8061 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8063 (thunderx_load): Split load_8 to load_16.
8064 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8066 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8067 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8068 (xgene1_store_pair): Split store_8 to store_16.
8069 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8070 (falkor_st_0_st_sd): Split store_8 to store_16.
8072 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8074 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8075 and store1/2/3/4 to store_4/8/12/16.
8076 * config/aarch64/aarch64.md: Update for rename.
8077 * config/arm/arm.md: Likewise.: Likewise.
8078 * config/arm/arm.c: Likewise.
8079 * config/arm/thumb1.md: Likewise.
8080 * config/arm/thumb2.md: Likewise.
8081 * config/arm/vfp.md: Likewise.
8082 * config/arm/arm-generic.md: Likewise.
8083 * config/arm/arm1020e.md: Likewise.
8084 * config/arm/arm1026ejs.md: Likewise.
8085 * config/arm/arm1136jfs.md: Likewise.
8086 * config/arm/arm926ejs.md: Likewise.
8087 * config/arm/cortex-a15.md: Likewise.
8088 * config/arm/cortex-a17.md: Likewise.
8089 * config/arm/cortex-a5.md: Likewise.
8090 * config/arm/cortex-a53.md: Likewise.
8091 * config/arm/cortex-a57.md: Likewise.
8092 * config/arm/cortex-a7.md: Likewise.
8093 * config/arm/cortex-a8.md: Likewise.
8094 * config/arm/cortex-a9.md: Likewise.
8095 * config/arm/cortex-m4.md: Likewise.
8096 * config/arm/cortex-m7.md: Likewise.
8097 * config/arm/cortex-r4.md: Likewise.
8098 * config/arm/exynos-m1.md: Likewise.
8099 * config/arm/fa526.md: Likewise.
8100 * config/arm/fa606te.md: Likewise.
8101 * config/arm/fa626te.md: Likewise.
8102 * config/arm/fa726te.md: Likewise.
8103 * config/arm/fmp626.md: Likewise.
8104 * config/arm/iwmmxt.md: Likewise.
8105 * config/arm/ldmstm.md: Likewise.
8106 * config/arm/marvell-pj4.md: Likewise.
8107 * config/arm/xgene1.md: Likewise.
8108 * config/aarch64/thunderx.md: Likewise.
8109 * config/aarch64/thunderx2t99.md: Likewise.
8110 * config/aarch64/falkor.md: Likewise.
8112 2017-09-12 Martin Liska <mliska@suse.cz>
8114 * attribs.c (private_lookup_attribute): New function.
8115 * attribs.h (private_lookup_attribute): Declared here.
8116 (lookup_attribute): Called from this place.
8118 2017-09-12 Richard Biener <rguenther@suse.de>
8120 PR tree-optimization/82157
8121 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8122 stmts with side-effects.
8124 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8125 Alan Hayward <alan.hayward@arm.com>
8126 David Sherwood <david.sherwood@arm.com>
8128 * target.def (hard_regno_nregs): New hook.
8129 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8130 * targhooks.h (default_hard_regno_nregs): Declare.
8131 * targhooks.c (default_hard_regno_nregs): New function.
8132 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8133 (TARGET_HARD_REGNO_NREGS): ...this hook.
8134 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8135 (CLASS_MAX_NREGS): Likewise.
8136 * doc/tm.texi: Regenerate.
8137 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8138 instead of HARD_REGNO_NREGS.
8139 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8140 HARD_REGNO_NREGS in the comment.
8141 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8142 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8143 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8144 Return an unsigned int.
8145 (TARGET_HARD_REGNO_NREGS): Redefine.
8146 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8147 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8148 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8149 (arc_hard_regno_nregs): New function.
8150 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8151 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8152 (arm_hard_regno_nregs): New function.
8153 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8154 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8155 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8156 (TARGET_HARD_REGNO_NREGS): Redefine.
8157 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8158 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8159 (HARD_REGNO_NREGS): Delete.
8160 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8161 (cr16_hard_regno_nregs): New function.
8162 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8163 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8164 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8165 (cris_hard_regno_nregs): New function.
8166 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8167 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8168 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8169 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8170 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8171 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8172 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8173 (frv_hard_regno_nregs): Make static. Take and return an
8175 (frv_class_max_nregs): Remove outdated copy of documentation.
8176 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8177 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8178 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8179 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8180 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8181 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8182 (TARGET_HARD_REGNO_NREGS): Redefine.
8183 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8184 (CLASS_MAX_NREGS): Update comment.
8185 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8186 (ia64_hard_regno_nregs): New function.
8187 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8188 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8189 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8190 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8191 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8193 (m32c_hard_regno_nregs): Likewise. Make static.
8194 (TARGET_HARD_REGNO_NREGS): Redefine.
8195 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8196 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8197 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8198 (m68k_hard_regno_nregs): New function.
8199 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8200 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8201 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8202 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8203 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8204 Take and return an unsigned int.
8205 (TARGET_HARD_REGNO_NREGS): Redefine.
8206 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8207 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8208 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8209 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8210 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8211 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8212 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8213 (msp430_hard_regno_nregs): Make static. Take and return an
8215 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8216 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8217 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8218 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8219 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8220 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8221 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8222 (TARGET_HARD_REGNO_NREGS): Redefine.
8223 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8224 (PA_HARD_REGNO_NREGS): ...this.
8225 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8226 (PA_HARD_REGNO_NREGS): ...this.
8227 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8228 (pa_hard_regno_nregs): New function.
8229 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8230 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8231 (pdp11_hard_regno_nregs): New function.
8232 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8233 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8234 (rs6000_hard_regno_nregs_hook): New function.
8235 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8236 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8237 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8238 Take and return an unsigned int. Move earlier in file.
8239 (TARGET_HARD_REGNO_NREGS): Redefine.
8240 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8241 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8242 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8243 (rl78_hard_regno_nregs): Make static. Take and return an
8245 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8246 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8247 (rs6000_hard_regno_nregs_hook): New function.
8248 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8249 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8250 (TARGET_HARD_REGNO_NREGS): Redefine.
8251 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8252 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8253 instead of HARD_REGNO_NREGS.
8254 (s390_hard_regno_nregs): New function.
8255 (s390_hard_regno_mode_ok): Add comment from s390.h.
8256 (TARGET_HARD_REGNO_NREGS): Redefine.
8257 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8258 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8259 (sh_hard_regno_nregs): New function.
8260 (sh_pass_in_reg_p): Use it.
8261 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8262 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8263 (sparc_hard_regno_nregs): New function.
8264 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8265 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8266 (spu_function_arg_advance): Use it, supplying a valid register number.
8267 (TARGET_HARD_REGNO_NREGS): Redefine.
8268 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8269 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8270 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8271 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8272 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8273 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8274 (CLASS_MAX_NREGS): Remove copy of old documentation.
8275 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8276 (visium_hard_regno_nregs): New function.
8277 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8278 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8279 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8280 xtensa_hard_regno_nregs): New function.
8281 * system.h (HARD_REGNO_NREGS): Poison.
8283 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8285 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8286 hard_regno_nregs instead of HARD_REGNO_NREGS.
8287 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8288 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8289 (c6x_expand_epilogue): Likewise.
8290 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8291 (frv_read_iacc_argument): Likewise.
8292 * config/sh/sh.c: Include regs.h.
8293 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8294 (regs_used): Likewise.
8295 (output_stack_adjust): Likewise.
8296 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8297 * expmed.c: Include regs.h.
8298 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8299 * ree.c: Include regs.h.
8300 (combine_reaching_defs): Use hard_regno_nregs instead of
8302 (add_removable_extension): Likewise.
8304 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8306 * regs.h (hard_regno_nregs): Turn into a function.
8307 (end_hard_regno): Update accordingly.
8308 * caller-save.c (setup_save_areas): Likewise.
8309 (save_call_clobbered_regs): Likewise.
8310 (replace_reg_with_saved_mem): Likewise.
8311 (insert_restore): Likewise.
8312 (insert_save): Likewise.
8313 * combine.c (can_change_dest_mode): Likewise.
8314 (move_deaths): Likewise.
8315 (distribute_notes): Likewise.
8316 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8317 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8318 (rs6000_split_multireg_move): Likewise.
8319 (rs6000_register_move_cost): Likewise.
8320 (rs6000_memory_move_cost): Likewise.
8321 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8322 (rs6000_split_multireg_move): Likewise.
8323 (rs6000_register_move_cost): Likewise.
8324 (rs6000_memory_move_cost): Likewise.
8325 * cselib.c (cselib_reset_table): Likewise.
8326 (cselib_lookup_1): Likewise.
8327 * emit-rtl.c (set_mode_and_regno): Likewise.
8328 * function.c (aggregate_value_p): Likewise.
8329 * ira-color.c (setup_profitable_hard_regs): Likewise.
8330 (check_hard_reg_p): Likewise.
8331 (calculate_saved_nregs): Likewise.
8332 (assign_hard_reg): Likewise.
8333 (improve_allocation): Likewise.
8334 (calculate_spill_cost): Likewise.
8335 * ira-emit.c (modify_move_list): Likewise.
8336 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8337 (ira_hard_reg_in_set_p): Likewise.
8338 * ira.c (setup_reg_mode_hard_regset): Likewise.
8339 (clarify_prohibited_class_mode_regs): Likewise.
8340 (check_allocation): Likewise.
8341 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8342 (lra_setup_reg_renumber): Likewise.
8343 (setup_try_hard_regno_pseudos): Likewise.
8344 (spill_for): Likewise.
8345 (assign_hard_regno): Likewise.
8346 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8347 * lra-constraints.c (in_class_p): Likewise.
8348 (lra_constraint_offset): Likewise.
8349 (simplify_operand_subreg): Likewise.
8350 (lra_constraints): Likewise.
8351 (split_reg): Likewise.
8352 (split_if_necessary): Likewise.
8353 (invariant_p): Likewise.
8354 (inherit_in_ebb): Likewise.
8355 * lra-lives.c (process_bb_lives): Likewise.
8356 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8357 (get_hard_regs): Likewise.
8358 (do_remat): Likewise.
8359 * lra-spills.c (assign_spill_hard_regs): Likewise.
8360 * mode-switching.c (create_pre_exit): Likewise.
8361 * postreload.c (reload_combine_recognize_pattern): Likewise.
8362 * recog.c (peep2_find_free_register): Likewise.
8363 * regcprop.c (kill_value_regno): Likewise.
8364 (set_value_regno): Likewise.
8365 (copy_value): Likewise.
8366 (maybe_mode_change): Likewise.
8367 (find_oldest_value_reg): Likewise.
8368 (copyprop_hardreg_forward_1): Likewise.
8369 * regrename.c (check_new_reg_p): Likewise.
8370 (regrename_do_replace): Likewise.
8371 * reload.c (push_reload): Likewise.
8372 (combine_reloads): Likewise.
8373 (find_dummy_reload): Likewise.
8374 (operands_match_p): Likewise.
8375 (find_reloads): Likewise.
8376 (find_equiv_reg): Likewise.
8377 (reload_adjust_reg_for_mode): Likewise.
8378 * reload1.c (count_pseudo): Likewise.
8379 (count_spilled_pseudo): Likewise.
8380 (find_reg): Likewise.
8381 (clear_reload_reg_in_use): Likewise.
8382 (free_for_value_p): Likewise.
8383 (allocate_reload_reg): Likewise.
8384 (choose_reload_regs): Likewise.
8385 (reload_adjust_reg_for_temp): Likewise.
8386 (emit_reload_insns): Likewise.
8387 (delete_output_reload): Likewise.
8388 * rtlanal.c (subreg_get_info): Likewise.
8389 * sched-deps.c (sched_analyze_reg): Likewise.
8390 * sel-sched.c (init_regs_for_mode): Likewise.
8391 (mark_unavailable_hard_regs): Likewise.
8392 (choose_best_reg_1): Likewise.
8393 (verify_target_availability): Likewise.
8394 * valtrack.c (dead_debug_insert_temp): Likewise.
8395 * var-tracking.c (track_loc_p): Likewise.
8396 (emit_note_insn_var_location): Likewise.
8397 * varasm.c (make_decl_rtl): Likewise.
8398 * reginfo.c (choose_hard_reg_mode): Likewise.
8399 (init_reg_modes_target): Refer directly to
8400 this_target_regs->x_hard_regno_nregs.
8402 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8404 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8405 instead of hard_regno_nregs.
8407 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8409 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8410 end_hard_regno instead of hard_regno_nregs.
8411 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8412 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8413 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8414 * ira-color.c (improve_allocation): Likewise.
8415 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8416 * lra-lives.c (mark_regno_live): Likewise.
8417 (mark_regno_dead): Likewise.
8418 * lra-remat.c (operand_to_remat): Likewise.
8419 * lra.c (collect_non_operand_hard_regs): Likewise.
8420 * postreload.c (reload_combine_note_store): Likewise.
8421 (move2add_valid_value_p): Likewise.
8422 * reload.c (regno_clobbered_p): Likewise.
8424 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8426 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8428 * config/v850/v850.c (v850_reorg): Likewise.
8429 * reload.c (refers_to_regno_for_reload_p): Likewise.
8430 (find_equiv_reg): Likewise.
8431 * reload1.c (reload_reg_reaches_end_p): Likewise.
8433 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8435 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8437 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8438 * config/arm/arm.c (output_move_neon): Likewise.
8439 (arm_attr_length_move_neon): Likewise.
8440 (neon_split_vcombine): Likewise.
8441 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8442 (c6x_mark_reg_written): Likewise.
8443 (c6x_dwarf_register_span): Likewise.
8444 * config/i386/i386.c (ix86_save_reg): Likewise.
8445 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8446 (rws_access_reg): Likewise.
8447 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8448 * mode-switching.c (create_pre_exit): Likewise.
8449 * ree.c (combine_reaching_defs): Likewise.
8450 (add_removable_extension): Likewise.
8451 * regcprop.c (find_oldest_value_reg): Likewise.
8452 (copyprop_hardreg_forward_1): Likewise.
8453 * reload.c (reload_inner_reg_of_subreg): Likewise.
8454 (push_reload): Likewise.
8455 (combine_reloads): Likewise.
8456 (find_dummy_reload): Likewise.
8457 (reload_adjust_reg_for_mode): Likewise.
8458 * reload1.c (find_reload_regs): Likewise.
8459 (forget_old_reloads_1): Likewise.
8460 (reload_reg_free_for_value_p): Likewise.
8461 (reload_adjust_reg_for_temp): Likewise.
8462 (emit_reload_insns): Likewise.
8463 (delete_output_reload): Likewise.
8464 * sel-sched.c (choose_best_reg_1): Likewise.
8465 (choose_best_pseudo_reg): Likewise.
8467 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8468 Alan Hayward <alan.hayward@arm.com>
8469 David Sherwood <david.sherwood@arm.com>
8471 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8472 * target.def (slow_unaligned_access): New hook.
8473 * targhooks.h (default_slow_unaligned_access): Declare.
8474 * targhooks.c (default_slow_unaligned_access): New function.
8475 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8476 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8477 * doc/tm.texi: Regenerate.
8478 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8479 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8480 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8482 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8483 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8485 (rs6000_slow_unaligned_access): New function.
8486 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8487 (expand_block_compare): Likewise.
8488 (expand_strn_compare): Likewise.
8489 (rs6000_rtx_costs): Likewise.
8490 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8491 (riscv_slow_unaligned_access): Likewise.
8492 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8493 (riscv_slow_unaligned_access_p): ...this and make static.
8494 (riscv_option_override): Update accordingly.
8495 (riscv_slow_unaligned_access): New function.
8496 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8497 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8498 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8499 (rs6000_slow_unaligned_access): New function.
8500 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8501 (rs6000_rtx_costs): Likewise.
8502 * config/rs6000/rs6000-string.c (expand_block_compare)
8503 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8504 of SLOW_UNALIGNED_ACCESS.
8505 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8506 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8507 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8508 of SLOW_UNALIGNED_ACCESS.
8509 * expmed.c (simple_mem_bitfield_p): Likewise.
8510 * expr.c (alignment_for_piecewise_move): Likewise.
8511 (emit_group_load_1): Likewise.
8512 (emit_group_store): Likewise.
8513 (copy_blkmode_from_reg): Likewise.
8514 (emit_push_insn): Likewise.
8515 (expand_assignment): Likewise.
8516 (store_field): Likewise.
8517 (expand_expr_real_1): Likewise.
8518 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8519 * lra-constraints.c (simplify_operand_subreg): Likewise.
8520 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8521 * gimple-ssa-store-merging.c: Likewise in block comment at start
8523 * tree-ssa-strlen.c: Include target.h.
8524 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8525 of SLOW_UNALIGNED_ACCESS.
8526 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8528 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8530 PR rtl-optimization/82185
8531 * expmed.c (emit_store_flag_int): Only test tem if it has been
8534 2017-09-12 Richard Biener <rguenther@suse.de>
8537 * match.pd ((FTYPE) N CMP CST): Fix typo.
8539 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8541 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8543 (mips_near_type_p): Add 'short_call' attribute as a synonym
8545 * doc/extend.texi (short_call): Document new function attribute.
8547 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8550 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8551 assertion check that in the condition.
8552 (get_atomic_generic_size): Likewise. Before testing if parameter
8553 has pointer type, if it has array type, call for C++
8554 default_conversion to perform array-to-pointer conversion.
8556 2017-09-12 Richard Biener <rguenther@suse.de>
8558 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8559 for operations we cannot scalarize.
8561 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8563 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8564 vectors heap vectors. Clean up comments.
8565 Make visited_bbs a reference.
8566 (profitable_jump_thread_path): Make GC
8567 vectors heap vectors. Clean up comments.
8569 (convert_and_register_jump_thread_path): Make GC vectors heap
8571 (check_subpath_and_update_thread_path): Same. Clean up comments.
8572 Make visited_bbs a reference.
8573 (handle_phi): Abstract common code to to
8574 register_jump_thread_path_if_profitable.
8575 Rename VAR_BB to DEF_BB.
8577 Make GC vectors heap vectors.
8578 Make visited_bbs a reference.
8579 (handle_assignment): Same.
8580 (register_jump_thread_path_if_profitable): New.
8581 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8583 Make GC vectors heap vectors. Clean up comments.
8584 Make visited_bbs a reference.
8585 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8586 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8589 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8592 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8593 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8595 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8597 Revert r251800 and r251799.
8599 2017-09-11 Martin Jambor <mjambor@suse.cz>
8602 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8603 arguments in advance.
8604 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8605 use it to find predecessor edges.
8606 (naive_outof_ssa): Collect vector of predecessors.
8608 2017-09-08 Jason Merrill <jason@redhat.com>
8610 PR c++/70029 - ICE with ref-qualifier and -flto
8611 * langhooks.h (struct lang_hooks_for_types): Add
8612 copy_lang_qualifiers.
8613 * attribs.c (build_type_attribute_qual_variant): Use it.
8614 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8616 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8617 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8619 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8622 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8623 (*mulsi3_sp64): New instruction.
8624 (mulsi3): New expander.
8626 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8628 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8630 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8632 * sancov.c: Include memmodel.h.
8634 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8637 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8640 2017-09-07 Carl Love <cel@us.ibm.com>
8642 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8643 the sldi instruction.
8645 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8647 * sancov.c: Include tm_p.h.
8649 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8652 * output.h (switch_to_other_text_partition): New declaration.
8653 * varasm.c (switch_to_other_text_partition): New function.
8654 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8655 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8656 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8657 to the other text partition before emitting LCL label and switch back
8658 after emitting the word after it.
8660 2017-09-07 Richard Biener <rguenther@suse.de>
8662 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8663 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8664 critical edges here, after loop init.
8665 (pass_data_pre): Remove PROP_no_crit_edges flags.
8666 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8667 for valueization of call args to avoid leaking VN_TOP.
8668 (visit_use): Assert we do not visit default defs.
8669 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8670 Use error_mark_node to more easily detect leaking VN_TOP.
8671 All default-defs are varying, not VN_TOP. Mark them visited.
8672 (run_scc_vn): Make code match comment.
8674 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8676 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8677 OPTION_MASK_FLOAT128_KEYWORD.
8678 (POWERPC_MASKS): Likewise.
8679 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8680 support for the -mfloat128-type option, and make -mfloat128
8681 default on PowerPC Linux systems. Define or undefine
8682 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8683 Define __float128 to be __ieee128 if IEEE 128-bit support is
8684 enabled, or undefine it.
8685 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8686 Delete defining __FLOAT128_TYPE__.
8687 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8688 -mfloat128-type option and make -mfloat128 default on PowerPC
8690 (TARGET_FLOAT128_TYPE): Likewise.
8691 (-mfloat128-type): Likewise.
8692 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8693 Delete the -mfloat128-type option and make -mfloat128 default on
8694 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8695 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8696 128-bit floating point is enabled. Change tests from using
8697 -mfloat128-type to -mfloat128.
8698 (rs6000_mangle_type): Use the correct mangling for the __float128
8699 type even if normal long double is restricted to 64-bits.
8700 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8702 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8703 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8704 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8706 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8707 documentation for -mfloat128.
8709 2017-09-06 Olivier Hainque <hainque@adacore.com>
8711 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8713 2017-09-06 Wish Wu <wishwu007@gmail.com>
8714 Jakub Jelinek <jakub@redhat.com>
8716 * asan.c (initialize_sanitizer_builtins): Add
8717 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8718 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8719 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8720 BT_FN_VOID_UINT64_PTR variables.
8721 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8722 (BT_FN_VOID_UINT16_UINT16): Likewise.
8723 (BT_FN_VOID_UINT32_UINT32): Likewise.
8724 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8725 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8726 (BT_FN_VOID_UINT64_PTR): Likewise.
8727 * common.opt (flag_sanitize_coverage): New variable.
8728 (fsanitize-coverage=trace-pc): Remove.
8729 (fsanitize-coverage=): Add.
8730 * flag-types.h (enum sanitize_coverage_code): New enum.
8731 * fold-const.c (fold_range_test): Disable non-short-circuit
8732 optimization if flag_sanitize_coverage.
8733 (fold_truth_andor): Likewise.
8734 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8735 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8736 (coverage_sanitizer_opts): New array.
8737 (get_closest_sanitizer_option): Add OPTS argument, handle also
8738 OPT_fsanitize_coverage_.
8739 (parse_sanitizer_options): Adjusted to also handle
8740 OPT_fsanitize_coverage_.
8741 (common_handle_option): Add OPT_fsanitize_coverage_.
8742 * sancov.c (instrument_comparison, instrument_switch): New function.
8743 (sancov_pass): Add trace-cmp support.
8744 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8745 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8746 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8747 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8748 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8749 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8750 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8751 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8752 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8753 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8755 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8757 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8758 error. Only quit immediately if parsing is complete.
8759 (BEGIN): Initialize fatal_err and parse_done.
8760 (begin fpu, end fpu): Check number of arguments.
8761 (begin arch, end arch): Likewise.
8762 (begin cpu, end cpu): Likewise.
8763 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8764 (optalias): Likewise.
8766 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8768 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8769 * config/arm/arm-isa.h: Delete. Move definitions to ...
8770 * arm-cpus.in: ... here. Use new feature and fgroup values.
8771 * config/arm/arm.c (arm_option_override): Use lower case for feature
8773 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8774 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8775 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8777 (print_isa_bits_for): New function.
8778 (gen_isa): New function.
8779 (gen_comm_data): Use print_isa_bits_for.
8780 (define feature): New keyword.
8781 (define fgroup): New keyword.
8782 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8783 (arm-isa.h): Add rule to generate file.
8784 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8785 case for feature bit names.
8787 2017-09-06 Richard Biener <rguenther@suse.de>
8789 * tree-ssa-pre.c (NECESSARY): Remove.
8790 (create_expression_by_pieces): Do not touch pass-local flags.
8791 (insert_into_preds_of_block): Likewise.
8792 (do_pre_regular_insertion): Likewise.
8793 (eliminate_insert): Likewise.
8794 (eliminate_dom_walker::before_dom_children): Likewise.
8795 (fini_eliminate): Do not look at inserted_exprs.
8796 (mark_operand_necessary): Remove.
8797 (remove_dead_inserted_code): Replace with simple work-list
8798 algorithm based on inserted_exprs and SSA uses.
8799 (pass_pre::execute): Re-order fini_eliminate and
8800 remove_dead_inserted_code.
8802 2017-09-06 Olivier Hainque <hainque@adacore.com>
8804 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8805 for VxWorks 7. Adjust surrounding comments.
8807 2017-09-06 Richard Biener <rguenther@suse.de>
8809 * gimple-ssa-strength-reduction.c
8810 (find_candidates_dom_walker::before_dom_children): Also allow
8813 2017-09-06 Richard Biener <rguenther@suse.de>
8815 PR tree-optimization/82108
8816 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8817 for gap in the non-permutation SLP case.
8819 2017-09-06 Martin Jambor <mjambor@suse.cz>
8821 PR tree-optimization/82078
8822 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8823 add_access_to_work_queue...
8824 (build_accesses_from_assign): ...here.
8825 (propagate_all_subaccesses): Make sure racc is the group
8826 representative, if there is one.
8828 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8831 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8834 2017-09-06 Richard Biener <rguenther@suse.de>
8836 * gimple-ssa-strength-reduction.c
8837 (find_candidates_dom_walker::before_doom_children): Use a
8838 type and not a mode check.
8840 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8843 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8845 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8847 2017-09-05 Jeff Law <law@redhat.com>
8849 PR tree-optimization/64910
8850 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8851 cases where we have 3 or more operands.
8853 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8856 * omp-low.c (lower_omp_for): Recompute tree invariant if
8857 gimple_omp_for_initial/final is ADDR_EXPR.
8860 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8861 into gimple val before gimplification fo the COND_EXPR.
8863 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8865 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8866 REGION_COPY argument.
8867 (thread_through_all_blocks): Remove unused argument to
8868 duplicate_thread_path.
8870 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8871 Alan Hayward <alan.hayward@arm.com>
8872 David Sherwood <david.sherwood@arm.com>
8874 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8875 Take a scalar_mode rather than a machine_mode.
8876 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8877 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8878 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8879 (aarch64_gen_adjusted_ldpstp): Likewise.
8880 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8882 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8883 Alan Hayward <alan.hayward@arm.com>
8884 David Sherwood <david.sherwood@arm.com>
8886 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8887 Take a scalar_int_mode instead of a machine_mode.
8888 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8889 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8890 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8891 (aarch64_simd_attr_length_rglist): Delete.
8892 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8893 a scalar_int_mode instead of a machine_mode.
8894 (aarch64_add_offset): Likewise.
8895 (aarch64_internal_mov_immediate): Likewise
8896 (aarch64_add_constant_internal): Likewise.
8897 (aarch64_add_constant): Likewise.
8898 (aarch64_movw_imm): Likewise.
8899 (aarch64_rtx_arith_op_extract_p): Likewise.
8900 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8901 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8902 Remove assert that the mode isn't a vector.
8903 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8904 (aarch64_expand_mov_immediate): Update calls after above changes.
8905 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8906 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8907 (aarch64_move_imm): Likewise.
8908 (aarch64_can_const_movi_rtx_p): Likewise.
8909 (aarch64_strip_extend): Likewise.
8910 (aarch64_extr_rtx_p): Likewise.
8911 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8912 a CONST_INT when the mode parameter is VOIDmode.
8913 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8915 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8917 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8918 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8919 (bitwise_type_for_mode): Update accordingly.
8921 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8923 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8924 * stor-layout.c (mode_for_size_tree): Likewise.
8925 (mode_for_array): Update accordingly.
8926 (layout_decl): Likewise.
8927 (compute_record_mode): Likewise. Only set the mode once.
8929 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8931 * target.def (get_mask_mode): Change return type to opt_mode.
8933 * doc/tm.texi: Regenerate.
8934 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8935 * targhooks.c (default_get_mask_mode): Likewise.
8936 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8937 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8938 targetm.get_mask_mode.
8939 * tree.c (build_truth_vector_type): Likewise.
8941 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8943 * machmode.h (mode_for_vector): Return an opt_mode.
8944 * stor-layout.c (mode_for_vector): Likewise.
8945 (mode_for_int_vector): Update accordingly.
8946 (layout_type): Likewise.
8947 * config/i386/i386.c (emit_memmov): Likewise.
8948 (ix86_expand_set_or_movmem): Likewise.
8949 (ix86_expand_vector_init): Likewise.
8950 (ix86_get_mask_mode): Likewise.
8951 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8953 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8954 * expmed.c (extract_bit_field_1): Likewise.
8955 * expr.c (expand_expr_real_2): Likewise.
8956 * optabs-query.c (can_vec_perm_p): Likewise.
8957 (can_vec_mask_load_store_p): Likewise.
8958 * optabs.c (expand_vec_perm): Likewise.
8959 * targhooks.c (default_get_mask_mode): Likewise.
8960 * tree-vect-stmts.c (vectorizable_store): Likewise.
8961 (vectorizable_load): Likewise.
8962 (get_vectype_for_scalar_type_and_size): Likewise.
8964 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8966 * machmode.h (mode_for_int_vector): New function.
8967 * stor-layout.c (mode_for_int_vector): Likewise.
8968 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8969 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8970 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8971 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8972 (s390_expand_vcond): Likewise.
8974 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8976 * machmode.h (opt_machine_mode): New type.
8977 (opt_mode<T>): Allow construction from anything that can be
8979 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8980 (mode_for_size): Return an opt_machine_mode.
8981 * stor-layout.c (mode_for_size): Likewise.
8982 (mode_for_size_tree): Update call accordingly.
8983 (bitwise_mode_for_mode): Likewise.
8984 (make_fract_type): Likewise.
8985 (make_accum_type): Likewise.
8986 * caller-save.c (replace_reg_with_saved_mem): Update call
8988 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8989 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8990 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8991 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8992 * expmed.c (extract_bit_field_1): Likewise.
8993 * reload.c (get_secondary_mem): Likewise.
8994 * varasm.c (assemble_integer): Likewise.
8995 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8998 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9000 * machmode.h (decimal_float_mode_for_size): New function.
9001 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9002 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9003 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9004 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9005 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9006 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9008 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9010 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9011 (get_builtin_sync_mode): Likewise.
9012 (expand_ifn_atomic_compare_exchange): Likewise.
9013 (expand_builtin_atomic_clear): Likewise.
9014 (expand_builtin_atomic_test_and_set): Likewise.
9015 (fold_builtin_atomic_always_lock_free): Likewise.
9016 * calls.c (compute_argument_addresses): Likewise.
9017 (emit_library_call_value_1): Likewise.
9018 (store_one_arg): Likewise.
9019 * combine.c (combine_instructions): Likewise.
9020 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9021 * config/arm/arm.c (arm_function_value): Likewise.
9022 (aapcs_allocate_return_reg): Likewise.
9023 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9024 * config/i386/i386.c (construct_container): Likewise.
9025 (ix86_gimplify_va_arg): Likewise.
9026 (ix86_expand_sse_cmp): Likewise.
9027 (emit_memmov): Likewise.
9028 (emit_memset): Likewise.
9029 (expand_small_movmem_or_setmem): Likewise.
9030 (ix86_expand_pextr): Likewise.
9031 (ix86_expand_pinsr): Likewise.
9032 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9033 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9035 * config/mips/mips.c (mips_function_value_1) Likewise.
9036 (mips_block_move_straight): Likewise.
9037 (mips_expand_ins_as_unaligned_store): Likewise.
9038 * config/powerpcspe/powerpcspe.c
9039 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9040 (rs6000_darwin64_record_arg_flush): Likewise.
9041 * config/rs6000/rs6000.c
9042 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9043 (rs6000_darwin64_record_arg_flush): Likewise.
9044 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9045 (sparc_function_value_1): Likewise.
9046 * config/spu/spu.c (adjust_operand): Likewise.
9047 (spu_emit_branch_or_set): Likewise.
9048 (arith_immediate_p): Likewise.
9049 * emit-rtl.c (gen_lowpart_common): Likewise.
9050 * expr.c (expand_expr_real_1): Likewise.
9051 * function.c (assign_parm_setup_block): Likewise.
9052 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9053 * reload1.c (alter_reg): Likewise.
9054 * stor-layout.c (mode_for_vector): Likewise.
9055 (layout_type): Likewise.
9057 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9059 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9060 (spu_convert_move): Likewise.
9061 * lower-subreg.c (resolve_simple_move): Likewise.
9063 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9066 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9067 define_insn to a define_expand.
9068 (altivec_vsum2sws_direct): New define_insn.
9069 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9071 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9073 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9076 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9081 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9082 realign stack if stack alignment needed is less than incoming
9085 2017-09-05 Marek Polacek <polacek@redhat.com>
9088 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9091 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9093 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9095 2017-09-05 Richard Biener <rguenther@suse.de>
9097 PR tree-optimization/82084
9098 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9100 2017-09-05 Richard Biener <rguenther@suse.de>
9102 PR tree-optimization/82102
9103 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9105 2017-09-05 Martin Liska <mliska@suse.cz>
9107 PR tree-optimization/82032
9108 * tree-cfg.c (generate_range_test): New function.
9109 * tree-cfg.h (generate_range_test): Declared here.
9110 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9111 (cleanup_control_expr_graph): Use it.
9112 * tree-switch-conversion.c (try_switch_expansion): Remove
9114 (emit_case_nodes): Use generate_range_test.
9116 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9119 * config/i386/i386.md (*<btsc><mode>_mask): Add
9120 TARGET_USE_BT to insn constraint.
9121 (*btr<mode>_mask): Ditto.
9123 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9125 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9126 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9128 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9131 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9132 TARGET_NEON and TARGET_IWMMXT.
9133 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9134 TARGET_NEON and TARGET_IWMMXT.
9135 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9137 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9139 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9140 (ix86_rewrite_tls_address): Ditto.
9141 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9142 (ix86_rewrite_tls_address_1): Ditto.
9143 (ix86_rewrite_tls_address): Ditto.
9144 * config/i386/predicates.md (tls_address_pattern): New predicate.
9145 * config/i386/i386.md (TLS address splitter): New splitter.
9147 2017-09-04 Richard Biener <rguenther@suse.de>
9149 PR tree-optimization/82084
9150 * fold-const.h (can_native_encode_string_p): Declare.
9151 * fold-const.c (can_native_encode_string_p): Factor out from ...
9152 (native_encode_string): ... here.
9153 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9154 vectorizing stores from constants we later cannot handle.
9156 2017-09-04 Marek Polacek <polacek@redhat.com>
9159 * doc/invoke.texi: Update -Wtautological-compare documentation.
9161 2017-09-04 Jeff Law <law@redhat.com>
9163 PR tree-optimization/64910
9164 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9165 swap the first and last operand if the last is a constant.
9167 2017-09-04 Marek Polacek <polacek@redhat.com>
9170 * convert.c (do_narrow): When sanitizing signed integer overflows,
9171 bail out for signed types.
9172 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9174 2017-09-04 Richard Biener <rguenther@suse.de>
9176 PR tree-optimization/82060
9177 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9178 Move devirtualization after stmt folding and before EH/AB/noreturn
9179 cleanup to get the stmt refs canonicalized. Use a bool instead
9180 of gimple_modified_p since that doesn't work for NOPs. Schedule
9181 NOPs generated by folding for removal.
9183 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9184 Alan Hayward <alan.hayward@arm.com>
9185 David Sherwood <david.sherwood@arm.com>
9187 * coretypes.h (pad_direction): New enum.
9188 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9189 (FUNCTION_ARG_PADDING): Likewise.
9190 * target.def (function_arg_padding): New hook.
9191 * targhooks.h (default_function_arg_padding): Declare.
9192 * targhooks.c (default_function_arg_padding): New function.
9193 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9194 (TARGET_FUNCTION_ARG_PADDING): ...this.
9195 * doc/tm.texi: Regenerate.
9196 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9197 instead of direction.
9198 (compute_argument_addresses): Likewise.
9199 (load_register_parameters): Likewise.
9200 (emit_library_call_value_1): Likewise.
9201 (store_one_arg): Use targetm.calls.function_arg_padding instead
9202 of FUNCTION_ARG_PADDING.
9203 (must_pass_in_stack_var_size_or_pad): Likewise.
9204 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9205 (emit_group_store): Likewise.
9206 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9207 instead of FUNCTION_ARG_PADDING.
9208 (emit_push_insn): Likewise, and propagate enum change throughout
9210 * function.h (direction): Delete.
9211 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9213 * function.c (assign_parm_find_stack_rtl): Likewise.
9214 (assign_parm_setup_block_p): Likewise.
9215 (assign_parm_setup_block): Likewise.
9216 (gimplify_parameters): Likewise.
9217 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9218 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9220 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9221 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9222 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9223 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9224 (aarch64_function_arg_padding): ...this new function.
9225 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9226 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9227 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9228 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9229 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9230 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9231 (arm_pad_arg_upward): Replace with...
9232 (arm_function_arg_padding): ...this new function.
9233 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9235 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9236 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9237 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9238 (ia64_hpux_function_arg_padding): Replace with...
9239 (ia64_function_arg_padding): ...this new function. Use pad_direction
9240 instead of direction. Check for TARGET_HPUX.
9241 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9242 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9243 (iq2000_function_arg_padding): New function.
9244 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9245 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9246 (mips_function_arg_padding): ...this new function.
9247 (mips_pad_reg_upward): Update accordingly.
9248 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9249 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9250 targetm.calls.function_arg_padding.
9251 (FUNCTION_ARG_PADDING): Delete.
9252 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9253 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9254 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9255 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9256 (nios2_block_reg_padding): Return pad_direction instead of direction.
9257 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9258 instead of direction.
9259 (nios2_function_arg_padding): Likewise. Make static.
9260 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9261 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9262 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9263 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9264 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9265 (pa_function_arg_padding): Make static. Return pad_direction instead
9267 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9268 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9269 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9270 instead of direction. Use targetm.calls.function_arg_padding.
9271 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9272 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9273 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9274 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9275 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9277 (function_arg_padding): Rename to...
9278 (rs6000_function_arg_padding): ...this. Make static. Return
9279 pad_direction instead of direction.
9280 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9281 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9282 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9283 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9284 instead of direction. Use targetm.calls.function_arg_padding.
9285 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9286 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9287 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9288 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9289 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9290 (function_arg_padding): Rename to...
9291 (rs6000_function_arg_padding): ...this. Make static. Return
9292 pad_direction instead of direction.
9293 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9294 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9295 * config/s390/s390.c (s390_function_arg_padding): New function.
9296 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9297 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9298 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9299 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9300 (function_arg_padding): Rename to...
9301 (sparc_function_arg_padding): ...this. Make static. Return
9302 pad_direction instead of direction.
9303 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9304 * config/spu/spu.c (spu_function_arg_padding): New function.
9305 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9306 * system.h (FUNCTION_ARG_PADDING): Poison.
9308 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9309 Alan Hayward <alan.hayward@arm.com>
9310 David Sherwood <david.sherwood@arm.com>
9312 * target.def (modes_tieable_p): New hook.
9313 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9314 (TARGET_MODES_TIEABLE_P): ...this.
9315 * doc/tm.texi.in: Regenerate.
9316 * hooks.h (hook_bool_mode_mode_true): Declare.
9317 * hooks.c (hook_bool_mode_mode_true): New function.
9318 * combine.c (subst): Use targetm.modes_tieable_p instead of
9320 * dse.c (find_shift_sequence): Likewise.
9321 * expmed.c (extract_low_bits): Likewise.
9322 * lower-subreg.c: Include target.h.
9323 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9325 * rtlanal.c (rtx_cost): Likewise.
9326 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9327 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9328 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9329 (TARGET_MODES_TIEABLE_P): Redefine.
9330 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9331 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9332 (TARGET_MODES_TIEABLE_P): Redefine.
9333 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9334 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9335 (arc_modes_tieable_p): New function.
9336 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9337 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9338 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9339 (arm_modes_tieable_p): Make static.
9340 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9341 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9342 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9343 (TARGET_MODES_TIEABLE_P): Redefine.
9344 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9345 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9346 (TARGET_MODES_TIEABLE_P): Redefine.
9347 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9348 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9349 (cr16_modes_tieable_p): New function.
9350 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9351 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9352 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9353 (TRULY_NOOP_TRUNCATION): Update comment.
9354 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9355 (TRULY_NOOP_TRUNCATION): Update comment.
9356 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9357 (frv_modes_tieable_p): New function.
9358 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9359 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9360 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9361 (TARGET_MODES_TIEABLE_P): Redefine.
9362 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9363 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9364 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9365 (TARGET_MODES_TIEABLE_P): Redefine.
9366 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9367 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9368 (ia64_modes_tieable_p): New function.
9369 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9370 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9371 (iq2000_modes_tieable_p): New function.
9372 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9373 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9374 (lm32_modes_tieable_p): New function.
9375 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9376 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9377 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9378 (TARGET_MODES_TIEABLE_P): Redefine.
9379 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9380 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9381 (m32r_modes_tieable_p): New function.
9382 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9383 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9384 (m68k_modes_tieable_p): New function.
9385 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9386 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9387 (mcore_modes_tieable_p): New function.
9388 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9389 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9391 (TARGET_MODES_TIEABLE_P): Redefine.
9392 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9393 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9394 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9395 (TARGET_MODES_TIEABLE_P): Redefine.
9396 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9397 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9398 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9399 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9400 (mn10300_modes_tieable_p): ...this and make static.
9401 (TARGET_MODES_TIEABLE_P): Redefine.
9402 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9403 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9404 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9405 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9406 (msp430_modes_tieable_p): Make static.
9407 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9408 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9409 (TARGET_MODES_TIEABLE_P): Redefine.
9410 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9411 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9412 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9413 (TARGET_MODES_TIEABLE_P): Redefine.
9414 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9415 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9416 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9417 (TARGET_MODES_TIEABLE_P): Redefine.
9418 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9419 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9420 (pdp11_modes_tieable_p): New function.
9421 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9422 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9423 (rs6000_modes_tieable_p): New function.
9424 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9425 * config/powerpcspe/powerpcspe.md: Update comment.
9426 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9427 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9428 (TARGET_MODES_TIEABLE_P): Redefine.
9429 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9430 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9431 (rl78_modes_tieable_p): New function.
9432 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9433 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9434 (rs6000_modes_tieable_p): New function.
9435 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9436 * config/rs6000/rs6000.md: Update comment.
9437 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9438 * config/rx/rx.c (rx_modes_tieable_p): New function.
9439 (TARGET_MODES_TIEABLE_P): Redefine.
9440 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9441 * config/s390/s390.c (s390_modes_tieable_p): New function.
9442 (TARGET_MODES_TIEABLE_P): Redefine.
9443 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9444 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9445 (sh_modes_tieable_p): New function.
9446 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9447 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9448 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9449 (sparc_modes_tieable_p): Make static.
9450 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9451 * config/spu/spu.c (spu_modes_tieable_p): New function.
9452 (TARGET_MODES_TIEABLE_P): Redefine.
9453 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9454 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9455 (TARGET_MODES_TIEABLE_P): Redefine.
9456 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9457 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9458 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9459 * config/v850/v850.c (v850_modes_tieable_p): New function.
9460 (TARGET_MODES_TIEABLE_P): Redefine.
9461 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9462 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9463 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9464 (visium_modes_tieable_p): New function.
9465 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9466 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9467 (xtensa_modes_tieable_p): New function.
9468 * system.h (MODES_TIEABLE_P): Poison.
9470 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9471 Alan Hayward <alan.hayward@arm.com>
9472 David Sherwood <david.sherwood@arm.com>
9474 * target.def (hard_regno_mode_ok): New hook.
9475 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9476 (TARGET_HARD_REGNO_MODE_OK): ...this.
9477 * doc/tm.texi.in: Regenerate.
9478 * hooks.h (hook_bool_uint_mode_true): Declare.
9479 * hooks.c (hook_bool_uint_mode_true): New function.
9480 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9482 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9483 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9484 instead of HARD_REGNO_MODE_OK.
9485 * caller-save.c: Include target.h.
9486 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9488 * combine.c (can_combine_p): Likewise.
9489 (combinable_i3pat): Likewise.
9490 (can_change_dest_mode): Likewise.
9491 * expr.c (init_expr_target): Likewise.
9492 (convert_move): Likewise.
9493 (convert_modes): Likewise.
9494 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9495 (setup_prohibited_mode_move_regs): Likewise.
9496 * ira.h (target_ira): Likewise.
9497 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9498 * lra-constraints.c (process_alt_operands): Likewise.
9499 (split_reg): Likewise.
9500 * recog.c (peep2_find_free_register): Likewise.
9501 * ree.c (combine_reaching_defs): Likewise.
9502 * regcprop.c (maybe_mode_change): Likewise.
9503 * reginfo.c (init_reg_sets_1): Likewise.
9504 (choose_hard_reg_mode): Likewise.
9505 (simplifiable_subregs): Likewise.
9506 * regrename.c (check_new_reg_p): Likewise.
9507 * reload.c (find_valid_class): Likewise.
9508 (find_valid_class_1): Likewise.
9509 (reload_inner_reg_of_subreg): Likewise.
9510 (push_reload): Likewise.
9511 (combine_reloads): Likewise.
9512 (find_dummy_reload): Likewise.
9513 (find_reloads): Likewise.
9514 * reload1.c (find_reg): Likewise.
9515 (set_reload_reg): Likewise.
9516 (allocate_reload_reg): Likewise.
9517 (choose_reload_regs): Likewise.
9518 (reload_adjust_reg_for_temp): Likewise.
9519 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9520 (simplify_subreg_regno): Likewise.
9521 * sel-sched.c (init_regs_for_mode): Likewise.
9522 * varasm.c (make_decl_rtl): Likewise.
9523 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9524 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9526 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9527 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9528 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9529 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9530 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9531 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9532 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9533 (arc_mode_class): Delete.
9534 (HARD_REGNO_MODE_OK): Delete.
9535 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9536 (arc_hard_regno_mode_ok): Rename old array to...
9537 (arc_hard_regno_mode_ok_modes): ...this.
9538 (arc_conditional_register_usage): Update accordingly.
9539 (arc_mode_class): Make static.
9540 (arc_hard_regno_mode_ok): New function.
9541 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9542 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9543 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9544 (arm_hard_regno_mode_ok): Make static.
9545 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9547 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9548 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9549 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9551 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9552 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9553 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9554 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9555 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9556 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9557 * config/bfin/predicates.md (valid_reg_operand): Use
9558 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9559 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9560 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9561 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9562 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9563 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9564 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9565 (cr16_hard_regno_mode_ok): Make static and return a bool.
9566 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9567 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9568 (cris_hard_regno_mode_ok): New function.
9569 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9570 (epiphany_mode_class): Delete.
9571 (HARD_REGNO_MODE_OK): Delete.
9572 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9573 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9574 (hard_regno_mode_ok): Rename to...
9575 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9576 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9577 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9579 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9580 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9581 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9582 (frv_hard_regno_mode_ok): Make static and return a bool.
9583 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9585 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9586 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9587 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9588 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9590 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9591 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9592 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9593 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9595 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9596 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9597 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9598 (ia64_hard_regno_mode_ok): New function.
9599 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9600 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9601 (iq2000_hard_regno_mode_ok): New function.
9602 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9603 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9604 (lm32_hard_regno_mode_ok): New function.
9605 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9606 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9607 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9608 instead of HARD_REGNO_MODE_OK.
9609 (m32c_hard_regno_ok): Rename to...
9610 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9611 (m32c_cannot_change_mode_class): Update accordingly.
9612 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9613 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9614 (m32r_mode_class): Delete.
9615 (HARD_REGNO_MODE_OK): Delete.
9616 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9617 (m32r_hard_regno_mode_ok): Rename to...
9618 (m32r_hard_regno_modes): ...this.
9619 (m32r_mode_class): Make static.
9620 (m32r_hard_regno_mode_ok): New function.
9621 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9622 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9623 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9624 (m68k_hard_regno_mode_ok): Make static.
9625 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9626 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9627 (mcore_hard_regno_mode_ok): New function.
9628 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9629 (HARD_REGNO_MODE_OK): Delete.
9630 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9632 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9633 (microblaze_hard_regno_mode_ok): New function.
9634 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9635 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9636 (mips_hard_regno_mode_ok): Delete.
9637 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9638 (mips_hard_regno_mode_ok_p): ...this and make static.
9639 (mips_hard_regno_mode_ok_p): Rename to...
9640 (mips_hard_regno_mode_ok_uncached): ...this.
9641 (mips_hard_regno_mode_ok): New function.
9642 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9643 of HARD_REGNO_MODE_OK.
9644 (mips_option_override): Update after above name changes.
9645 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9646 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9647 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9648 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9649 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9650 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9651 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9652 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9653 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9654 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9655 (msp430_hard_regno_mode_ok): Make static and return a bool.
9656 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9657 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9658 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9660 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9661 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9662 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9663 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9664 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9665 (PA_HARD_REGNO_MODE_OK): ...this
9666 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9667 (PA_HARD_REGNO_MODE_OK): ...this.
9668 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9669 (pa_hard_regno_mode_ok): New function.
9670 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9671 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9672 (pdp11_hard_regno_mode_ok): New function.
9673 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9674 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9676 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9678 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9679 (rs6000_hard_regno_mode_ok): Rename to...
9680 (rs6000_hard_regno_mode_ok_uncached): ...this.
9681 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9682 (rs6000_hard_regno_mode_ok): New function.
9683 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9684 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9685 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9686 (riscv_hard_regno_mode_ok): ...this and make static.
9687 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9688 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9689 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9690 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9691 (rl78_hard_regno_mode_ok): Make static and return bool.
9692 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9693 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9695 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9696 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9697 (rs6000_hard_regno_mode_ok): Rename to...
9698 (rs6000_hard_regno_mode_ok_uncached): ...this.
9699 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9700 (rs6000_hard_regno_mode_ok): New function.
9701 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9702 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9703 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9704 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9705 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9706 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9707 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9708 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9709 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9710 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9711 (sh_hard_regno_mode_ok): Make static.
9712 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9713 instead of HARD_REGNO_MODE_OK.
9714 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9715 (sparc_mode_class): Delete.
9716 (HARD_REGNO_MODE_OK): Delete.
9717 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9718 (hard_regno_mode_classes): Make static.
9719 (sparc_mode_class): Likewise.
9720 (sparc_hard_regno_mode_ok): New function.
9721 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9722 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9723 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9725 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9726 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9727 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9728 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9729 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9730 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9731 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9732 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9733 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9734 (visium_hard_regno_mode_ok): New function.
9735 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9736 instead of HARD_REGNO_MODE_OK.
9737 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9738 (HARD_REGNO_MODE_OK): Delete.
9739 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9740 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9741 (xtensa_option_override): Update accordingly.
9742 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9743 (xtensa_hard_regno_mode_ok): New function.
9744 * system.h (HARD_REGNO_MODE_OK): Poison.
9746 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9747 Alan Hayward <alan.hayward@arm.com>
9748 David Sherwood <david.sherwood@arm.com>
9750 * target.def (hard_regno_call_part_clobbered): New hook.
9751 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9752 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9753 * doc/tm.texi: Regenerate.
9754 * hooks.h (hook_bool_uint_mode_false): Declare.
9755 * hooks.c (hook_bool_uint_mode_false): New function.
9756 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9757 * cselib.c (cselib_process_insn): Use
9758 targetm.hard_regno_call_part_clobbered instead of
9759 HARD_REGNO_CALL_PART_CLOBBERED.
9760 * ira-conflicts.c (ira_build_conflicts): Likewise.
9761 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9762 * lra-constraints.c (need_for_call_save_p): Likewise.
9763 * lra-lives.c: Include target.h.
9764 (check_pseudos_live_through_calls): Use
9765 targetm.hard_regno_call_part_clobbered instead of
9766 HARD_REGNO_CALL_PART_CLOBBERED.
9767 * regcprop.c: Include target.h.
9768 (copyprop_hardreg_forward_1): Use
9769 targetm.hard_regno_call_part_clobbered instead of
9770 HARD_REGNO_CALL_PART_CLOBBERED.
9771 * reginfo.c (choose_hard_reg_mode): Likewise.
9772 * regrename.c (check_new_reg_p): Likewise.
9773 * reload.c (find_equiv_reg): Likewise.
9774 * reload1.c (emit_reload_insns): Likewise.
9775 * sched-deps.c (deps_analyze_insn): Likewise.
9776 * sel-sched.c (init_regs_for_mode): Likewise.
9777 (mark_unavailable_hard_regs): Likewise.
9778 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9779 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9780 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9782 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9783 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9784 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9786 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9788 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9789 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9790 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9792 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9793 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9794 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9796 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9797 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9799 * config/powerpcspe/powerpcspe.c
9800 (rs6000_hard_regno_call_part_clobbered): New function.
9801 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9802 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9803 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9805 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9806 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9807 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9809 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9810 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9811 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9813 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9814 Alan Hayward <alan.hayward@arm.com>
9815 David Sherwood <david.sherwood@arm.com>
9817 * rtl.h (subreg_memory_offset): Declare.
9818 * emit-rtl.c (subreg_memory_offset): New function.
9819 * expmed.c (store_bit_field_1): Use it.
9820 * expr.c (undefined_operand_subword_p): Likewise.
9821 * simplify-rtx.c (simplify_subreg): Likewise.
9823 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9825 PR rtl-optimization/57448
9828 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9829 using atomic_load pattern.
9830 (expand_atomic_store): Likewise.
9832 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9835 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9836 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9837 handling. Use replace_call_with_value with NULL instead of
9838 gsi_replace, unlink_stmt_vdef and release_defs.
9840 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9843 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9845 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9848 * rtl.h (emit_library_call_value_1): Declare.
9849 (emit_library_call): Replace declaration with a series of overloads.
9850 Remove the parameter count argument.
9851 (emit_library_call_value): Likewise.
9852 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9853 with an "rtx_mode_t *".
9854 (emit_library_call_value): Delete.
9855 (emit_library_call): Likewise.
9856 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9857 (asan_emit_allocas_unpoison): Likewise.
9858 * builtins.c (expand_builtin_powi): Likewise.
9859 (expand_asan_emit_allocas_unpoison): Likewise.
9860 * cfgexpand.c (expand_main_function): Likewise.
9861 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9862 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9863 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9864 * config/arm/arm.c (arm_trampoline_init): Likewise.
9865 (arm_call_tls_get_addr): Likewise.
9866 (arm_expand_divmod_libfunc): Likewise.
9867 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9868 (smulsi3_highpart): Likewise.
9869 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9870 (c6x_expand_compare): Likewise.
9871 (c6x_expand_movmem): Likewise.
9872 * config/frv/frv.c (frv_trampoline_init): Likewise.
9873 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9874 (ix86_expand_divmod_libfunc): Likewise.
9875 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9876 (ia64_expand_compare): Likewise.
9877 (ia64_profile_hook): Likewise.
9878 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9879 (nonlocal_goto): Likewise.
9880 (restore_stack_nonlocal): Likewise.
9881 * config/m32r/m32r.c (block_move_call): Likewise.
9882 (m32r_trampoline_init): Likewise.
9883 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9884 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9885 (m68k_call_m68k_read_tp): Likewise.
9886 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9887 (microblaze_expand_divide): Likewise.
9888 * config/mips/mips.h (mips_args): Likewise.
9889 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9890 (MIPS_ICACHE_SYNC): Likewise.
9891 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9892 (nios2_trampoline_init): Likewise.
9893 * config/pa/pa.c (hppa_tls_call): Likewise.
9894 (pa_trampoline_init): Likewise.
9895 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9896 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9897 (expand_strn_compare): Likewise.
9898 (rs6000_generate_compare): Likewise.
9899 (rs6000_expand_float128_convert): Likewise.
9900 (output_profile_hook): Likewise.
9901 (rs6000_trampoline_init): Likewise.
9902 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9903 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9904 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9905 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9906 (rs6000_generate_compare): Likewise.
9907 (rs6000_expand_float128_convert): Likewise.
9908 (output_profile_hook): Likewise.
9909 (rs6000_trampoline_init): Likewise.
9910 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9911 * config/sh/sh.c (sh_trampoline_init): Likewise.
9912 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9913 (sparc_emit_float_lib_cmp): Likewise.
9914 (sparc32_initialize_trampoline): Likewise.
9915 (sparc64_initialize_trampoline): Likewise.
9916 (sparc_profile_hook): Likewise.
9917 * config/spu/spu.c (ea_load_store): Likewise.
9918 * config/spu/spu.md (floatunssidf2): Likewise.
9919 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9920 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9921 * config/visium/visium.c (expand_block_move_4): Likewise.
9922 (expand_block_move_2): Likewise.
9923 (expand_block_move_1): Likewise.
9924 (expand_block_set_4): Likewise.
9925 (expand_block_set_2): Likewise.
9926 (expand_block_set_1): Likewise.
9927 (visium_trampoline_init): Likewise.
9928 (visium_profile_hook): Likewise.
9929 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9930 (xtensa_setup_frame_addresses): Likewise.
9931 (xtensa_trampoline_init): Likewise.
9932 * except.c (sjlj_emit_function_enter): Likewise.
9933 (sjlj_emit_function_exit): Likewise.
9934 * explow.c (allocate_dynamic_stack_space): Likewise.
9935 (probe_stack_range): Likewise.
9936 * expr.c (convert_mode_scalar): Likewise.
9937 * optabs.c (expand_binop): Likewise.
9938 (expand_twoval_binop_libfunc): Likewise.
9939 (expand_unop): Likewise.
9940 (prepare_cmp_insn): Likewise.
9941 (prepare_float_lib_cmp): Likewise.
9942 (expand_float): Likewise.
9943 (expand_fix): Likewise.
9944 (expand_fixed_convert): Likewise.
9945 (maybe_emit_sync_lock_test_and_set): Likewise.
9946 (expand_atomic_compare_and_swap): Likewise.
9947 (expand_mem_thread_fence): Likewise.
9948 (expand_atomic_fetch_op): Likewise.
9950 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9952 * doc/generic.texi (OpenACC): Adjust URL.
9953 * doc/invoke.texi (C Dialect Options): Ditto.
9955 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9957 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9958 predicate for operand 1. Add (m,<S>) constraint.
9959 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9960 Prevent memory operand 1 with register operand 2.
9962 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9964 PR rtl-optimization/82024
9965 * combine.c (try_combine): If the combination result is a PARALLEL,
9966 and we only need to retain the SET in there that would be placed
9967 at I2, check that we can place that at I3 instead, before doing so.
9969 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9972 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9974 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9975 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9978 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9979 Jeff Law <law@redhat.com>
9981 * varasm.c (bss_initializer_p): Do not put constants into .bss
9982 (categorize_decl_for_section): Handle bss_initializer_p returning
9983 false when DECL_INITIAL is NULL.
9985 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9988 * config/s390/s390.c (s390_can_inline_p): New function.
9990 2017-09-01 Jeff Law <law@redhat.com>
9992 PR tree-optimization/82052
9993 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9994 Always initialize the returned slot after a hash table miss
9995 when INSERT is true.
9997 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9999 * config/s390/s390.md (mem_signal_fence): Remove.
10000 * doc/md.texi (mem_signal_fence): Remove.
10001 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10003 * target-insns.def (mem_signal_fence): Remove.
10005 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10008 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10011 * asan.c (create_odr_indicator): Strip name encoding from assembler
10012 name before appending it after __odr_asan_.
10014 2017-09-01 Martin Liska <mliska@suse.cz>
10016 PR tree-optimization/82059
10017 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10018 frequency only when an edge is redirected.
10020 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10022 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10023 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10024 (arc_conditional_register_usage): Remove ARC600 lp_count
10026 (arc_file_start): Emit Tag_ARC_CPU_variation.
10027 (arc_can_use_doloop_p): New conditions to use ZOLs.
10028 (hwloop_fail): New function.
10029 (hwloop_optimize): Likewise.
10030 (hwloop_pattern_reg): Likewise.
10031 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10032 (arc_reorg_loops): New function, calls reorg_loops.
10033 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10034 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10036 (arc_loop_hazard): Remove function, functionality moved into
10038 (arc_hazard): Remove arc_loop_hazard call.
10039 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10040 into hwloop_optimize.
10041 (arc_label_align): Remove ZOL handling.
10042 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10043 * config/arc/arc.md (doloop_begin): Remove pattern.
10044 (doloop_begin_i): Likewise.
10045 (doloop_end_i): Likewise.
10046 (doloop_fallback): Likewise.
10047 (doloop_fallback_m): Likewise.
10048 (doloop_end): Reimplement expand.
10049 (arc_lp): New pattern for LP instruction.
10050 (loop_end): New pattern.
10051 (loop_fail): Likewise.
10052 (decrement_and_branch_until_zero): Likewise.
10053 * config/arc/arc.opt (mlpc-width): New option.
10054 * doc/invoke.texi (mlpc-width): Document option.
10056 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10058 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10059 (arc_ccfsm_advance): Fix checking for delay slots.
10060 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10062 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10064 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10066 (movhi_insn): Update store instruction constraint which are saving
10067 6-bit short immediates.
10068 (movsi_insn): Consider also short scaled load operations.
10069 (zero_extendhisi2_i): Use Usd constraint instead of T.
10070 (extendhisi2_i): Add q constraint.
10071 (arc_clzsi2): Add type and length attributes.
10072 (arc_ctzsi2): Likewise.
10073 * config/arc/constraints.md (Usc): Update constraint, the
10074 assembler can parse two relocations for a single instruction.
10076 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10078 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10079 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10081 2017-08-31 Olivier Hainque <hainque@adacore.com>
10083 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10084 match as powerpc-wrs-vxworks*.
10086 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10088 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10089 register constraint for by-element operand.
10090 (aarch64_mls_elt_merge<mode>): Likewise.
10092 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10094 * config/arc/arc.c (arc_can_follow_jump): Check for short
10097 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10099 * config.gcc: Use g.opt for arc.
10100 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10101 functionality moved to ...
10102 (legitimate_scaled_address_p): New function, ...here.
10103 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10104 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10105 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10107 (arc_override_options): Handle G option.
10108 (arc_output_pic_addr_const): Correct function definition.
10109 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10110 (arc_decl_anon_ns_mem_p): Delete.
10111 (arc_in_small_data_p): Overhaul this function to take into
10112 consideration the value given via G option.
10113 (arc_rewrite_small_data_1): Renamed and corrected old
10114 arc_rewrite_small_data function.
10115 (arc_rewrite_small_data): New function.
10116 (small_data_pattern): Don't use pic_offset_table_rtx.
10117 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10118 * config/arc/simdext.md (movmisalignv2hi): Use
10119 prepare_move_operands function.
10121 (movmisalign*): Likewise.
10122 * doc/invoke.texi (ARC options): Document -G option.
10124 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10126 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10128 * config/arc/arc.c (arc_print_operand): Output scalled address for
10129 sdata whenever is possible.
10130 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10131 load/stores are available.
10132 (compact_sda_memory_operand): Check for the alignment required by
10133 code density instructions.
10134 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10136 * config/arc/constraints.md (Usd): Update constraint.
10137 (Us0): New constraint.
10138 (Usc): Update constraint.
10140 2017-08-31 Richard Biener <rguenther@suse.de>
10142 PR middle-end/82054
10143 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10144 function only once.
10146 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10148 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10149 Resize type_signature.
10151 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10152 Alan Hayward <alan.hayward@arm.com>
10153 David Sherwood <david.sherwood@arm.com>
10155 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10156 subregs whose inner modes can be stored in GPRs.
10157 (aarch64_classify_index): Likewise.
10159 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10160 Alan Hayward <alan.hayward@arm.com>
10161 David Sherwood <david.sherwood@arm.com>
10163 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10164 (V_INT_EQUIV): ...this.
10165 (v_cmp_result): Rename to...
10166 (v_int_equiv): ...this.
10167 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10168 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10169 (copysign<mode>3): Likewise.
10170 (aarch64_simd_bsl<mode>_internal): Likewise.
10171 (aarch64_simd_bsl<mode>): Likewise.
10172 (vec_cmp<mode><mode>): Likewise.
10173 (vcond<mode><mode>): Likewise.
10174 (vcond<v_cmp_mixed><mode>): Likewise.
10175 (vcondu<mode><v_cmp_mixed>): Likewise.
10176 (aarch64_cm<optab><mode>): Likewise.
10177 (aarch64_cmtst<mode>): Likewise.
10178 (aarch64_fac<optab><mode>): Likewise.
10179 (vec_perm_const<mode>): Likewise.
10180 (vcond_mask_<mode><v_cmp_result>): Rename to...
10181 (vcond_mask_<mode><v_int_equiv>): ...this.
10182 (vec_cmp<mode><v_cmp_result>): Rename to...
10183 (vec_cmp<mode><v_int_equiv>): ...this.
10185 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10186 Alan Hayward <alan.hayward@arm.com>
10187 David Sherwood <david.sherwood@arm.com>
10189 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10191 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10192 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10193 (UNSPEC_LD4_DREG): New unspecs.
10194 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10195 (aarch64_ld2<mode>_dreg_be): Replace with...
10196 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10198 (aarch64_ld3<mode>_dreg_le)
10199 (aarch64_ld3<mode>_dreg_be): Replace with...
10200 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10202 (aarch64_ld4<mode>_dreg_le)
10203 (aarch64_ld4<mode>_dreg_be): Replace with...
10204 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10207 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10209 PR tree-optimization/81987
10210 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10211 insert an initializer in a location not dominated by the stride
10214 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10216 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10217 on the entire header of the finally block in the fallthru case.
10219 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10221 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10223 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10225 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10226 rs6000_emit_move_from_cr and call renamed function.
10227 (rs6000_emit_prologue): Call renamed functions.
10228 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10229 movesi_from_cr, remove volatile CRs.
10231 2017-08-30 Jon Beniston <jon@beniston.com>
10232 Richard Biener <rguenther@suse.de>
10234 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10235 of VECTOR_MODE_P check.
10236 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10237 element vector types.
10239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10241 * df.h (df_read_modify_subreg_p): Remove in favor of...
10242 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10243 const_rtx instead of an rtx.
10244 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10245 * df-problems.c (df_word_lr_mark_ref): Likewise.
10246 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10247 (mark_pseudo_reg_dead): Likewise.
10248 (mark_ref_dead): Likewise.
10249 * reginfo.c (init_subregs_of_mode): Likewise.
10250 * sched-deps.c (sched_analyze_1): Likewise.
10251 * df-scan.c (df_def_record_1): Likewise.
10252 (df_uses_record): Likewise.
10253 (df_read_modify_subreg_p): Remove in favor of...
10254 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10255 const_rtx instead of an rtx.
10257 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10258 Alan Hayward <alan.hayward@arm.com>
10259 David Sherwood <david.sherwood@arm.com>
10261 * rtl.h (partial_subreg_p): New function.
10262 * caller-save.c (save_call_clobbered_regs): Use it.
10263 * calls.c (expand_call): Likewise.
10264 * combine.c (combinable_i3pat): Likewise.
10265 (simplify_set): Likewise.
10266 (make_extraction): Likewise.
10267 (make_compound_operation_int): Likewise.
10268 (gen_lowpart_or_truncate): Likewise.
10269 (force_to_mode): Likewise.
10270 (make_field_assignment): Likewise.
10271 (reg_truncated_to_mode): Likewise.
10272 (record_truncated_value): Likewise.
10273 (move_deaths): Likewise.
10274 * cse.c (record_jump_cond): Likewise.
10275 (cse_insn): Likewise.
10276 * cselib.c (cselib_lookup_1): Likewise.
10277 * expmed.c (extract_bit_field_using_extv): Likewise.
10278 * function.c (assign_parm_setup_reg): Likewise.
10279 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10280 * ira-build.c (create_insn_allocnos): Likewise.
10281 * lra-coalesce.c (merge_pseudos): Likewise.
10282 * lra-constraints.c (match_reload): Likewise.
10283 (simplify_operand_subreg): Likewise.
10284 (curr_insn_transform): Likewise.
10285 * lra-lives.c (process_bb_lives): Likewise.
10286 * lra.c (new_insn_reg): Likewise.
10287 (lra_substitute_pseudo): Likewise.
10288 * regcprop.c (mode_change_ok): Likewise.
10289 (maybe_mode_change): Likewise.
10290 (copyprop_hardreg_forward_1): Likewise.
10291 * reload.c (push_reload): Likewise.
10292 (find_reloads): Likewise.
10293 (find_reloads_subreg_address): Likewise.
10294 * reload1.c (alter_reg): Likewise.
10295 (eliminate_regs_1): Likewise.
10296 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10298 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10300 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10301 back to if statements, including unpack.
10303 2017-08-30 Martin Liska <mliska@suse.cz>
10305 PR inline-asm/82001
10306 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10308 (func_checker::compare_asm_inputs_outputs): ... this function.
10309 (func_checker::compare_gimple_asm): Use the function to compare
10310 also ASM constrains.
10311 * ipa-icf-gimple.h: Rename the function.
10313 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10314 Alan Hayward <alan.hayward@arm.com>
10315 David Sherwood <david.sherwood@arm.com>
10317 * coretypes.h (complex_mode): New type.
10318 * gdbhooks.py (build_pretty_printer): Handle it.
10319 * machmode.h (complex_mode): New class.
10320 (complex_mode::includes_p): New function.
10321 (is_complex_int_mode): Likewise.
10322 (is_complex_float_mode): Likewise.
10323 * genmodes.c (get_mode_class): Handle complex mode classes.
10324 * function.c (expand_function_end): Use is_complex_int_mode.
10326 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10327 Alan Hayward <alan.hayward@arm.com>
10328 David Sherwood <david.sherwood@arm.com>
10330 * coretypes.h (scalar_mode_pod): New typedef.
10331 * gdbhooks.py (build_pretty_printer): Handle it.
10332 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10333 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10334 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10335 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10336 as_a <scalar_mode>.
10338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10339 Alan Hayward <alan.hayward@arm.com>
10340 David Sherwood <david.sherwood@arm.com>
10342 * machmode.h (mode_for_vector): Take a scalar_mode instead
10344 * stor-layout.c (mode_for_vector): Likewise.
10345 * explow.c (promote_mode): Use as_a <scalar_mode>.
10346 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10349 Alan Hayward <alan.hayward@arm.com>
10350 David Sherwood <david.sherwood@arm.com>
10352 * target.def (preferred_simd_mode): Take a scalar_mode
10353 instead of a machine_mode.
10354 * targhooks.h (default_preferred_simd_mode): Likewise.
10355 * targhooks.c (default_preferred_simd_mode): Likewise.
10356 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10357 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10358 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10359 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10360 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10361 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10362 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10363 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10365 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10366 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10367 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10368 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10369 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10370 * doc/tm.texi: Regenerate.
10371 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10375 Alan Hayward <alan.hayward@arm.com>
10376 David Sherwood <david.sherwood@arm.com>
10378 * target.def (scalar_mode_supported_p): Take a scalar_mode
10379 instead of a machine_mode.
10380 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10381 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10382 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10383 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10384 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10385 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10386 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10387 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10388 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10389 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10390 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10391 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10392 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10393 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10395 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10396 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10397 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10398 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10399 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10401 * doc/tm.texi: Regenerate.
10403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10404 Alan Hayward <alan.hayward@arm.com>
10405 David Sherwood <david.sherwood@arm.com>
10407 * coretypes.h (opt_scalar_mode): New typedef.
10408 * gdbhooks.py (build_pretty_printers): Handle it.
10409 * machmode.h (mode_iterator::get_2xwider): Add overload for
10411 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10413 * expr.c (convert_mode_scalar): Likewise.
10414 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10415 * optabs.c (expand_float): Likewise.
10416 (expand_fix): Likewise.
10417 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10420 Alan Hayward <alan.hayward@arm.com>
10421 David Sherwood <david.sherwood@arm.com>
10423 * optabs.c (expand_float): Explicitly check for scalars before
10424 using a branching expansion.
10425 (expand_fix): Likewise.
10427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10428 Alan Hayward <alan.hayward@arm.com>
10429 David Sherwood <david.sherwood@arm.com>
10431 * expr.c (convert_mode): Split scalar handling out into...
10432 (convert_mode_scalar): ...this new function. Treat the modes
10435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10436 Alan Hayward <alan.hayward@arm.com>
10437 David Sherwood <david.sherwood@arm.com>
10439 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10441 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10444 Alan Hayward <alan.hayward@arm.com>
10445 David Sherwood <david.sherwood@arm.com>
10447 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10448 rather than a machine_mode.
10449 (fixed_from_string): Likewise.
10450 (fixed_convert): Likewise.
10451 (fixed_convert_from_int): Likewise.
10452 (fixed_convert_from_real): Likewise.
10453 (real_convert_from_fixed): Likewise.
10454 * fixed-value.c (fixed_from_double_int): Likewise.
10455 (fixed_from_string): Likewise.
10456 (fixed_convert): Likewise.
10457 (fixed_convert_from_int): Likewise.
10458 (fixed_convert_from_real): Likewise.
10459 (real_convert_from_fixed): Likewise.
10460 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10463 Alan Hayward <alan.hayward@arm.com>
10464 David Sherwood <david.sherwood@arm.com>
10466 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10467 of separate mode class checks. Do not allow vector modes here.
10468 (immed_wide_int_const): Use as_a <scalar_mode>.
10469 * explow.c (trunc_int_for_mode): Likewise.
10470 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10471 (wi::shwi): Likewise.
10472 (wi::min_value): Likewise.
10473 (wi::max_value): Likewise.
10474 * dwarf2out.c (loc_descriptor): Likewise.
10475 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10476 for CONST_WIDE_INT.
10478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10479 Alan Hayward <alan.hayward@arm.com>
10480 David Sherwood <david.sherwood@arm.com>
10482 * tree.h (SCALAR_TYPE_MODE): New macro.
10483 * expr.c (expand_expr_addr_expr_1): Use it.
10484 (expand_expr_real_2): Likewise.
10485 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10486 (fold_convert_const_fixed_from_int): Likewise.
10487 (fold_convert_const_fixed_from_real): Likewise.
10488 (native_encode_fixed): Likewise
10489 (native_encode_complex): Likewise
10490 (native_encode_vector): Likewise.
10491 (native_interpret_fixed): Likewise.
10492 (native_interpret_real): Likewise.
10493 (native_interpret_complex): Likewise.
10494 (native_interpret_vector): Likewise.
10495 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10496 (simd_clone_adjust_argument_types): Likewise.
10497 (simd_clone_init_simd_arrays): Likewise.
10498 (simd_clone_adjust): Likewise.
10499 * stor-layout.c (layout_type): Likewise.
10500 * tree.c (build_minus_one_cst): Likewise.
10501 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10502 * tree-inline.c (estimate_move_cost): Likewise.
10503 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10504 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10505 (vectorizable_reduction): Likewise.
10506 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10507 (vect_recog_mixed_size_cond_pattern): Likewise.
10508 (check_bool_pattern): Likewise.
10509 (adjust_bool_pattern): Likewise.
10510 (search_type_for_mask_1): Likewise.
10511 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10512 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10513 (vectorizable_load): Likewise.
10514 (vectorizable_store): Likewise.
10515 * ubsan.c (ubsan_encode_value): Likewise.
10516 * varasm.c (output_constant): Likewise.
10518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10519 Alan Hayward <alan.hayward@arm.com>
10520 David Sherwood <david.sherwood@arm.com>
10522 * coretypes.h (scalar_mode): New class.
10523 * machmode.h (scalar_mode): Likewise.
10524 (scalar_mode::includes_p): New function.
10525 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10526 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10527 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10528 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10529 * expmed.c (store_bit_field_1): Likewise.
10530 (extract_bit_field_1): Likewise.
10531 * expr.c (write_complex_part): Likewise.
10532 (read_complex_part): Likewise.
10533 (emit_move_complex_push): Likewise.
10534 (expand_expr_real_2): Likewise.
10535 * function.c (assign_parm_setup_reg): Likewise.
10536 (assign_parms_unsplit_complex): Likewise.
10537 * optabs.c (expand_binop): Likewise.
10538 * rtlanal.c (subreg_get_info): Likewise.
10539 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10540 * varasm.c (output_constant_pool_2): Likewise.
10542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10543 Alan Hayward <alan.hayward@arm.com>
10544 David Sherwood <david.sherwood@arm.com>
10546 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10548 (expmed_mult_highpart_optab): Likewise.
10549 (expmed_mult_highpart): Likewise.
10551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10552 Alan Hayward <alan.hayward@arm.com>
10553 David Sherwood <david.sherwood@arm.com>
10555 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10556 instead of a machine_mode.
10557 (builtin_memset_read_str): Likewise.
10558 * builtins.c (c_readstr): Likewise.
10559 (builtin_memcpy_read_str): Likewise.
10560 (builtin_strncpy_read_str): Likewise.
10561 (builtin_memset_read_str): Likewise.
10562 (builtin_memset_gen_str): Likewise.
10563 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10564 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10565 instead of a machine_mode.
10566 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10568 (make_extraction): Likewise.
10569 (try_widen_shift_mode): Take and return scalar_int_modes instead
10571 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10572 a scalar_int_mode instead of a machine_mode.
10573 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10574 (avr_addr_space_pointer_mode): Likewise.
10575 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10576 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10577 (msp430_unwind_word_mode): Likewise.
10578 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10579 (spu_addr_space_pointer_mode): Likewise.
10580 (spu_addr_space_address_mode): Likewise.
10581 (spu_libgcc_cmp_return_mode): Likewise.
10582 (spu_libgcc_shift_count_mode): Likewise.
10583 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10584 (rl78_addr_space_pointer_mode): Likewise.
10585 (fl78_unwind_word_mode): Likewise.
10586 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10588 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10589 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10590 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10591 (mips_valid_pointer_mode): Likewise.
10592 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10593 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10594 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10596 (ft32_addr_space_address_mode): Likewise.
10597 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10598 scalar_int_mode instead of a machine_mode.
10599 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10601 (m32c_addr_space_address_mode): Likewise.
10602 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10603 (rs6000_eh_return_filter_mode): Likewise.
10604 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10605 (rs6000_eh_return_filter_mode): Likewise.
10606 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10607 (s390_libgcc_shift_count_mode): Likewise.
10608 (s390_unwind_word_mode): Likewise.
10609 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10611 * target.def (mode_rep_extended): Likewise.
10612 (valid_pointer_mode): Likewise.
10613 (addr_space.valid_pointer_mode): Likewise.
10614 (eh_return_filter_mode): Return a scalar_int_mode rather than
10616 (libgcc_cmp_return_mode): Likewise.
10617 (libgcc_shift_count_mode): Likewise.
10618 (unwind_word_mode): Likewise.
10619 (addr_space.pointer_mode): Likewise.
10620 (addr_space.address_mode): Likewise.
10621 * doc/tm.texi: Regenerate.
10622 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10624 (do_jump): Use scalar_int_mode for local variables.
10625 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10626 rather than a machine_mode.
10627 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10628 (scompare_loc_descriptor_wide): Likewise.
10629 (scompare_loc_descriptor_narrow): Likewise.
10630 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10632 * except.c (sjlj_emit_dispatch_table): Likewise.
10633 (expand_builtin_eh_copy_values): Likewise.
10634 * explow.c (convert_memory_address_addr_space_1): Likewise.
10635 Take a scalar_int_mode rather than a machine_mode.
10636 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10637 than a machine_mode.
10638 (memory_address_addr_space): Use scalar_int_mode for local variables.
10639 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10640 rather than a machine_mode.
10641 * expmed.c (mask_rtx): Likewise.
10642 (init_expmed_one_conv): Likewise.
10643 (expand_mult_highpart_adjust): Likewise.
10644 (extract_high_half): Likewise.
10645 (expmed_mult_highpart_optab): Likewise.
10646 (expmed_mult_highpart): Likewise.
10647 (expand_smod_pow2): Likewise.
10648 (expand_sdiv_pow2): Likewise.
10649 (emit_store_flag_int): Likewise.
10650 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10652 (extract_low_bits): Likewise.
10653 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10655 * expr.c (pieces_addr::adjust): Likewise.
10656 (can_store_by_pieces): Likewise.
10657 (store_by_pieces): Likewise.
10658 (clear_by_pieces_1): Likewise.
10659 (expand_expr_addr_expr_1): Likewise.
10660 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10661 (expand_expr_real_1): Likewise.
10662 (try_casesi): Likewise.
10663 * final.c (shorten_branches): Likewise.
10664 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10665 type of "mode" to machine_mode.
10666 * internal-fn.c (expand_arith_overflow_result_store): Take a
10667 scalar_int_mode rather than a machine_mode.
10668 (expand_mul_overflow): Use scalar_int_mode for local variables.
10669 * loop-doloop.c (doloop_modify): Likewise.
10670 (doloop_optimize): Likewise.
10671 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10672 than a machine_mode.
10673 (expand_doubleword_shift_condmove): Likewise.
10674 (expand_doubleword_shift): Likewise.
10675 (expand_doubleword_clz): Likewise.
10676 (expand_doubleword_popcount): Likewise.
10677 (expand_doubleword_parity): Likewise.
10678 (expand_absneg_bit): Use scalar_int_mode for local variables.
10679 (prepare_float_lib_cmp): Likewise.
10680 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10681 rather than a machine_mode.
10682 (convert_memory_address_addr_space): Likewise.
10683 (get_mode_bounds): Likewise.
10684 (get_address_mode): Return a scalar_int_mode rather than a
10686 * rtlanal.c (get_address_mode): Likewise.
10687 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10688 than a machine_mode.
10689 * targhooks.c (default_mode_rep_extended): Likewise.
10690 (default_valid_pointer_mode): Likewise.
10691 (default_addr_space_valid_pointer_mode): Likewise.
10692 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10693 than a machine_mode.
10694 (default_libgcc_cmp_return_mode): Likewise.
10695 (default_libgcc_shift_count_mode): Likewise.
10696 (default_unwind_word_mode): Likewise.
10697 (default_addr_space_pointer_mode): Likewise.
10698 (default_addr_space_address_mode): Likewise.
10699 * targhooks.h (default_eh_return_filter_mode): Likewise.
10700 (default_libgcc_cmp_return_mode): Likewise.
10701 (default_libgcc_shift_count_mode): Likewise.
10702 (default_unwind_word_mode): Likewise.
10703 (default_addr_space_pointer_mode): Likewise.
10704 (default_addr_space_address_mode): Likewise.
10705 (default_mode_rep_extended): Take a scalar_int_mode rather than
10707 (default_valid_pointer_mode): Likewise.
10708 (default_addr_space_valid_pointer_mode): Likewise.
10709 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10711 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10712 rather than a machine_mode.
10713 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10714 for local variables.
10715 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10716 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10717 than a machine_mode.
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 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10724 the mode argument to scalar_int_mode.
10725 (do_jump_by_parts_zero_rtx): Likewise.
10726 (do_jump_by_parts_equality_rtx): Likewise.
10727 (do_jump_by_parts_greater): Take a mode argument.
10728 (do_jump_by_parts_equality): Likewise.
10729 (do_jump_1): Update calls accordingly.
10731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10732 Alan Hayward <alan.hayward@arm.com>
10733 David Sherwood <david.sherwood@arm.com>
10735 * is-a.h (safe_dyn_cast): New function.
10736 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10737 (jump_table_for_label): Likewise.
10738 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10739 instead of an rtx_insn *.
10740 (shorten_branches): Use dyn_cast instead of LABEL_P and
10741 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10742 rtx_jump_table_data::get_data_mode.
10743 (final_scan_insn): Likewise.
10745 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10746 Alan Hayward <alan.hayward@arm.com>
10747 David Sherwood <david.sherwood@arm.com>
10749 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10750 trying to combine a full-register integer set with a subreg
10753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10754 Alan Hayward <alan.hayward@arm.com>
10755 David Sherwood <david.sherwood@arm.com>
10757 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10758 that is always either address_mode or pointer_mode.
10760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10761 Alan Hayward <alan.hayward@arm.com>
10762 David Sherwood <david.sherwood@arm.com>
10764 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10765 when the two are known to be equal.
10767 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10768 Alan Hayward <alan.hayward@arm.com>
10769 David Sherwood <david.sherwood@arm.com>
10771 * simplify-rtx.c (simplify_const_unary_operation): Use
10772 is_a <scalar_int_mode> instead of checking for a nonzero
10773 precision. Forcibly convert op_mode to a scalar_int_mode
10774 in that case. More clearly differentiate the operand and
10775 result modes and use the former when deciding what the value
10776 of a count-bits operation should be. Use is_int_mode instead
10777 of checking for a MODE_INT. Remove redundant check for whether
10778 this mode has a zero precision.
10780 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10781 Alan Hayward <alan.hayward@arm.com>
10782 David Sherwood <david.sherwood@arm.com>
10784 * optabs.c (widen_leading): Change the type of the mode argument
10785 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10786 (widen_bswap): Likewise.
10787 (expand_parity): Likewise.
10788 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10789 (expand_ffs): Likewise.
10790 (epand_unop): Check for scalar integer modes before calling the
10793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10794 Alan Hayward <alan.hayward@arm.com>
10795 David Sherwood <david.sherwood@arm.com>
10797 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10799 (expand_expr_real_1): Update call accordingly.
10801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10802 Alan Hayward <alan.hayward@arm.com>
10803 David Sherwood <david.sherwood@arm.com>
10805 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10806 value_mode arguments. Use scalar_int_mode internally.
10807 (store_bit_field_1): Rename the new integer mode from imode
10808 to op0_mode and use it instead of GET_MODE (op0). Update calls
10809 to store_split_bit_field, store_bit_field_using_insv and
10810 store_fixed_bit_field.
10811 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10812 Use scalar_int_mode internally. Use a bit count rather than a mode
10813 when calculating the largest bit size for get_best_mode.
10814 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10815 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10816 Remove assertion that OP0 has a scalar integer mode.
10817 (store_split_bit_field): Add op0_mode and value_mode arguments.
10818 Update calls to extract_fixed_bit_field.
10819 (extract_bit_field_using_extv): Add an op0_mode argument.
10820 Use scalar_int_mode internally.
10821 (extract_bit_field_1): Rename the new integer mode from imode to
10822 op0_mode and use it instead of GET_MODE (op0). Update calls to
10823 extract_split_bit_field, extract_bit_field_using_extv and
10824 extract_fixed_bit_field.
10825 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10826 to extract_split_bit_field and extract_fixed_bit_field_1.
10827 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10828 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10829 on the target mode.
10830 (extract_split_bit_field): Add an op0_mode argument. Update call
10831 to extract_fixed_bit_field.
10833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10834 Alan Hayward <alan.hayward@arm.com>
10835 David Sherwood <david.sherwood@arm.com>
10837 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10838 * explow.c (hard_function_value): Likewise.
10839 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10840 convert_to_mode call outside the loop.
10841 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10842 for the mode iterator. Require the mode specified by max_pieces
10844 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10846 (copy_blkmode_to_reg): Likewise.
10847 (set_storage_via_setmem): Likewise.
10848 * optabs.c (prepare_cmp_insn): Likewise.
10849 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10850 * stor-layout.c (finish_bitfield_representative): Likewise.
10852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10853 Alan Hayward <alan.hayward@arm.com>
10854 David Sherwood <david.sherwood@arm.com>
10856 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10857 * expr.c (convert_move): Use them.
10858 (convert_modes): Likewise.
10859 (store_expr_with_bounds): Likewise.
10861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10862 Alan Hayward <alan.hayward@arm.com>
10863 David Sherwood <david.sherwood@arm.com>
10865 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10866 parameter for the mode of "x". Remove the "known_x", "known_mode"
10867 and "known_ret" arguments. Change the type of the mode argument
10868 to scalar_int_mode.
10869 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10870 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10871 (reg_num_sign_bit_copies_for_combine): Likewise.
10872 * rtlanal.c (nonzero_bits1): Likewise.
10873 (num_sign_bit_copies1): Likewise.
10874 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10875 (reg_num_sign_bit_copies_general): Likewise.
10876 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10877 (reg_nonzero_bits_general): Likewise.
10879 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10880 Alan Hayward <alan.hayward@arm.com>
10881 David Sherwood <david.sherwood@arm.com>
10883 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10884 than in subroutines. Return 1 for non-integer modes.
10885 (cached_num_sign_bit_copies): Change the type of the mode parameter
10886 to scalar_int_mode.
10887 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10888 classes. Handle CONST_INT_P first and then check whether X also
10889 has a scalar integer mode. Check the same thing for inner registers
10890 of a SUBREG and for values that are being extended or truncated.
10892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10893 Alan Hayward <alan.hayward@arm.com>
10894 David Sherwood <david.sherwood@arm.com>
10896 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10897 in subroutines. Return the mode mask for non-integer modes.
10898 (cached_nonzero_bits): Change the type of the mode parameter
10899 to scalar_int_mode.
10900 (nonzero_bits1): Likewise. Remove early exit for other mode
10901 classes. Handle CONST_INT_P first and then check whether X
10902 also has a scalar integer mode.
10904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10905 Alan Hayward <alan.hayward@arm.com>
10906 David Sherwood <david.sherwood@arm.com>
10908 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10909 Return a scalar_int_mode and assert that the size is greater than
10911 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10912 (op_by_pieces_d::op_by_pieces_d): Likewise.
10913 (op_by_pieces_d::run): Likewise.
10914 (can_store_by_pieces): Likewise.
10916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10917 Alan Hayward <alan.hayward@arm.com>
10918 David Sherwood <david.sherwood@arm.com>
10920 * combine.c (extract_left_shift): Add a mode argument and update
10922 (make_compound_operation_int): Change the type of the mode parameter
10923 to scalar_int_mode and update the call to extract_left_shift.
10925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10926 Alan Hayward <alan.hayward@arm.com>
10927 David Sherwood <david.sherwood@arm.com>
10929 * combine.c (simplify_and_const_int): Change the type of the mode
10930 parameter to scalar_int_mode.
10931 (simplify_and_const_int_1): Likewise. Update recursive call.
10933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10934 Alan Hayward <alan.hayward@arm.com>
10935 David Sherwood <david.sherwood@arm.com>
10937 * combine.c (simplify_compare_const): Check that the mode is a
10938 scalar_int_mode (rather than VOIDmode) before testing its
10940 (simplify_comparison): Move COMPARISON_P handling out of the
10941 loop and restrict the latter part of the loop to scalar_int_modes.
10942 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10943 and when considering SUBREG_REGs. Use is_int_mode instead of
10944 checking GET_MODE_CLASS against MODE_INT.
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_widen_shift_mode): Move check for equal modes to...
10951 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10952 shift_unit_mode and for modes involved in scalar shifts.
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 * combine.c (force_int_to_mode): New function, split out from...
10959 (force_to_mode): ...here. Keep xmode up-to-date and use it
10960 instead of GET_MODE (x).
10962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10963 Alan Hayward <alan.hayward@arm.com>
10964 David Sherwood <david.sherwood@arm.com>
10966 * optabs-query.h (extraction_insn::struct_mode): Change type to
10967 opt_scalar_int_mode and update comment.
10968 (extraction_insn::field_mode): Change type to scalar_int_mode.
10969 (extraction_insn::pos_mode): Likewise.
10970 * combine.c (make_extraction): Update accordingly.
10971 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10972 (get_optab_extraction_insn): Likewise.
10973 * recog.c (simplify_while_replacing): Likewise.
10974 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10975 parameter to opt_scalar_int_mode.
10977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10978 Alan Hayward <alan.hayward@arm.com>
10979 David Sherwood <david.sherwood@arm.com>
10981 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10982 to a scalar_int_mode instead of a machine_mode.
10983 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10984 (get_best_mode): Return a boolean and use a pointer argument to store
10985 the selected mode. Replace the limit mode parameter with a bit limit.
10986 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10987 for the values returned by bit_field_mode_iterator::next_mode.
10988 (store_bit_field): Update call to get_best_mode.
10989 (store_fixed_bit_field): Likewise.
10990 (extract_fixed_bit_field): Likewise.
10991 * expr.c (optimize_bitfield_assignment_op): Likewise.
10992 * fold-const.c (optimize_bit_field_compare): Likewise.
10993 (fold_truth_andor_1): Likewise.
10994 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10995 Update for new type of m_mode.
10996 (get_best_mode): As above.
10998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10999 Alan Hayward <alan.hayward@arm.com>
11000 David Sherwood <david.sherwood@arm.com>
11002 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11003 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11004 (store_bit_field): Check is_a <scalar_int_mode> before calling
11005 strict_volatile_bitfield_p.
11006 (extract_bit_field): Likewise.
11008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11009 Alan Hayward <alan.hayward@arm.com>
11010 David Sherwood <david.sherwood@arm.com>
11012 * target.def (cstore_mode): Return a scalar_int_mode.
11013 * doc/tm.texi: Regenerate.
11014 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11015 * targhooks.h (default_cstore_mode): Likewise.
11016 * targhooks.c (default_cstore_mode): Likewise, using a forced
11018 * expmed.c (emit_cstore): Expect the target of the cstore to be
11021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11022 Alan Hayward <alan.hayward@arm.com>
11023 David Sherwood <david.sherwood@arm.com>
11025 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11027 (niter_desc): Likewise mode.
11028 (iv_analyze): Add a mode parameter.
11030 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11031 and change its type to scalar_int_mode.
11032 * loop-iv.c: Update commentary at head of file.
11033 (iv_constant): Pass the mode paraeter before the rtx it describes
11034 and change its type to scalar_int_mode. Remove VOIDmode handling.
11035 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11036 (iv_extend): Likewise.
11037 (shorten_into_mode): Likewise.
11038 (iv_add): Use scalar_int_mode.
11039 (iv_mult): Likewise.
11040 (iv_shift): Likewise.
11041 (canonicalize_iv_subregs): Likewise.
11042 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11043 it describes and change its mode to scalar_int_mode. Also change
11044 the type of the returned inner_mode to scalar_int_mode.
11045 (get_biv_step): Likewise, turning outer_mode from a pointer
11046 into a direct parameter. Update call to get_biv_step_1.
11047 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11048 iv_constant and get_biv_step.
11049 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11050 and change its type to scalar_int_mode. Don't initialise iv->mode
11051 to VOIDmode and remove later checks for its still being VOIDmode.
11052 Update calls to iv_analyze_op and iv_analyze_expr. Check
11053 is_a <scalar_int_mode> when changing the mode under consideration.
11054 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11055 Update call to iv_analyze_expr.
11056 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11057 inner register is not also a scalar_int_mode. Update call to
11059 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11060 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11061 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11062 separate mode class checks. Update calls to iv_analyze. Remove
11063 fix-up of VOIDmodes after iv_analyze_biv.
11064 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11065 don't have a scalar_int_mode. Update call to biv_p.
11067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11068 Alan Hayward <alan.hayward@arm.com>
11069 David Sherwood <david.sherwood@arm.com>
11071 * cfgexpand.c (convert_debug_memory_address): Use
11072 as_a <scalar_int_mode>.
11073 * combine.c (expand_compound_operation): Likewise.
11074 (make_extraction): Likewise.
11075 (change_zero_ext): Likewise.
11076 (simplify_comparison): Likewise.
11077 * cse.c (cse_insn): Likewise.
11078 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11079 (mem_loc_descriptor): Likewise.
11080 (loc_descriptor): Likewise.
11081 * expmed.c (init_expmed_one_mode): Likewise.
11082 (synth_mult): Likewise.
11083 (emit_store_flag_1): Likewise.
11084 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11085 of a comparison with size.
11086 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11087 (reduce_to_bit_field_precision): Likewise.
11088 * function.c (expand_function_end): Likewise.
11089 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11090 * loop-doloop.c (doloop_modify): Likewise.
11091 * optabs.c (expand_binop): Likewise.
11092 (expand_unop): Likewise.
11093 (expand_copysign_absneg): Likewise.
11094 (prepare_cmp_insn): Likewise.
11095 (maybe_legitimize_operand): Likewise.
11096 * recog.c (const_scalar_int_operand): Likewise.
11097 * rtlanal.c (get_address_mode): Likewise.
11098 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11099 (simplify_cond_clz_ctz): Likewise.
11100 * tree-nested.c (get_nl_goto_field): Likewise.
11101 * tree.c (build_vector_type_for_mode): Likewise.
11102 * var-tracking.c (use_narrower_mode): Likewise.
11104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11105 Alan Hayward <alan.hayward@arm.com>
11106 David Sherwood <david.sherwood@arm.com>
11108 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11109 * builtins.c (expand_builtin_signbit): Use it.
11110 * cfgexpand.c (expand_debug_expr): Likewise.
11111 * dojump.c (do_jump): Likewise.
11112 (do_compare_and_jump): Likewise.
11113 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11114 * expmed.c (make_tree): Likewise.
11115 * expr.c (expand_expr_real_2): Likewise.
11116 (expand_expr_real_1): Likewise.
11117 (try_casesi): Likewise.
11118 * fold-const-call.c (fold_const_call_ss): Likewise.
11119 * fold-const.c (unextend): Likewise.
11120 (extract_muldiv_1): Likewise.
11121 (fold_single_bit_test): Likewise.
11122 (native_encode_int): Likewise.
11123 (native_encode_string): Likewise.
11124 (native_interpret_int): Likewise.
11125 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11126 * internal-fn.c (expand_addsub_overflow): Likewise.
11127 (expand_neg_overflow): Likewise.
11128 (expand_mul_overflow): Likewise.
11129 (expand_arith_overflow): Likewise.
11130 * match.pd: Likewise.
11131 * stor-layout.c (layout_type): Likewise.
11132 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11133 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11134 * tree-ssanames.c (get_range_info): Likewise.
11135 * tree-switch-conversion.c (array_value_type) Likewise.
11136 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11137 (vect_recog_divmod_pattern): Likewise.
11138 (vect_recog_mixed_size_cond_pattern): Likewise.
11139 * tree-vrp.c (extract_range_basic): Likewise.
11140 (simplify_float_conversion_using_ranges): Likewise.
11141 * tree.c (int_fits_type_p): Likewise.
11142 * ubsan.c (instrument_bool_enum_load): Likewise.
11143 * varasm.c (mergeable_string_section): Likewise.
11144 (narrowing_initializer_constant_valid_p): Likewise.
11145 (output_constant): Likewise.
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 * machmode.h (NARROWEST_INT_MODE): New macro.
11152 * expr.c (alignment_for_piecewise_move): Use it instead of
11153 GET_CLASS_NARROWEST_MODE (MODE_INT).
11154 (push_block): Likewise.
11155 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11157 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11160 Alan Hayward <alan.hayward@arm.com>
11161 David Sherwood <david.sherwood@arm.com>
11163 * postreload.c (move2add_valid_value_p): Change the type of the
11164 mode parameter to scalar_int_mode.
11165 (move2add_use_add2_insn): Add a mode parameter and use it instead
11167 (move2add_use_add3_insn): Likewise.
11168 (reload_cse_move2add): Update accordingly.
11170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11171 Alan Hayward <alan.hayward@arm.com>
11172 David Sherwood <david.sherwood@arm.com>
11174 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11176 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11177 * optabs.c (expand_unop): Likewise.
11179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11180 Alan Hayward <alan.hayward@arm.com>
11181 David Sherwood <david.sherwood@arm.com>
11183 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11184 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11185 (popcount_loc_descriptor): Likewise.
11186 (bswap_loc_descriptor): Likewise.
11187 (rotate_loc_descriptor): Likewise.
11188 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11189 calling the functions above.
11191 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11192 Alan Hayward <alan.hayward@arm.com>
11193 David Sherwood <david.sherwood@arm.com>
11195 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11197 (try_combine): Likewise.
11198 (simplify_if_then_else): Likewise.
11199 * cse.c (cse_insn): Likewise.
11200 * dwarf2out.c (mem_loc_descriptor): Likewise.
11201 * emit-rtl.c (gen_lowpart_common): Likewise.
11202 * simplify-rtx.c (simplify_truncation): Likewise.
11203 (simplify_binary_operation_1): Likewise.
11204 (simplify_const_relational_operation): Likewise.
11205 (simplify_ternary_operation): Likewise.
11206 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11209 Alan Hayward <alan.hayward@arm.com>
11210 David Sherwood <david.sherwood@arm.com>
11212 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11213 * reload.c (push_reload): Likewise.
11214 (find_reloads): Likewise.
11216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11217 Alan Hayward <alan.hayward@arm.com>
11218 David Sherwood <david.sherwood@arm.com>
11220 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11221 (make_compound_operation_int): Likewise.
11222 (change_zero_ext): Likewise.
11223 * expr.c (convert_move): Likewise.
11224 (convert_modes): Likewise.
11225 * fwprop.c (forward_propagate_subreg): Likewise.
11226 * loop-iv.c (get_biv_step_1): Likewise.
11227 * optabs.c (widen_operand): Likewise.
11228 * postreload.c (move2add_valid_value_p): Likewise.
11229 * recog.c (simplify_while_replacing): Likewise.
11230 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11231 (simplify_binary_operation_1): Likewise. Remove redundant
11232 mode equality check.
11234 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11235 Alan Hayward <alan.hayward@arm.com>
11236 David Sherwood <david.sherwood@arm.com>
11238 * combine.c (combine_simplify_rtx): Add checks for
11239 is_a <scalar_int_mode>.
11240 (simplify_if_then_else): Likewise.
11241 (make_field_assignment): Likewise.
11242 (simplify_comparison): Likewise.
11243 * ifcvt.c (noce_try_bitop): Likewise.
11244 * loop-invariant.c (canonicalize_address_mult): Likewise.
11245 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11248 Alan Hayward <alan.hayward@arm.com>
11249 David Sherwood <david.sherwood@arm.com>
11251 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11252 is_a <scalar_int_mode> instead of != BLKmode.
11254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11255 Alan Hayward <alan.hayward@arm.com>
11256 David Sherwood <david.sherwood@arm.com>
11258 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11259 instead of != VOIDmode.
11260 * combine.c (if_then_else_cond): Likewise.
11261 (change_zero_ext): Likewise.
11262 * dwarf2out.c (mem_loc_descriptor): Likewise.
11263 (loc_descriptor): Likewise.
11264 * rtlanal.c (canonicalize_condition): Likewise.
11265 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11267 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11268 Alan Hayward <alan.hayward@arm.com>
11269 David Sherwood <david.sherwood@arm.com>
11271 * simplify-rtx.c (simplify_binary_operation_1): Use
11272 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11275 Alan Hayward <alan.hayward@arm.com>
11276 David Sherwood <david.sherwood@arm.com>
11278 * wide-int.h (int_traits<unsigned char>) New class.
11279 (int_traits<unsigned short>) Likewise.
11280 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11281 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11283 * combine.c (set_nonzero_bits_and_sign_copies): Use
11284 is_a <scalar_int_mode>.
11285 (find_split_point): Likewise.
11286 (combine_simplify_rtx): Likewise.
11287 (simplify_logical): Likewise.
11288 (expand_compound_operation): Likewise.
11289 (expand_field_assignment): Likewise.
11290 (make_compound_operation): Likewise.
11291 (extended_count): Likewise.
11292 (change_zero_ext): Likewise.
11293 (simplify_comparison): Likewise.
11294 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11295 (ucompare_loc_descriptor): Likewise.
11296 (minmax_loc_descriptor): Likewise.
11297 (mem_loc_descriptor): Likewise.
11298 (loc_descriptor): Likewise.
11299 * expmed.c (init_expmed_one_mode): Likewise.
11300 * lra-constraints.c (lra_constraint_offset): Likewise.
11301 * optabs.c (prepare_libcall_arg): Likewise.
11302 * postreload.c (move2add_note_store): Likewise.
11303 * reload.c (operands_match_p): Likewise.
11304 * rtl.h (load_extend_op): Likewise.
11305 * rtlhooks.c (gen_lowpart_general): Likewise.
11306 * simplify-rtx.c (simplify_truncation): Likewise.
11307 (simplify_unary_operation_1): Likewise.
11308 (simplify_binary_operation_1): Likewise.
11309 (simplify_const_binary_operation): Likewise.
11310 (simplify_const_relational_operation): Likewise.
11311 (simplify_subreg): Likewise.
11312 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11313 * var-tracking.c (adjust_mems): Likewise.
11314 (prepare_call_arguments): Likewise.
11316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11317 Alan Hayward <alan.hayward@arm.com>
11318 David Sherwood <david.sherwood@arm.com>
11320 * machmode.h (is_int_mode): New fuction.
11321 * combine.c (find_split_point): Use it.
11322 (combine_simplify_rtx): Likewise.
11323 (simplify_if_then_else): Likewise.
11324 (simplify_set): Likewise.
11325 (simplify_shift_const_1): Likewise.
11326 (simplify_comparison): Likewise.
11327 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11328 * cse.c (notreg_cost): Likewise.
11329 (cse_insn): Likewise.
11330 * cselib.c (cselib_lookup_1): Likewise.
11331 * dojump.c (do_jump_1): Likewise.
11332 (do_compare_rtx_and_jump): Likewise.
11333 * dse.c (get_call_args): Likewise.
11334 * dwarf2out.c (rtl_for_decl_init): Likewise.
11335 (native_encode_initializer): Likewise.
11336 * expmed.c (emit_store_flag_1): Likewise.
11337 (emit_store_flag): Likewise.
11338 * expr.c (convert_modes): Likewise.
11339 (store_field): Likewise.
11340 (expand_expr_real_1): Likewise.
11341 * fold-const.c (fold_read_from_constant_string): Likewise.
11342 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11343 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11344 * optabs.c (expand_binop): Likewise.
11345 (expand_unop): Likewise.
11346 (expand_abs_nojump): Likewise.
11347 (expand_one_cmpl_abs_nojump): Likewise.
11348 * simplify-rtx.c (mode_signbit_p): Likewise.
11349 (val_signbit_p): Likewise.
11350 (val_signbit_known_set_p): Likewise.
11351 (val_signbit_known_clear_p): Likewise.
11352 (simplify_relational_operation_1): Likewise.
11353 * tree.c (vector_type_mode): Likewise.
11355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11356 Alan Hayward <alan.hayward@arm.com>
11357 David Sherwood <david.sherwood@arm.com>
11359 * machmode.h (smallest_mode_for_size): Fix formatting.
11360 (smallest_int_mode_for_size): New function.
11361 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11362 instead of smallest_mode_for_size.
11363 * combine.c (make_extraction): Likewise.
11364 * config/arc/arc.c (arc_expand_movmem): Likewise.
11365 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11366 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11367 * config/s390/s390.c (s390_expand_insv): Likewise.
11368 * config/sparc/sparc.c (assign_int_registers): Likewise.
11369 * config/spu/spu.c (spu_function_value): Likewise.
11370 (spu_function_arg): Likewise.
11371 * coverage.c (get_gcov_type): Likewise.
11372 (get_gcov_unsigned_t): Likewise.
11373 * dse.c (find_shift_sequence): Likewise.
11374 * expmed.c (store_bit_field_1): Likewise.
11375 * expr.c (convert_move): Likewise.
11376 (store_field): Likewise.
11377 * internal-fn.c (expand_arith_overflow): Likewise.
11378 * optabs-query.c (get_best_extraction_insn): Likewise.
11379 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11380 * stor-layout.c (layout_type): Likewise.
11381 (initialize_sizetypes): Likewise.
11382 * targhooks.c (default_get_mask_mode): Likewise.
11383 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11386 Alan Hayward <alan.hayward@arm.com>
11387 David Sherwood <david.sherwood@arm.com>
11389 * machmode.h (opt_mode::else_blk): New function.
11390 (int_mode_for_mode): Declare.
11391 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11392 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11394 * cfgexpand.c (expand_debug_expr): Likewise.
11395 * combine.c (gen_lowpart_or_truncate): Likewise.
11396 (gen_lowpart_for_combine): Likewise.
11397 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11398 * config/avr/avr.c (avr_to_int_mode): Likewise.
11399 (avr_out_plus_1): Likewise.
11400 (avr_out_plus): Likewise.
11401 (avr_out_round): Likewise.
11402 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11403 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11404 (s390_expand_vcond): Likewise.
11405 * config/spu/spu.c (spu_split_immediate): Likewise.
11406 (spu_expand_mov): Likewise.
11407 * dse.c (get_stored_val): Likewise.
11408 * expmed.c (store_bit_field_1): Likewise.
11409 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11411 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11412 (extract_low_bits): Likewise.
11413 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11414 handling rather than repeating the check.
11415 (emit_group_store): Likewise.
11416 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11417 * optabs.c (expand_absneg_bit): Likewise.
11418 (expand_copysign_absneg): Likewise.
11419 (expand_copysign_bit): Likewise.
11420 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11421 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11422 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11423 * var-tracking.c (prepare_call_arguments): Likewise.
11424 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11425 int_mode_for_mode instead of mode_for_size.
11426 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11429 Alan Hayward <alan.hayward@arm.com>
11430 David Sherwood <david.sherwood@arm.com>
11432 * machmode.h (int_mode_for_size): New function.
11433 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11434 instead of mode_for_size.
11435 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11437 * combine.c (expand_field_assignment): Use int_mode_for_size
11438 instead of mode_for_size.
11439 (make_extraction): Likewise.
11440 (simplify_shift_const_1): Likewise.
11441 (simplify_comparison): Likewise.
11442 * dojump.c (do_jump): Likewise.
11443 * dwarf2out.c (mem_loc_descriptor): Likewise.
11444 * emit-rtl.c (init_derived_machine_modes): Likewise.
11445 * expmed.c (flip_storage_order): Likewise.
11446 (convert_extracted_bit_field): Likewise.
11447 * expr.c (copy_blkmode_from_reg): Likewise.
11448 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11449 * internal-fn.c (expand_mul_overflow): Likewise.
11450 * lower-subreg.c (simple_move): Likewise.
11451 * optabs-libfuncs.c (init_optabs): Likewise.
11452 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11453 * tree.c (vector_type_mode): Likewise.
11454 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11455 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11456 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11457 * tree-vect-stmts.c (vectorizable_load): Likewise.
11458 (vectorizable_store): Likewise.
11460 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11461 Alan Hayward <alan.hayward@arm.com>
11462 David Sherwood <david.sherwood@arm.com>
11464 * coretypes.h (pod_mode): New type.
11465 (scalar_int_mode_pod): New typedef.
11466 * machmode.h (pod_mode): New class.
11467 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11468 * genmodes.c (emit_mode_int_n): Update accordingly.
11469 * lower-subreg.h (target_lower_subreg): Change type to
11470 scalar_int_mode_pod.
11471 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11472 scalar_int_mode_pod.
11474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11475 Alan Hayward <alan.hayward@arm.com>
11476 David Sherwood <david.sherwood@arm.com>
11478 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11479 machine_mode to scalar_int_mode.
11480 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11481 (rs6000_option_override_internal): Remove cast to int.
11482 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11483 machine_mode to scalar_int_mode.
11484 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11485 (rs6000_option_override_internal): Remove cast to int.
11486 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11487 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11489 * config/s390/s390.c (s390_expand_builtin): Likewise.
11490 * coretypes.h (scalar_int_mode): New type.
11491 (opt_scalar_int_mode): New typedef.
11492 * machmode.h (scalar_int_mode): New class.
11493 (scalar_int_mode::includes_p): New function.
11494 (byte_mode): Change type to scalar_int_mode.
11495 (word_mode): Likewise.
11496 (ptr_mode): Likewise.
11497 * emit-rtl.c (byte_mode): Likewise.
11498 (word_mode): Likewise.
11499 (ptr_mode): Likewise.
11500 (init_derived_machine_modes): Update accordingly.
11501 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11502 and MODE_PARTIAL_INT.
11503 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11504 opt_scalar_int_mode.
11506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11507 Alan Hayward <alan.hayward@arm.com>
11508 David Sherwood <david.sherwood@arm.com>
11510 * target.def (libgcc_floating_mode_supported_p): Take a
11512 * doc/tm.texi: Regenerate.
11513 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11515 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11516 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11520 Alan Hayward <alan.hayward@arm.com>
11521 David Sherwood <david.sherwood@arm.com>
11523 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11524 * doc/tm.texi: Regenerate.
11525 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11526 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11527 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11528 * targhooks.h (default_floatn_mode): Likewise.
11529 * targhooks.c (default_floatn_mode): Likewise.
11530 * tree.c (build_common_tree_nodes): Update accordingly.
11532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11533 Alan Hayward <alan.hayward@arm.com>
11534 David Sherwood <david.sherwood@arm.com>
11536 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11537 (mode_iterator::iterate_p): Likewise.
11538 (mode_iterator::get_wider): Likewise.
11539 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11541 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11542 Alan Hayward <alan.hayward@arm.com>
11543 David Sherwood <david.sherwood@arm.com>
11545 * coretypes.h (opt_scalar_float_mode): New typedef.
11546 * machmode.h (float_mode_for_size): New function.
11547 * emit-rtl.c (double_mode): Delete.
11548 (init_emit_once): Use float_mode_for_size.
11549 * stor-layout.c (layout_type): Likewise.
11550 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11552 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11553 Alan Hayward <alan.hayward@arm.com>
11554 David Sherwood <david.sherwood@arm.com>
11556 * output.h (assemble_real): Take a scalar_float_mode.
11557 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11558 * config/arm/arm.md (consttable_4): Likewise.
11559 (consttable_8): Likewise.
11560 (consttable_16): Likewise.
11561 * config/mips/mips.md (consttable_float): Likewise.
11562 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11563 * varasm.c (assemble_real): Take a scalar_float_mode.
11564 (output_constant_pool_2): Update accordingly.
11565 (output_constant): Likewise.
11567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11568 Alan Hayward <alan.hayward@arm.com>
11569 David Sherwood <david.sherwood@arm.com>
11571 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11572 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11573 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11574 (native_encode_real): Likewise.
11575 (native_interpret_real): Likewise.
11576 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11577 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11580 Alan Hayward <alan.hayward@arm.com>
11581 David Sherwood <david.sherwood@arm.com>
11583 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11584 <scalar_float_mode>. Simplify.
11585 (gen_extend_conv_libfunc): Likewise.
11587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11588 Alan Hayward <alan.hayward@arm.com>
11589 David Sherwood <david.sherwood@arm.com>
11591 * coretypes.h (scalar_float_mode): New type.
11592 * machmode.h (mode_traits::from_int): Use machine_mode if
11593 USE_ENUM_MODES is defined.
11594 (is_a): New function.
11596 (dyn_cast): Likewise.
11597 (scalar_float_mode): New class.
11598 (scalar_float_mode::includes_p): New function.
11599 (is_float_mode): Likewise.
11600 * gdbhooks.py (MachineModePrinter): New class.
11601 (build_pretty_printer): Use it for scalar_float_mode.
11602 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11603 (format_helper::format_helper): Turn into a template.
11604 * genmodes.c (get_mode_class): New function.
11605 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11606 or machine_mode if none.
11607 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11608 as_a <scalar_float_mode>.
11609 * dwarf2out.c (mem_loc_descriptor): Likewise.
11610 (insert_float): Likewise.
11611 (add_const_value_attribute): Likewise.
11612 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11613 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11614 (expand_unop): Update accordingly.
11615 (expand_abs_nojump): Likewise.
11616 (expand_copysign_absneg): Take a scalar_float_mode.
11617 (expand_copysign_bit): Likewise.
11618 (expand_copysign): Update accordingly.
11620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11621 Alan Hayward <alan.hayward@arm.com>
11622 David Sherwood <david.sherwood@arm.com>
11624 * coretypes.h (opt_mode): New class.
11625 * machmode.h (opt_mode): Likewise.
11626 (opt_mode::else_void): New function.
11627 (opt_mode::require): Likewise.
11628 (opt_mode::exists): Likewise.
11629 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11630 (GET_MODE_2XWIDER_MODE): Likewise.
11631 (mode_iterator::get_wider): Update accordingly.
11632 (mode_iterator::get_2xwider): Likewise.
11633 (mode_iterator::get_known_wider): Likewise, turning into a template.
11634 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11635 forcing a wider mode to exist.
11636 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11637 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11638 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11639 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11640 * lower-subreg.c (init_lower_subreg): Likewise.
11641 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11642 on the final iteration.
11643 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11644 a wider mode exists before asking for a move pattern.
11645 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11646 forcing a wider mode to exist.
11647 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11648 returning false if no such mode exists.
11649 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11650 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11651 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11652 Avoid checking for a MODE_INT if we already know the mode is not a
11654 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11655 forcing a wider mode to exist.
11656 (expmed_mult_highpart_optab): Likewise.
11657 (expmed_mult_highpart): Likewise.
11658 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11660 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11661 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11662 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11663 * internal-fn.c (expand_mul_overflow): Update use of
11664 GET_MODE_2XWIDER_MODE.
11665 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11666 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11667 GET_MODE_WIDER_MODE.
11668 (convert_plusminus_to_widen): Likewise.
11669 * tree-switch-conversion.c (array_value_type): Likewise.
11670 * var-tracking.c (emit_note_insn_var_location): Likewise.
11671 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11672 Return false inside rather than outside the loop if no wider mode
11674 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11675 and GET_MODE_2XWIDER_MODE
11676 (can_compare_p): Use else_void.
11677 * gdbhooks.py (OptMachineModePrinter): New class.
11678 (build_pretty_printer): Use it for opt_mode.
11680 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11681 Alan Hayward <alan.hayward@arm.com>
11682 David Sherwood <david.sherwood@arm.com>
11684 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11685 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11688 Alan Hayward <alan.hayward@arm.com>
11689 David Sherwood <david.sherwood@arm.com>
11691 * machmode.h (mode_traits): New structure.
11692 (get_narrowest_mode): New function.
11693 (mode_iterator::start): Likewise.
11694 (mode_iterator::iterate_p): Likewise.
11695 (mode_iterator::get_wider): Likewise.
11696 (mode_iterator::get_known_wider): Likewise.
11697 (mode_iterator::get_2xwider): Likewise.
11698 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11699 (FOR_EACH_MODE): Likewise.
11700 (FOR_EACH_MODE_FROM): Likewise.
11701 (FOR_EACH_MODE_UNTIL): Likewise.
11702 (FOR_EACH_WIDER_MODE): Likewise.
11703 (FOR_EACH_2XWIDER_MODE): Likewise.
11704 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11705 * combine.c (simplify_comparison): Likewise
11706 * config/i386/i386.c (type_natural_mode): Likewise.
11707 * cse.c (cse_insn): Likewise.
11708 * dse.c (find_shift_sequence): Likewise.
11709 * emit-rtl.c (init_derived_machine_modes): Likewise.
11710 (init_emit_once): Likewise.
11711 * explow.c (hard_function_value): Likewise.
11712 * expmed.c (extract_fixed_bit_field_1): Likewise.
11713 (extract_bit_field_1): Likewise.
11714 (expand_divmod): Likewise.
11715 (emit_store_flag_1): Likewise.
11716 * expr.c (init_expr_target): Likewise.
11717 (convert_move): Likewise.
11718 (alignment_for_piecewise_move): Likewise.
11719 (widest_int_mode_for_size): Likewise.
11720 (emit_block_move_via_movmem): Likewise.
11721 (copy_blkmode_to_reg): Likewise.
11722 (set_storage_via_setmem): Likewise.
11723 (compress_float_constant): Likewise.
11724 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11725 * optabs-query.c (get_best_extraction_insn): Likewise.
11726 * optabs.c (expand_binop): Likewise.
11727 (expand_twoval_unop): Likewise.
11728 (expand_twoval_binop): Likewise.
11729 (widen_leading): Likewise.
11730 (widen_bswap): Likewise.
11731 (expand_parity): Likewise.
11732 (expand_unop): Likewise.
11733 (prepare_cmp_insn): Likewise.
11734 (prepare_float_lib_cmp): Likewise.
11735 (expand_float): Likewise.
11736 (expand_fix): Likewise.
11737 (expand_sfix_optab): Likewise.
11738 * postreload.c (move2add_use_add2_insn): Likewise.
11739 * reg-stack.c (reg_to_stack): Likewise.
11740 * reginfo.c (choose_hard_reg_mode): Likewise.
11741 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11742 * stor-layout.c (mode_for_size): Likewise.
11743 (smallest_mode_for_size): Likewise.
11744 (mode_for_vector): Likewise.
11745 (finish_bitfield_representative): Likewise.
11746 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11747 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11748 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11749 * var-tracking.c (prepare_call_arguments): Likewise.
11751 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11752 Alan Hayward <alan.hayward@arm.com>
11753 David Sherwood <david.sherwood@arm.com>
11755 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11756 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11757 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11758 * machmode.h (mode_size): Move earlier in file.
11759 (mode_precision): Likewise.
11760 (mode_inner): Likewise.
11761 (mode_nunits): Likewise.
11762 (mode_unit_size): Likewise.
11763 (unit_unit_precision): Likewise.
11764 (mode_wider): Likewise.
11765 (mode_2xwider): Likewise.
11766 (machine_mode): New class.
11767 (mode_to_bytes): New function.
11768 (mode_to_bits): Likewise.
11769 (mode_to_precision): Likewise.
11770 (mode_to_inner): Likewise.
11771 (mode_to_unit_size): Likewise.
11772 (mode_to_unit_precision): Likewise.
11773 (mode_to_nunits): Likewise.
11774 (GET_MODE_SIZE): Use mode_to_bytes.
11775 (GET_MODE_BITSIZE): Use mode_to_bits.
11776 (GET_MODE_PRECISION): Use mode_to_precision.
11777 (GET_MODE_INNER): Use mode_to_inner.
11778 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11779 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11780 (GET_MODE_NUNITS): Use mode_to_nunits.
11781 * system.h (ALWAYS_INLINE): New macro.
11782 * config/powerpcspe/powerpcspe-c.c
11783 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11784 int for arg1_mode and arg2_mode.
11786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11787 Alan Hayward <alan.hayward@arm.com>
11788 David Sherwood <david.sherwood@arm.com>
11790 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11791 Prefix mode names with E_ in case statements.
11792 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11793 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11794 (aarch64_split_simd_move): Likewise.
11795 (aarch64_gen_storewb_pair): Likewise.
11796 (aarch64_gen_loadwb_pair): Likewise.
11797 (aarch64_gen_store_pair): Likewise.
11798 (aarch64_gen_load_pair): Likewise.
11799 (aarch64_get_condition_code_1): Likewise.
11800 (aarch64_constant_pool_reload_icode): Likewise.
11801 (get_rsqrte_type): Likewise.
11802 (get_rsqrts_type): Likewise.
11803 (get_recpe_type): Likewise.
11804 (get_recps_type): Likewise.
11805 (aarch64_gimplify_va_arg_expr): Likewise.
11806 (aarch64_simd_container_mode): Likewise.
11807 (aarch64_emit_load_exclusive): Likewise.
11808 (aarch64_emit_store_exclusive): Likewise.
11809 (aarch64_expand_compare_and_swap): Likewise.
11810 (aarch64_gen_atomic_cas): Likewise.
11811 (aarch64_emit_bic): Likewise.
11812 (aarch64_emit_atomic_swap): Likewise.
11813 (aarch64_emit_atomic_load_op): Likewise.
11814 (aarch64_evpc_trn): Likewise.
11815 (aarch64_evpc_uzp): Likewise.
11816 (aarch64_evpc_zip): Likewise.
11817 (aarch64_evpc_ext): Likewise.
11818 (aarch64_evpc_rev): Likewise.
11819 (aarch64_evpc_dup): Likewise.
11820 (aarch64_gen_ccmp_first): Likewise.
11821 (aarch64_gen_ccmp_next): Likewise.
11822 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11823 (alpha_emit_xfloating_libcall): Likewise.
11824 (emit_insxl): Likewise.
11825 (alpha_arg_type): Likewise.
11826 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11827 (arc_preferred_simd_mode): Likewise.
11828 (arc_secondary_reload): Likewise.
11829 (get_arc_condition_code): Likewise.
11830 (arc_print_operand): Likewise.
11831 (arc_legitimate_constant_p): Likewise.
11832 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11833 * config/arc/arc.md (casesi_load): Likewise.
11834 (casesi_compact_jump): Likewise.
11835 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11836 (cc_use_register): Likewise.
11837 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11838 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11839 (arm_init_iwmmxt_builtins): Likewise.
11840 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11841 (neon_expand_vector_init): Likewise.
11842 (arm_attr_length_move_neon): Likewise.
11843 (maybe_get_arm_condition_code): Likewise.
11844 (arm_emit_vector_const): Likewise.
11845 (arm_preferred_simd_mode): Likewise.
11846 (arm_output_iwmmxt_tinsr): Likewise.
11847 (thumb1_output_casesi): Likewise.
11848 (thumb2_output_casesi): Likewise.
11849 (arm_emit_load_exclusive): Likewise.
11850 (arm_emit_store_exclusive): Likewise.
11851 (arm_expand_compare_and_swap): Likewise.
11852 (arm_evpc_neon_vuzp): Likewise.
11853 (arm_evpc_neon_vzip): Likewise.
11854 (arm_evpc_neon_vrev): Likewise.
11855 (arm_evpc_neon_vtrn): Likewise.
11856 (arm_evpc_neon_vext): Likewise.
11857 (arm_validize_comparison): Likewise.
11858 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11859 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11860 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11861 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11862 (c6x_preferred_simd_mode): Likewise.
11863 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11864 (epiphany_rtx_costs): Likewise.
11865 * config/epiphany/predicates.md (proper_comparison_operator):
11867 * config/frv/frv.c (condexec_memory_operand): Likewise.
11868 (frv_emit_move): Likewise.
11869 (output_move_single): Likewise.
11870 (output_condmove_single): Likewise.
11871 (frv_hard_regno_mode_ok): Likewise.
11872 (frv_matching_accg_mode): Likewise.
11873 * config/h8300/h8300.c (split_adds_subs): Likewise.
11874 (h8300_rtx_costs): Likewise.
11875 (h8300_print_operand): Likewise.
11876 (compute_mov_length): Likewise.
11877 (output_logical_op): Likewise.
11878 (compute_logical_op_length): Likewise.
11879 (compute_logical_op_cc): Likewise.
11880 (h8300_shift_needs_scratch_p): Likewise.
11881 (output_a_shift): Likewise.
11882 (compute_a_shift_length): Likewise.
11883 (compute_a_shift_cc): Likewise.
11884 (expand_a_rotate): Likewise.
11885 (output_a_rotate): Likewise.
11886 * config/i386/i386.c (classify_argument): Likewise.
11887 (function_arg_advance_32): Likewise.
11888 (function_arg_32): Likewise.
11889 (function_arg_64): Likewise.
11890 (function_value_64): Likewise.
11891 (ix86_gimplify_va_arg): Likewise.
11892 (ix86_legitimate_constant_p): Likewise.
11893 (put_condition_code): Likewise.
11894 (split_double_mode): Likewise.
11895 (ix86_avx256_split_vector_move_misalign): Likewise.
11896 (ix86_expand_vector_logical_operator): Likewise.
11897 (ix86_split_idivmod): Likewise.
11898 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11899 (ix86_build_const_vector): Likewise.
11900 (ix86_build_signbit_mask): Likewise.
11901 (ix86_match_ccmode): Likewise.
11902 (ix86_cc_modes_compatible): Likewise.
11903 (ix86_expand_branch): Likewise.
11904 (ix86_expand_sse_cmp): Likewise.
11905 (ix86_expand_sse_movcc): Likewise.
11906 (ix86_expand_int_sse_cmp): Likewise.
11907 (ix86_expand_vec_perm_vpermi2): Likewise.
11908 (ix86_expand_vec_perm): Likewise.
11909 (ix86_expand_sse_unpack): Likewise.
11910 (ix86_expand_int_addcc): Likewise.
11911 (ix86_split_to_parts): Likewise.
11912 (ix86_vectorize_builtin_gather): Likewise.
11913 (ix86_vectorize_builtin_scatter): Likewise.
11914 (avx_vpermilp_parallel): Likewise.
11915 (inline_memory_move_cost): Likewise.
11916 (ix86_tieable_integer_mode_p): Likewise.
11917 (x86_maybe_negate_const_int): Likewise.
11918 (ix86_expand_vector_init_duplicate): Likewise.
11919 (ix86_expand_vector_init_one_nonzero): Likewise.
11920 (ix86_expand_vector_init_one_var): Likewise.
11921 (ix86_expand_vector_init_concat): Likewise.
11922 (ix86_expand_vector_init_interleave): Likewise.
11923 (ix86_expand_vector_init_general): Likewise.
11924 (ix86_expand_vector_set): Likewise.
11925 (ix86_expand_vector_extract): Likewise.
11926 (emit_reduc_half): Likewise.
11927 (ix86_emit_i387_round): Likewise.
11928 (ix86_mangle_type): Likewise.
11929 (ix86_expand_round_sse4): Likewise.
11930 (expand_vec_perm_blend): Likewise.
11931 (canonicalize_vector_int_perm): Likewise.
11932 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11933 (expand_vec_perm_1): Likewise.
11934 (expand_vec_perm_interleave3): Likewise.
11935 (expand_vec_perm_even_odd_pack): Likewise.
11936 (expand_vec_perm_even_odd_1): Likewise.
11937 (expand_vec_perm_broadcast_1): Likewise.
11938 (ix86_vectorize_vec_perm_const_ok): Likewise.
11939 (ix86_expand_vecop_qihi): Likewise.
11940 (ix86_expand_mul_widen_hilo): Likewise.
11941 (ix86_expand_sse2_abs): Likewise.
11942 (ix86_expand_pextr): Likewise.
11943 (ix86_expand_pinsr): Likewise.
11944 (ix86_preferred_simd_mode): Likewise.
11945 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11946 * config/i386/sse.md (*andnot<mode>3): Likewise.
11947 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11948 (*<code><mode>3): Likewise.
11949 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11950 (ia64_expand_atomic_op): Likewise.
11951 (ia64_arg_type): Likewise.
11952 (ia64_mode_to_int): Likewise.
11953 (ia64_scalar_mode_supported_p): Likewise.
11954 (ia64_vector_mode_supported_p): Likewise.
11955 (expand_vec_perm_broadcast): Likewise.
11956 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11957 (iq2000_function_arg_advance): Likewise.
11958 (iq2000_function_arg): Likewise.
11959 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11960 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11961 (m68k_libcall_value): Likewise.
11962 (m68k_function_value): Likewise.
11963 (sched_attr_op_type): Likewise.
11964 * config/mcore/mcore.c (mcore_output_move): Likewise.
11965 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11967 (microblaze_function_arg): Likewise.
11968 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11969 (mips_print_operand): Likewise.
11970 (mips_mode_ok_for_mov_fmt_p): Likewise.
11971 (mips_vector_mode_supported_p): Likewise.
11972 (mips_preferred_simd_mode): Likewise.
11973 (mips_expand_vpc_loongson_even_odd): Likewise.
11974 (mips_expand_vec_unpack): Likewise.
11975 (mips_expand_vi_broadcast): Likewise.
11976 (mips_expand_vector_init): Likewise.
11977 (mips_expand_vec_reduc): Likewise.
11978 (mips_expand_msa_cmp): Likewise.
11979 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11980 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11981 (cc_flags_for_mode): Likewise.
11982 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11983 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11984 (nds32_output_casesi_pc_relative): Likewise.
11985 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11986 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11987 (nvptx_gen_unpack): Likewise.
11988 (nvptx_gen_pack): Likewise.
11989 (nvptx_gen_shuffle): Likewise.
11990 (nvptx_gen_wcast): Likewise.
11991 (nvptx_preferred_simd_mode): Likewise.
11992 * config/pa/pa.c (pa_secondary_reload): Likewise.
11993 * config/pa/predicates.md (base14_operand): Likewise.
11994 * config/powerpcspe/powerpcspe-c.c
11995 (altivec_resolve_overloaded_builtin): Likewise.
11996 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11998 (rs6000_preferred_simd_mode): Likewise.
11999 (output_vec_const_move): Likewise.
12000 (rs6000_expand_vector_extract): Likewise.
12001 (rs6000_split_vec_extract_var): Likewise.
12002 (reg_offset_addressing_ok_p): Likewise.
12003 (rs6000_legitimate_offset_address_p): Likewise.
12004 (rs6000_legitimize_address): Likewise.
12005 (rs6000_emit_set_const): Likewise.
12006 (rs6000_const_vec): Likewise.
12007 (rs6000_emit_move): Likewise.
12008 (spe_build_register_parallel): Likewise.
12009 (rs6000_darwin64_record_arg_recurse): Likewise.
12010 (swap_selector_for_mode): Likewise.
12011 (spe_init_builtins): Likewise.
12012 (paired_init_builtins): Likewise.
12013 (altivec_init_builtins): Likewise.
12014 (do_load_for_compare): Likewise.
12015 (rs6000_generate_compare): Likewise.
12016 (rs6000_expand_float128_convert): Likewise.
12017 (emit_load_locked): Likewise.
12018 (emit_store_conditional): Likewise.
12019 (rs6000_output_function_epilogue): Likewise.
12020 (rs6000_handle_altivec_attribute): Likewise.
12021 (rs6000_function_value): Likewise.
12022 (emit_fusion_gpr_load): Likewise.
12023 (emit_fusion_p9_load): Likewise.
12024 (emit_fusion_p9_store): Likewise.
12025 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12026 (fusion_gpr_mem_load): Likewise.
12027 (fusion_addis_mem_combo_load): Likewise.
12028 (fusion_addis_mem_combo_store): Likewise.
12029 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12030 (fusion_gpr_mem_load): Likewise.
12031 (fusion_addis_mem_combo_load): Likewise.
12032 (fusion_addis_mem_combo_store): Likewise.
12033 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12035 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12036 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12037 (rs6000_preferred_simd_mode): Likewise.
12038 (output_vec_const_move): Likewise.
12039 (rs6000_expand_vector_extract): Likewise.
12040 (rs6000_split_vec_extract_var): Likewise.
12041 (reg_offset_addressing_ok_p): Likewise.
12042 (rs6000_legitimate_offset_address_p): Likewise.
12043 (rs6000_legitimize_address): Likewise.
12044 (rs6000_emit_set_const): Likewise.
12045 (rs6000_const_vec): Likewise.
12046 (rs6000_emit_move): Likewise.
12047 (rs6000_darwin64_record_arg_recurse): Likewise.
12048 (swap_selector_for_mode): Likewise.
12049 (paired_init_builtins): Likewise.
12050 (altivec_init_builtins): Likewise.
12051 (rs6000_expand_float128_convert): Likewise.
12052 (emit_load_locked): Likewise.
12053 (emit_store_conditional): Likewise.
12054 (rs6000_output_function_epilogue): Likewise.
12055 (rs6000_handle_altivec_attribute): Likewise.
12056 (rs6000_function_value): Likewise.
12057 (emit_fusion_gpr_load): Likewise.
12058 (emit_fusion_p9_load): Likewise.
12059 (emit_fusion_p9_store): Likewise.
12060 * config/rx/rx.c (rx_gen_move_template): Likewise.
12061 (flags_from_mode): Likewise.
12062 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12063 (s390_slb_comparison): Likewise.
12064 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12065 (s390_vector_mode_supported_p): Likewise.
12066 (s390_cc_modes_compatible): Likewise.
12067 (s390_match_ccmode_set): Likewise.
12068 (s390_canonicalize_comparison): Likewise.
12069 (s390_emit_compare_and_swap): Likewise.
12070 (s390_branch_condition_mask): Likewise.
12071 (s390_rtx_costs): Likewise.
12072 (s390_secondary_reload): Likewise.
12073 (__SECONDARY_RELOAD_CASE): Likewise.
12074 (s390_expand_cs): Likewise.
12075 (s390_preferred_simd_mode): Likewise.
12076 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12077 * config/sh/sh.c (sh_print_operand): Likewise.
12078 (dump_table): Likewise.
12079 (sh_secondary_reload): Likewise.
12080 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12081 * config/sh/sh.md (casesi_worker_1): Likewise.
12082 (casesi_worker_2): Likewise.
12083 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12084 (fcc_comparison_operator): Likewise.
12085 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12086 (emit_soft_tfmode_cvt): Likewise.
12087 (sparc_preferred_simd_mode): Likewise.
12088 (output_cbranch): Likewise.
12089 (sparc_print_operand): Likewise.
12090 (sparc_expand_vec_perm_bmask): Likewise.
12091 (vector_init_bshuffle): Likewise.
12092 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12093 (spu_vector_mode_supported_p): Likewise.
12094 (spu_expand_insv): Likewise.
12095 (spu_emit_branch_or_set): Likewise.
12096 (spu_handle_vector_attribute): Likewise.
12097 (spu_builtin_splats): Likewise.
12098 (spu_builtin_extract): Likewise.
12099 (spu_builtin_promote): Likewise.
12100 (spu_expand_sign_extend): Likewise.
12101 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12102 (tilegx_simd_int): Likewise.
12103 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12104 (tilepro_simd_int): Likewise.
12105 * config/v850/v850.c (const_double_split): Likewise.
12106 (v850_print_operand): Likewise.
12107 (ep_memory_offset): Likewise.
12108 * config/vax/vax.c (vax_rtx_costs): Likewise.
12109 (vax_output_int_move): Likewise.
12110 (vax_output_int_add): Likewise.
12111 (vax_output_int_subtract): Likewise.
12112 * config/visium/predicates.md (visium_branch_operator): Likewise.
12113 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12114 (visium_print_operand_address): Likewise.
12115 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12116 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12117 (xtensa_expand_conditional_branch): Likewise.
12118 (xtensa_copy_incoming_a7): Likewise.
12119 (xtensa_output_literal): Likewise.
12120 * dfp.c (decimal_real_maxval): Likewise.
12121 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12124 Alan Hayward <alan.hayward@arm.com>
12125 David Sherwood <david.sherwood@arm.com>
12127 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12128 (mode_nunits_inline): Likewise.
12129 (mode_inner_inline): Likewise.
12130 (mode_unit_size_inline): Likewise.
12131 (mode_unit_precision_inline): Likewise.
12132 (emit_insn_modes_h): Likewise. Also emit a #define of the
12134 (emit_mode_wider): Add an E_ prefix to mode names.
12135 (emit_mode_complex): Likewise.
12136 (emit_mode_inner): Likewise.
12137 (emit_mode_adjustments): Likewise.
12138 (emit_mode_int_n): Likewise.
12139 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12140 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12141 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12142 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12143 (CRC32_BUILTIN, ENTRY): Likewise.
12144 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12145 (aarch64_pop_regs): Likewise.
12146 (aarch64_process_components): Likewise.
12147 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12148 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12149 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12150 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12151 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12152 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12154 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12155 (pcmpistr): Likewise.
12156 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12157 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12158 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12160 * config/rl78/rl78.c (mduc_regs): Likewise.
12161 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12162 (htm_expand_builtin): Likewise.
12163 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12164 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12165 * config/xtensa/xtensa.c (print_operand): Likewise.
12166 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12167 (NUM_MODE_VECTOR_INT): Likewise.
12168 * genoutput.c (null_operand): Likewise.
12169 (output_operand_data): Likewise.
12170 * genrecog.c (print_parameter_value): Likewise.
12171 * lra.c (debug_operand_data): Likewise.
12173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12175 * dwarf2out.c (scompare_loc_descriptor_wide)
12176 (scompare_loc_descriptor_narrow): New functions, split out from...
12177 (scompare_loc_descriptor): ...here.
12178 * expmed.c (emit_store_flag_int): New function, split out from...
12179 (emit_store_flag): ...here.
12181 2017-08-30 Richard Biener <rguenther@suse.de>
12183 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12184 (dwarf2out_early_finish): Move setting of AT_pubnames from
12185 early debug output to early finish.
12187 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12189 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12190 and -mdata-region to the assembler.
12192 2017-08-30 Richard Biener <rguenther@suse.de>
12194 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12196 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12197 (dwarf2out_early_global_decl): Only generate a DIE for the
12198 abstract origin if it doesn't already exist or is a declaration DIE.
12199 (resolve_addr): Do not add the linkage name twice when
12200 generating a stub DIE for the DW_TAG_GNU_call_site target.
12202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12204 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12205 Use machine_mode rather than int for arg1_mode.
12207 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12210 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12211 that the second argument of the built-in functions to unpack
12212 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12213 switch statement instead a lot of if statements.
12214 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12215 Allow 64-bit values to be in Altivec registers as well as
12216 traditional floating point registers.
12217 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12219 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12221 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12222 MAX_REGS_PER_ADDRESS == 1.
12224 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12226 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12227 * config/i386/i386.c (ix86_option_override_internal): Simplify
12228 setting of opts->x_flag_entry.
12230 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12231 Jakub Jelinek <jakub@redhat.com>
12232 Richard Biener <rguenther@suse.de>
12234 PR tree-optimization/81503
12235 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12236 folded constant fits in the target type; reorder tests for clarity.
12238 2017-08-29 Martin Liska <mliska@suse.cz>
12240 * passes.def: Include pass_lower_switch.
12241 * stmt.c (dump_case_nodes): Remove and move to
12242 tree-switch-conversion.
12243 (case_values_threshold): Likewise.
12244 (expand_switch_as_decision_tree_p): Likewise.
12245 (emit_case_decision_tree): Likewise.
12246 (expand_case): Likewise.
12247 (balance_case_nodes): Likewise.
12248 (node_has_low_bound): Likewise.
12249 (node_has_high_bound): Likewise.
12250 (node_is_bounded): Likewise.
12251 (emit_case_nodes): Likewise.
12252 (struct simple_case_node): New struct.
12253 (add_case_node): Remove.
12254 (emit_case_dispatch_table): Use vector instead of case_list.
12255 (reset_out_edges_aux): Remove.
12256 (compute_cases_per_edge): Likewise.
12257 (expand_case): Build list of simple_case_node.
12258 (expand_sjlj_dispatch_table): Use it.
12259 * tree-switch-conversion.c (struct case_node): Moved from
12260 stmt.c and adjusted.
12261 (emit_case_nodes): Likewise.
12262 (node_has_low_bound): Likewise.
12263 (node_has_high_bound): Likewise.
12264 (node_is_bounded): Likewise.
12265 (case_values_threshold): Likewise.
12266 (reset_out_edges_aux): Likewise.
12267 (compute_cases_per_edge): Likewise.
12268 (add_case_node): Likewise.
12269 (dump_case_nodes): Likewise.
12270 (balance_case_nodes): Likewise.
12271 (expand_switch_as_decision_tree_p): Likewise.
12272 (emit_jump): Likewise.
12273 (emit_case_decision_tree): Likewise.
12274 (try_switch_expansion): Likewise.
12275 (do_jump_if_equal): Likewise.
12276 (emit_cmp_and_jump_insns): Likewise.
12277 (fix_phi_operands_for_edge): New function.
12278 (record_phi_operand_mapping): Likewise.
12279 (class pass_lower_switch): New pass.
12280 (pass_lower_switch::execute): New function.
12281 (make_pass_lower_switch): Likewise.
12282 (conditional_probability):
12283 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12284 * tree-pass.h: Add make_pass_lower_switch.
12286 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12289 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12292 2017-08-29 Richard Biener <rguenther@suse.de>
12294 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12295 we do not add a DW_AT_inline attribute twice.
12296 (gen_subprogram_die): Remove code setting DW_AT_inline on
12297 DECL_ABSTRACT_P nodes.
12299 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12301 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12302 calls to internal functions.
12303 (gimplify_modify_expr): Likewise.
12304 * tree-call-cdce.c (use_internal_fn): Likewise.
12305 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12306 (convert_to_divmod): Set the nothrow flag.
12307 * tree-if-conv.c (predicate_mem_writes): Likewise.
12308 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12309 (vectorizable_call): Likewise.
12310 (vectorizable_store): Likewise.
12311 (vectorizable_load): Likewise.
12312 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12313 (vect_recog_mask_conversion_pattern): Likewise.
12315 2017-08-29 Martin Liska <mliska@suse.cz>
12318 * gcc.c (driver_handle_option): Add new argument.
12319 * opts-common.c (handle_option): Pass
12320 target_option_override_hook.
12321 * opts-global.c (lang_handle_option): Add new option.
12322 (set_default_handlers): Add new argument.
12323 (decode_options): Likewise.
12324 * opts.c (target_handle_option): Likewise.
12325 (common_handle_option): Call target_option_override_hook.
12326 * opts.h (struct cl_option_handler_func): Add hook for
12327 target option override.
12328 (struct cl_option_handlers): Likewise.
12329 (set_default_handlers): Add new argument.
12330 (decode_options): Likewise.
12331 (common_handle_option): Likewise.
12332 (target_handle_option): Likewise.
12333 * toplev.c (toplev::main): Pass targetm.target_option.override
12336 2017-08-29 Richard Biener <rguenther@suse.de>
12337 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12339 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12340 life to the active subtree.
12342 2017-08-28 Jeff Law <law@redhat.com>
12344 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12345 derive_equivalences.
12346 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12347 Code moved into....
12348 (edge_info::derive_equivalences): New private member function
12350 * tree-ssa-dom.c (class edge_info): Changed from a struct
12351 to a class. Add ctor/dtor, methods and data members.
12352 (edge_info::edge_info): Renamed from allocate_edge_info.
12353 Initialize additional members.
12354 (edge_info::~edge_info): New.
12355 (free_dom_edge_info): Delete the edge info.
12356 (record_edge_info): Use new class & associated member functions.
12357 Tighten forms for testing for edge equivalences.
12358 (record_temporary_equivalences): Iterate over the simple
12359 equivalences rather than assuming there's only one per edge.
12360 (cprop_into_successor_phis): Iterate over the simple
12361 equivalences rather than assuming there's only one per edge.
12362 (optimize_stmt): Use operand_equal_p rather than pointer
12363 equality for mini-DSE code.
12365 2017-08-28 Nathan Sidwell <nathan@acm.org>
12367 * gcc.c (execute): Fold SIGPIPE handling into switch
12368 statement. Adjust internal error message.
12370 2017-08-28 Richard Biener <rguenther@suse.de>
12373 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12374 Do nothing for removed DIEs.
12376 2017-08-28 Richard Biener <rguenther@suse.de>
12378 PR tree-optimization/81977
12379 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12382 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12385 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12386 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12387 using targetm.gen_mem_thread_fence.
12389 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12392 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12393 predicate to register_operand. Reorder operands.
12394 (*btr<mode>): Ditto.
12395 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12396 (*btr<mode>_mask): Ditto.
12398 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12400 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12401 * config/rs6000/xmmintrin.h: New file.
12402 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12404 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12407 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12408 parameter and_insn and return it.
12409 (recombine_lvx_pattern): Insert a copy to ensure availability of
12410 the base register of the copied masking operation at the point of
12411 the instruction replacement.
12412 (recombine_stvx_pattern): Likewise.
12414 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12416 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12417 undocumented switches.
12418 (-mpower9-dform-vector): Likewise.
12419 (-mpower9-dform): Likewise.
12420 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12421 comments to delete references to -mpower9-dform* switches.
12422 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12423 Delete reference to -mpower9-dform* switches, test for
12424 -mpower9-vector instead.
12425 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12426 (OTHER_P9_VECTOR_MASKS): Likewise.
12427 (POWERPC_MASKS): Likewise.
12428 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12429 tests against -mpower9-dform* to -mpower9-vector. Delete code
12430 that checked for -mpower9-dform* consistancy with other options.
12431 Add test for -mpower9-misc to enable other power9 switches.
12432 (rs6000_init_hard_regno_mode_ok): Likewise.
12433 (rs6000_option_override_internal): Likewise.
12434 (rs6000_emit_prologue): Likewise.
12435 (rs6000_emit_epilogue): Likewise.
12436 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12437 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12438 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12440 (emit_fusion_p9_store): Likewise.
12441 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12442 resetting these macros if the assembler does not support ISA 3.0
12444 (TARGET_P9_DFORM_VECTOR): Likewise.
12445 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12446 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12448 2017-08-25 Alan Modra <amodra@gmail.com>
12450 PR rtl-optimization/81747
12451 * cse.c (cse_extended_basic_block): Don't attempt to record
12452 equivalences for degenerate conditional branches that branch
12453 to their fall-through.
12455 2017-08-24 Martin Sebor <msebor@redhat.com>
12457 PR middle-end/81908
12458 * gimple-fold.c (size_must_be_zero_p): New function.
12459 (gimple_fold_builtin_memory_op): Call it.
12461 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12463 * config/rs6000/mm_malloc.h: New file.
12465 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12467 PR tree-optimization/81913
12468 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12469 analysis when either IVs in condition can wrap.
12471 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12473 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12474 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12476 2017-08-24 Richard Biener <rguenther@suse.de>
12479 * targhooks.c (default_target_can_inline_p): Properly
12480 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12481 is present and always compare.
12482 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12483 infer -mfpmath=sse from TARGET_SSE_P.
12484 (ix86_can_inline_p): Properly use target_option_default_node when
12485 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12487 2017-08-24 Richard Biener <rguenther@suse.de>
12490 * dwarf2out.c (output_die): Handle flag_generate_offload like
12492 (output_comp_unit): Likewise.
12493 (gen_array_type_die): Likewise.
12494 (dwarf2out_early_finish): Likewise.
12495 (note_variable_value_in_expr): Likewise.
12496 (dwarf2out_finish): Likewise. Adjust assert.
12497 * cgraphunit.c (symbol_table::compile): Move setting of
12498 flag_generate_offload earlier ...
12499 (symbol_table::finalize_compilation_unit): ... here, before
12500 early debug finalization.
12502 2017-08-24 Richard Biener <rguenther@suse.de>
12504 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12505 and ipa-fnsummary.h.
12506 (ix86_can_inline_p): When ix86_fpmath flags do not match
12507 check whether the callee uses FP math at all.
12509 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12511 PR middle-end/81931
12512 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12513 instead of TYPE_PRECISION.
12515 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12516 Alan Hayward <alan.hayward@arm.com>
12517 David Sherwood <david.sherwood@arm.com>
12519 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12521 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12523 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12525 (ix86_frame::stack_realign_allocate): New field.
12526 (struct machine_frame_state): Modify comments.
12527 (machine_frame_state::sp_realigned_fp_end): New field.
12528 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12529 layout calculation.
12530 (sp_valid_at): Add assertion to assure no attempt to access invalid
12531 offset of a realigned stack.
12532 (fp_valid_at): Likewise.
12533 (choose_baseaddr): Modify comments.
12534 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12535 ix86_expand_prologue.
12536 (ix86_expand_prologue): Modify stack realignment and allocation.
12537 (ix86_expand_epilogue): Modify comments.
12538 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12539 avx2_runtime, avx512f, and avx512f_runtime.
12541 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12543 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12544 (mstackrealign): Do not init to -1.
12545 * config/i386/i386.c (ix86_option_override_internal):
12546 Check opts_set, not opts when setting default value of
12547 opts->x_ix86_force_align_arg_pointer.
12549 2017-08-23 Richard Biener <rguenther@suse.de>
12551 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12552 lang_hooks.decl_printable_name.
12553 * print-rtl-function.c (print_rtx_function): Likewise.
12554 * tree-pretty-print.c (dump_function_header): Likewise.
12556 2017-08-23 Richard Biener <rguenther@suse.de>
12559 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12560 -g0 at compile-time.
12562 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12564 PR middle-end/19706
12565 * doc/sourcebuild.texi (Other hardware attributes):
12568 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12570 PR middle-end/19706
12571 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12572 Fix single-use check.
12574 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12576 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12578 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12580 * doc/install.texi: Modify to add more details on running selected
12583 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12585 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12586 is combined with -mabi=ms.
12587 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12590 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12592 PR tree-optimization/81488
12593 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12594 and cached_basis fields.
12595 (MAX_SPREAD): New constant.
12596 (alloc_cand_and_find_basis): Initialize new fields.
12597 (clear_visited): New function.
12598 (create_phi_basis_1): Rename from create_phi_basis, set visited
12599 and cached_basis fields.
12600 (create_phi_basis): New wrapper function.
12601 (phi_add_costs_1): Rename from phi_add_costs, add spread
12602 parameter, set visited field, short-circuit when limits reached.
12603 (phi_add_costs): New wrapper function.
12604 (record_phi_increments_1): Rename from record_phi_increments, set
12606 (record_phi_increments): New wrapper function.
12607 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12608 (phi_incr_cost): New wrapper function.
12609 (all_phi_incrs_profitable_1): Rename from
12610 all_phi_incrs_profitable, set visited field.
12611 (all_phi_incrs_profitable): New wrapper function.
12613 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12614 Alan Hayward <alan.hayward@arm.com>
12615 David Sherwood <david.sherwood@arm.com>
12617 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12618 that takes the outer and inner modes.
12619 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12620 comparison as the canonical test for a paradoxical subreg.
12621 * combine.c (simplify_set): Use paradoxical_subreg_p.
12622 (make_extraction): Likewise.
12623 (force_to_mode): Likewise.
12624 (rtx_equal_for_field_assignment_p): Likewise.
12625 (gen_lowpart_for_combine): Likewise.
12626 (simplify_comparison): Likewise.
12627 * cse.c (equiv_constant): Likewise.
12628 * expmed.c (store_bit_field_1): Likewise.
12629 * final.c (alter_subreg): Likewise.
12630 * fwprop.c (propagate_rtx): Likewise.
12631 (forward_propagate_subreg): Likewise.
12632 * ira-conflicts.c (ira_build_conflicts): Likewise.
12633 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12634 * lra-constraints.c (curr_insn_transform): Likewise.
12635 (split_reg): Likewise.
12636 * lra-eliminations.c (move_plus_up): Likewise.
12637 (lra_eliminate_regs_1): Likewise.
12638 * recog.c (general_operand): Likewise.
12639 * ree.c (combine_reaching_defs): Likewise.
12640 * reload.c (push_reload): Likewise.
12641 (find_reloads): Likewise.
12642 * reload1.c (elimination_effects): Likewise.
12643 (compute_reload_subreg_offset): Likewise.
12644 (choose_reload_regs): Likewise.
12645 * rtlanal.c (subreg_lsb_1): Likewise.
12646 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12647 (simplify_subreg): Likewise.
12648 * var-tracking.c (track_loc_p): Likewise.
12649 * emit-rtl.c (byte_lowpart_offset): Likewise.
12650 (paradoxical_subreg_p): Delete out-of-line definition.
12652 2017-08-22 Jeff Law <law@redhat.com>
12654 PR tree-optimization/81741
12655 PR tree-optimization/71947
12656 * tree-ssa-dom.c: Include tree-inline.h.
12657 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12658 equivalences if one is more expensive to compute than the other.
12659 * tree-ssa-scopedtables.h (class const_or_copies): Make
12660 record_const_or_copy_raw method private.
12661 (class avail_exprs_stack): New method simplify_binary_operation.
12662 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12663 avail_exprs_stack::simplify_binary_operation as needed.
12664 (avail_exprs_stack::simplify_binary_operation): New function.
12666 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12668 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12669 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12670 (DOT_SYMBOLS): Likewise.
12671 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12672 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12673 (RS6000_ABI_NAME): Likewise.
12674 (TARGET_CMODEL): Likewise.
12675 (TOC_SECTION_ASM_OP): Likewise.
12676 (SET_CMODEL): New macro.
12677 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12679 2017-08-22 Richard Biener <rguenther@suse.de>
12681 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12682 to allow for free-lang-data replacements similar to verify_type_variant.
12684 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12685 Alan Hayward <alan.hayward@arm.com>
12686 David Sherwood <david.sherwood@arm.com>
12688 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12689 VOIDmode for the LABEL_REF.
12691 2017-08-22 Richard Biener <rguenther@suse.de>
12693 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12695 2017-08-22 Richard Biener <rguenther@suse.de>
12697 * common.opt (feliminate-dwarf2-dups): Ignore.
12698 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12699 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12700 same_die_p_wrap, compute_section_prefix,
12701 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12702 (comdat_symbol_id, comdat_symbol_number): Likewise.
12703 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12705 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12706 (output_die): Mark unreachable path unreachable.
12707 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12708 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12709 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12710 (dwarf2out_early_finish): Likewise.
12712 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12714 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12716 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12719 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12720 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12721 (avr_attribute_table) <io, io_low, address>: Initialize
12722 .decl_required with true.
12724 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12726 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12727 undocumented debugging options.
12728 (-mvsx-scalar-double): Likewise.
12729 (-mallow-df-permute): Likewise.
12730 (-mvectorize-builtins): Likewise.
12731 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12732 (rs6000_builtin_vectorized_function): Likewise.
12733 (rs6000_builtin_md_vectorized_function): Likewise.
12734 (rs6000_opt_vars): Likewise.
12736 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12739 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12740 (*btrq_imm): Rename from *btrq.
12741 (*btcq_imm): Rename from *btcq.
12742 (btsc): New code attribute.
12743 (*<btsc><mode>): New insn pattern.
12744 (*btr<mode>): Ditto.
12745 (*<btsc><mode>_mask): New insn_and_split pattern.
12746 (*btr<mode>_mask): Ditto.
12748 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12749 Alan Hayward <alan.hayward@arm.com>
12750 David Sherwood <david.sherwood@arm.com>
12752 * function.c (pad_below): Simplify padding calculation.
12754 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12755 Alan Hayward <alan.hayward@arm.com>
12756 David Sherwood <david.sherwood@arm.com>
12758 * target.def (function_prologue): Remove frame size argument.
12759 (function_epilogue): Likewise.
12760 * doc/tm.texi: Regenerate.
12761 * final.c (final_start_function): Update call to function_prologue.
12762 (final_end_function): Update call to function_epilogue.
12763 (default_function_pro_epilogue): Remove frame size argument.
12764 * output.h (default_function_pro_epilogue): Likewise.
12765 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12766 (arm_output_function_prologue): Likewise.
12767 * config/frv/frv.c (frv_function_prologue): Likewise.
12768 (frv_function_epilogue): Likewise.
12769 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12770 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12771 (ia64_output_function_epilogue): Likewise.
12772 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12773 (m32r_output_function_epilogue): Likewise.
12774 * config/microblaze/microblaze.c (microblaze_function_prologue)
12775 (microblaze_function_epilogue): Likewise.
12776 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12777 (mips_output_function_epilogue): Likewise.
12778 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12779 (mmix_target_asm_function_epilogue): Likewise.
12780 * config/msp430/msp430.c (msp430_start_function): Likewise.
12781 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12782 (nds32_asm_function_epilogue): Likewise.
12783 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12784 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12785 (pa_output_function_epilogue): Likewise.
12786 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12787 (rs6000_output_function_epilogue): Likewise.
12788 * config/rl78/rl78.c (rl78_start_function): Likewise.
12789 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12790 (rs6000_output_function_epilogue): Likewise.
12791 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12792 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12793 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12794 (sparc_asm_function_epilogue): Likewise.
12796 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12798 * tree.h (type_has_mode_precision_p): New function.
12799 * convert.c (convert_to_integer_1): Use it.
12800 * expr.c (expand_expr_real_2): Likewise.
12801 (expand_expr_real_1): Likewise.
12802 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12803 * match.pd: Likewise.
12804 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12805 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12806 * tree-tailcall.c (process_assignment): Likewise.
12807 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12808 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12809 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12810 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12811 (vectorizable_assignment): Likewise.
12812 (vectorizable_shift): Likewise.
12813 (vectorizable_operation): Likewise.
12814 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12816 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12818 * match.pd: Add pow (C, x) simplification.
12820 2017-08-21 Richard Biener <rguenther@suse.de>
12822 PR tree-optimization/81900
12823 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12824 for blocks with abnormal predecessors.
12825 (compute_antic): Do not set visited flag prematurely.
12827 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12830 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12832 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12834 * stor-layout.h (vector_type_mode): Move to...
12835 * tree.h (vector_type_mode): ...here.
12836 * stor-layout.c (vector_type_mode): Move to...
12837 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12839 2017-08-21 Richard Biener <rguenther@suse.de>
12841 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12842 register_external_die hooks.
12843 (debug_false_tree_charstarstar_uhwistar): Declare.
12844 (debug_nothing_tree_charstar_uhwi): Likewise.
12845 * debug.c (do_nothing_debug_hooks): Adjust.
12846 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12847 (debug_nothing_tree_charstar_uhwi): Likewise.
12848 * dbxout.c (dbx_debug_hooks): Adjust.
12849 (xcoff_debug_hooks): Likewise.
12850 * sdbout.c (sdb_debug_hooks): Likewise.
12851 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12852 * dwarf2out.c (macinfo_label_base): New global.
12853 (dwarf2out_register_external_die): New function for the
12854 register_external_die hook.
12855 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12856 (dwarf2_debug_hooks): Use them.
12857 (dwarf2_lineno_debug_hooks): Adjust.
12858 (struct die_struct): Add with_offset flag.
12859 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12860 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12861 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12862 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12863 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12864 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12865 defining section names for the early LTO debug variants.
12866 (reset_indirect_string): New helper.
12867 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12868 (print_dw_val): Add support for offsetted symbol references.
12869 (get_ultimate_context): Split out from is_cxx.
12870 (is_cxx): Use get_ultimate_context.
12871 (is_fortran): Add decl overload.
12872 (compute_comp_unit_symbol): Split out worker from
12873 compute_section_prefix.
12874 (compute_section_prefix): Call compute_comp_unit_symbol and
12875 set comdat_type_p here.
12876 (output_die): Skip DIE symbol output for the LTO added one.
12877 Handle DIE symbol references with offset.
12878 (output_comp_unit): Guard section name mangling properly.
12879 For LTO debug sections emit a symbol at the section beginning
12880 which we use to refer to its DIEs.
12881 (add_abstract_origin_attribute): For DIEs registered via
12882 dwarf2out_register_external_die directly refer to the early
12883 DIE rather than indirectly through the shadow one we created.
12884 Remove obsolete call to dwarf2out_abstract_function for
12885 non-function/block origins.
12886 (gen_array_type_die): When generating early LTO debug do
12887 not emit DW_AT_string_length.
12888 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12889 late when in LTO. As suggested place a gcc_unreachable for
12890 the DECL_ABSTRACT_P case.
12891 (gen_subprogram_die): Avoid another specification DIE
12892 for early built declarations/definitions for the late LTO case.
12893 (gen_variable_die): Add type references for late duplicated VLA dies
12895 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12896 we have the abstract instance already.
12897 (process_scope_var): Adjust decl DIE contexts in LTO which
12898 first puts them in limbo.
12899 (gen_decl_die): Do not generate type DIEs late apart from
12900 types for VLAs or for decls we do not yet have a DIE. Do not
12901 call dwarf2out_abstract_function late.
12902 (dwarf2out_early_global_decl): Make sure to create DIEs
12903 for abstract instances of a decl first.
12904 (dwarf2out_late_global_decl): Adjust comment.
12905 (output_macinfo_op): With multiple macro sections use
12906 macinfo_label_base to distinguish labels.
12907 (output_macinfo): Likewise. Update macinfo_label_base.
12908 Pass in the line info label.
12909 (note_variable_value_in_expr): When generating LTO resolve
12910 all variable values here by generating DIEs as needed.
12911 (init_sections_and_labels): Add early LTO debug flag parameter
12912 and generate different sections and names if set. Add generation
12913 counter for the labels so we can have multiple of them.
12914 (reset_dies): Helper to allow DIEs to be output multiple times.
12915 (dwarf2out_finish): When outputting DIEs to the fat part of an
12916 LTO object first reset DIEs.
12917 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12918 (modified_type_die): Check for decl_ultimate_origin being self
12920 (gen_type_die_with_usage): Likewise.
12921 (gen_typedef_die): Allow decl_ultimate_origin being self.
12922 (set_decl_abstract_flags): Remove.
12923 (set_block_abstract_flags): Likewise.
12924 (dwarf2out_abstract_function): Treat the early generated DIEs
12925 as the abstract copy and only add DW_AT_inline and
12926 DW_AT_artificial here and call set_decl_origin_self.
12927 If the DIE has an abstract origin don't do anything.
12928 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12929 if we have none yet (Go fails to build one, PR78628).
12930 (variably_modified_type_p): Prevent endless recursion for Ada
12931 cyclic pointer types.
12932 * lto-streamer-in.c: Include debug.h.
12933 (dref_queue): New global.
12934 (lto_read_tree_1): Stream in DIE references.
12935 (lto_input_tree): Register DIE references.
12936 (input_function): Stream DECL_DEBUG_ARGS.
12937 * lto-streamer-out.c: Include debug.h.
12938 (lto_write_tree_1): Output DIE references.
12939 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12940 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12941 (output_function): Stream DECL_DEBUG_ARGS.
12942 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12943 Stream DECL_ABSTRACT_ORIGIN.
12944 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12945 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12946 DECL_CONTEXT for file-scope decls.
12947 * lto-streamer.h (struct dref_entry): Declare.
12948 (dref_queue): Likewise.
12949 * cfgexpand.c (pass_expand::execute): Do not call the
12950 outlining_inline_function hook here.
12951 * lto-wrapper.c (debug_obj): New global.
12952 (tool_cleanup): Unlink it if required.
12953 (debug_objcopy): New function.
12954 (run_gcc): Handle early debug sections in the IL files by
12955 extracting them to separate files, partially linkin them and
12956 feeding the result back as result to the linker.
12957 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12958 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12959 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12960 sections into a separate segment.
12961 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12963 (darwin_asm_dwarf_section): Likewise.
12964 (darwin_asm_output_dwarf_offset): Likewise.
12965 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12967 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12968 Alan Hayward <alan.hayward@arm.com>
12969 David Sherwood <david.sherwood@arm.com>
12971 * read-md.h (md_reader::record_potential_iterator_use): Replace
12972 pointer argument with an rtx and an index.
12973 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12974 (apply_mode_iterator): Likewise.
12975 (apply_code_iterator): Likewise.
12976 (apply_int_iterator): Likewise.
12977 (apply_subst_iterator): Likewise.
12978 (record_iterator_use): Likewise.
12979 (record_attribute_use): Likewise.
12980 (md_reader::record_potential_iterator_use): Likewise. Update calls
12981 to record_iterator_use and apply_iterator.
12982 (iterator_use): Replace ptr with x and index.
12983 (attribute_use): Likewise.
12984 (apply_attribute_uses): Update calls to apply_iterator.
12985 (apply_iterators): Likewise. Update initialization of iterator_use.
12986 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12987 and record_potential_iterator_use.
12988 (rtx_reader::read_rtx_operand): Likewise.
12990 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12991 Alan Hayward <alan.hayward@arm.com>
12992 David Sherwood <david.sherwood@arm.com>
12994 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12997 2017-08-21 Richard Biener <rguenther@suse.de>
12999 PR middle-end/81884
13000 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13001 at struct end conservatively when comparing common bases.
13003 2017-08-21 Richard Biener <rguenther@suse.de>
13005 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13006 (mem_ref_in_stmt): Remove.
13007 (determine_max_movement): Use ref index to get at the reference.
13008 (invariantness_dom_walker::before_dom_children): Deal with
13009 lim data already initialized.
13010 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13012 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13014 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13015 (TARGET_ISA_ROUND): Ditto.
13016 (TARGET_ROUND): Ditto.
13017 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13018 * config/i386/i386.md: Ditto.
13019 * config/i386/sse.md: Ditto.
13020 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13021 with OPTION_MASK_ISA_SSE4_1.
13023 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13026 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13027 __builtin_ia32_lzcnt_u16.
13029 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13032 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13033 (rs6000_set_current_function): Rewrite function to use it.
13035 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13038 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13039 and TYPE_WARN_IF_NOT_ALIGN.
13040 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13041 (handle_warn_if_not_align): New.
13042 (place_union_field): Call handle_warn_if_not_align.
13043 (place_field): Call handle_warn_if_not_align.
13044 Copy TYPE_WARN_IF_NOT_ALIGN.
13045 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13046 (layout_type): Likewise.
13047 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13049 (tree_decl_common): Add warn_if_not_align.
13050 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13051 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13052 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13053 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13054 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13055 * doc/extend.texi: Document warn_if_not_aligned attribute.
13056 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13058 2017-08-17 Martin Liska <mliska@suse.cz>
13061 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13062 (get_data_dependence): Use it as pointer type.
13063 (distribute_loop): Likewise.
13065 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13067 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13068 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13069 (p8_vmrgow_<mode>_direct): New define_insn.
13070 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13071 handle endianness for vmrgew and vmrgow permute patterns.
13073 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13075 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13076 * config/rs6000/rs6000-cpus.def: Remove comment.
13077 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13078 (POWERPC_MASKS): Likewise.
13079 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13080 use of TARGET_VSX_TIMODE.
13081 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13082 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13083 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13084 (rs6000_option_override_internal): Remove dead code.
13085 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13086 (rs6000_legitimize_reload_address): Likewise.
13087 (rs6000_legitimate_address_p): Likewise.
13088 (rs6000_opt_masks): Delete "vsx-timode".
13089 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13090 from function comment.
13091 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13092 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13093 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13095 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13096 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13097 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13098 (VSX_TI): Likewise.
13100 (define_peephole2): Likewise.
13102 2017-08-17 Martin Sebor <msebor@redhat.com>
13105 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13106 past the end of an array.
13107 (test_pp_format): Add test cases.
13109 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13111 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13112 missing ECF_NOTHROW flags.
13114 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13117 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13118 operands residing in integer registers.
13119 (*vsx_le_perm_load_<mode>): Likewise.
13120 (*vsx_le_perm_store_<mode>): Likewise.
13121 (define_peephole2): Add peepholes to optimize the above.
13123 2017-08-17 Marek Polacek <polacek@redhat.com>
13125 PR middle-end/81814
13126 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13127 to mimic what shorten_compare did. Change the return type to bool.
13128 (fold_cond_expr_with_comparison): Update call to
13129 operand_equal_for_comparison_p.
13130 (fold_ternary_loc): Likewise.
13132 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13134 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13136 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13138 2017-08-17 Richard Biener <rguenther@suse.de>
13140 * tree-ssa-structalias.c (solve_graph): When propagating
13141 to successors update the graphs succ edges and avoid duplicate work.
13143 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13146 * config/i386/i386.c (ix86_option_override_internal): Save target
13147 specific options after ix86_stack_protector_guard_reg was changed.
13149 2017-08-17 Richard Biener <rguenther@suse.de>
13151 PR tree-optimization/81827
13152 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13153 (new_var_info): Initialize it conservatively.
13154 (get_call_vi): Mark register vars.
13155 (new_scalar_tmp_constraint_exp): Likewise.
13156 (handle_rhs_call): Likewise.
13157 (handle_const_call): Likewise.
13158 (create_function_info_for): Likewise.
13159 (solve_constraints): Sort varinfos to separate register from
13160 non-register vars to pack points-to solution bitmaps during
13163 2017-08-17 Marek Polacek <polacek@redhat.com>
13165 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13167 2017-08-17 Richard Biener <rguenther@suse.de>
13169 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13170 to true when overflow is undefined and we saturated the result.
13172 2017-08-17 Alan Modra <amodra@gmail.com>
13175 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13176 Don't use store multiple if only one reg needs saving.
13177 (interesting_frame_related_regno): New function.
13178 (rs6000_frame_related): Don't emit frame info for regs that
13180 (rs6000_emit_epilogue): Likewise.
13182 2017-08-16 Nathan Sidwell <nathan@acm.org>
13184 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13185 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13186 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13187 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13188 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13189 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13190 (verify_type): Adjust for TYPE_BINFO move.
13191 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13192 process TYPE_BINFO directly.
13193 (hash_tree): Likewise.
13194 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13196 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13199 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13201 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13203 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13206 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13207 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13208 Add ix86_binary_operator_ok to insn constraint.
13209 (*iordi_1_bts): Ditto.
13210 (*xordi_1_btc): Ditto.
13211 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13212 Update corresponding peephole2 pattern.
13216 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13218 PR tree-optimization/81832
13219 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13220 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13222 2017-08-16 Marek Polacek <polacek@redhat.com>
13225 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13226 regarding size_zero_node.
13228 2017-08-16 Martin Liska <mliska@suse.cz>
13231 * config.gcc: Respect previously set extra_objs in case
13234 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13236 PR tree-optimization/81835
13237 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13238 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13239 not depend on the phi.
13241 2017-08-16 Alan Modra <amodra@gmail.com>
13243 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13246 2017-08-16 Alan Modra <amodra@gmail.com>
13248 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13249 (save_reg_p): ..into this. Update all callers.
13250 (first_reg_to_save): Simplify.
13252 2017-08-16 Alan Modra <amodra@gmail.com>
13254 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13257 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13261 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13262 constant count if that count is less than 32.
13264 2017-08-15 Nathan Sidwell <nathan@acm.org>
13266 * gcc.c (execute): Emit friendlier message if inferior is killed
13267 by an external cause.
13269 2017-08-15 Richard Biener <rguenther@suse.de>
13271 PR tree-optimization/81790
13272 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13273 CONSTRUCTORs from simplifying and VN.
13275 2017-08-14 Martin Sebor <msebor@redhat.com>
13277 * builtin-attrs.def: Add comments.
13279 2017-08-14 Martin Sebor <msebor@redhat.com>
13282 * doc/extend.texi (attribute nonstring): Document new attribute.
13284 2017-08-14 Martin Sebor <msebor@redhat.com>
13287 * tree-diagnostic.c (default_tree_printer): Handle %G.
13288 * gimple-pretty-print.h (percent_G_format): Declare new function.
13289 * gimple-pretty-print.c (percent_G_format): Define.
13290 * tree-pretty-print.c (percent_K_format): Add argument.
13292 2017-08-14 Martin Sebor <msebor@redhat.com>
13294 PR translation/79998
13295 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13296 Remove a stray space.
13298 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13301 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13302 (*iordi_1_bts): Ditto.
13303 (*xordi_1_btc): Ditto.
13305 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13308 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13309 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13311 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13312 quoted strings, and make more translator-friendly.
13313 (darwin_rs6000_override_options): Likewise.
13314 (rs6000_option_override_internal): Likewise.
13315 (rs6000_return_in_memory): Fix overlong line.
13316 (init_cmulative_args): Use quoted strings, and make more
13317 translator-friendly.
13318 (rs6000_pass_by_reference): Fix overlong line.
13319 (def_builtin): Use quoted strings.
13320 (altivec_expand_predicate_builtin): Use quoted strings, and make
13321 more translator-friendly.
13322 (htm_expand_builtin): Use quoted strings.
13323 (cpu_expand_builtin): Use quoted strings, and make more
13324 translator-friendly.
13325 (altivec_expand_builtin): Likewise.
13326 (paired_expand_predicate_builtin): Likewise.
13327 (rs6000_invalid_builtin): Likewise.
13328 (builtin_function_type): Use quoted strings.
13329 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13330 more translator-friendly.
13331 (rs6000_trampoline_init): Likewise.
13332 (rs6000_handle_altivec_attribute): Likewise.
13333 (rs6000_inner_target_options): Use quoted strings.
13334 (rs6000_disable_incompatible_switches): Likewise.
13335 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13336 strings, and make more translator-friendly.
13337 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13339 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13341 PR tree-optimization/81799
13342 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13343 cond_expr to simple gimple operand.
13345 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13347 PR middle-end/46932
13348 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13350 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13354 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13355 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13356 TARGET_GASISR_PROLOGUES.
13357 * config/avr/avr.c (avr_option_override): Same.
13358 (avr_pass_pre_proep::execute): Same.
13360 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13363 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13364 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13366 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13368 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13369 define_insn_and_split. Split to a memory load from 0 in
13370 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13371 using PTR mode iterator.
13372 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13373 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13374 (*add_tp_<mode>): Redefine as define_insn_and_split.
13375 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13376 address space. Merge with *add_tp_x32 using PTR mode iterator.
13377 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13378 Split to an add with a memory load from 0 in
13379 DEFAULT_TLS_SEG_REG address space.
13381 2017-08-12 Andrew Pinski <apinski@cavium.com>
13383 * config/aarch64/aarch64-option-extensions.def (rdma):
13384 Fix feature string to what Linux prints out in /proc/cpuinfo.
13386 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13389 * dwarf2out.c (modified_type_die): For C typedef types that have
13390 an ultimate origin, process the ultimate origin instead of the
13392 (gen_typedef_die): Assert that input DECLs have no ultimate
13394 (gen_type_die_with_usage): For typedef variants that have an
13395 ultimate origin, just call gen_decl_die on the original DECL.
13396 (process_scope_var): Avoid creating DIEs for local typedefs and
13397 concrete static variables.
13399 2017-08-12 Alan Modra <amodra@gmail.com>
13403 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13404 match gnu-user.h startfile.
13405 (ENDFILE_LINUX_SPEC): Similarly.
13407 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13410 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13412 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13414 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13415 * config/aarch64/aarch64.md (mov<mode>): Change.
13416 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13417 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13418 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13420 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13422 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13423 for storage order barriers.
13425 2017-08-11 Martin Liska <mliska@suse.cz>
13428 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13429 variables of void type.
13431 2017-08-11 Martin Liska <mliska@suse.cz>
13433 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13434 TARGET_SUPPORTS_ALIASES.
13435 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13436 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13437 (optimize_weakref): Likewise.
13438 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13439 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13440 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13442 2017-08-11 Martin Liska <mliska@suse.cz>
13445 * config/i386/i386.c (make_resolver_func): Do complete
13446 refactoring of the function.
13448 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13451 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13452 * config/i386/i386.c (ix86_stack_protect_guard): Use
13453 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13454 * doc/invoke.texi (x86 Options): Document
13455 -mstack-protector-guard-symbol= option.
13457 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13459 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13460 * config/i386/i386.c (ix86_split_stack_guard): New function.
13461 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13462 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13463 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13464 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13465 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13466 (split_stack_space_check): Call ix86_split_stack_guard.
13468 2017-08-10 Martin Sebor <msebor@redhat.com>
13470 * print-tree.c (print_node): Print location using the established
13472 Replace spaces with colons.
13473 (debug_raw, debug): Ditto.
13475 2017-08-10 Martin Sebor <msebor@redhat.com>
13478 * pretty-print.c (pp_format): Correct the handling of %s precision.
13480 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13483 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13485 (ix86_finalize_stack_frame_flags): This. Also clear
13486 frame_pointer_needed if -fno-omit-frame-pointer is used without
13488 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13489 with ix86_finalize_stack_frame_flags.
13490 (ix86_expand_epilogue): Likewise.
13491 (ix86_expand_split_stack_prologue): Likewise.
13492 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13494 2017-08-10 Martin Liska <mliska@suse.cz>
13497 * c-attribs.c (handle_target_attribute):
13498 Report warning for an empty string argument of target attribute.
13500 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13503 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13505 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13506 or DECL_NONLOCAL labels.
13507 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13508 or DECL_NONLOCAL labels here.
13510 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13512 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13513 to indicate when early gimple folding has been disabled.
13514 (rs6000_gimple_fold_builtin): Add debug content.
13515 (rs6000_invalid_builtin): Fix whitespace.
13516 (rs6000_expand_builtin): Fix whitespace.
13517 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13519 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13522 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13523 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13525 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13527 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13528 (qdf24xx): Likewise.
13529 * config/aarch64/aarch64.md: Include falkor.md.
13530 * config/aarch64/falkor.md: New.
13532 2017-08-09 Marek Polacek <polacek@redhat.com>
13535 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13536 * diagnostic.c (emit_diagnostic): Add a comment.
13537 (emit_diagnostic_valist): New function.
13539 2017-08-09 Marek Polacek <polacek@redhat.com>
13542 * input.c (make_location): New overload.
13543 * input.h (make_location): Declare.
13545 2017-08-08 Alan Modra <amodra@gmail.com>
13546 H.J. Lu <hongjiu.lu@intel.com>
13549 * gcc.c (NO_PIE_SPEC): Delete.
13550 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13552 (LINK_PIE_SPEC): ..to here.
13553 (LINK_COMMAND_SPEC): Support -no-pie.
13554 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13555 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13557 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13558 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13559 (ENDFILE_CRTEND_SPEC): Similarly.
13561 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13564 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13565 (mstack-protector-guard-offset=): Ditto.
13566 * config/i386/i386.c (ix86_option_override): Handle
13567 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13569 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13570 ix86_stack_protect_guard_offset variables.
13571 (TARGET_STACK_PROTECT_GUARD): Always define.
13572 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13573 and -mstack-protector-guard-offset= options.
13575 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13577 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13578 boundary case for the last candidate.
13580 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13582 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13583 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13585 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13587 PR middle-end/19706
13588 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13589 * config/aarch64/aarch64-builtins.c
13590 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13591 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13592 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13594 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13595 Andrew Pinski <pinskia@gmail.com>
13597 PR middle-end/19706
13598 * internal-fn.def (XORSIGN): New.
13599 * optabs.def (xorsign_optab): New.
13600 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13601 (convert_expand_mult_copysign): New.
13602 (pass_optimize_widening_mul::execute): Call
13603 convert_expand_mult_copysign.
13605 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13607 PR tree-optimization/81354
13608 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13609 Insert on edges rather than explicitly creating landing pads.
13610 (analyze_candidates_and_replace): Commit edge inserts.
13612 2017-08-08 Richard Biener <rguenther@suse.de>
13614 PR middle-end/81719
13615 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13616 (expand_simple_operations): Also look through ADDR_EXPRs with
13617 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13619 2017-08-08 Richard Biener <rguenther@suse.de>
13621 PR tree-optimization/81723
13622 * tree-vect-slp.c (struct bst_traits): New hash traits.
13623 (bst_fail): New global.
13624 (vect_build_slp_tree_2): New worker, split out from ...
13625 (vect_build_slp_tree): ... this now wrapping it with using
13626 bst_fail set to cache SLP tree build fails. Properly handle
13628 (vect_analyze_slp_instance): Allocate and free bst_fail.
13630 2017-08-08 Martin Liska <mliska@suse.cz>
13633 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13634 LABEL_DECLs that can be from a different function.
13636 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13638 PR tree-optimization/81744
13639 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13640 loop's number of iterations.
13642 2017-08-08 Martin Liska <mliska@suse.cz>
13644 * asan.c: Include header files.
13645 * attribs.c (build_decl_attribute_variant): New function moved
13647 (build_type_attribute_qual_variant): Likewise.
13648 (cmp_attrib_identifiers): Likewise.
13649 (simple_cst_list_equal): Likewise.
13650 (omp_declare_simd_clauses_equal): Likewise.
13651 (attribute_value_equal): Likewise.
13652 (comp_type_attributes): Likewise.
13653 (build_type_attribute_variant): Likewise.
13654 (lookup_ident_attribute): Likewise.
13655 (remove_attribute): Likewise.
13656 (merge_attributes): Likewise.
13657 (merge_type_attributes): Likewise.
13658 (merge_decl_attributes): Likewise.
13659 (merge_dllimport_decl_attributes): Likewise.
13660 (handle_dll_attribute): Likewise.
13661 (attribute_list_equal): Likewise.
13662 (attribute_list_contained): Likewise.
13663 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13664 (lookup_attribute_by_prefix): Likewise.
13665 * bb-reorder.c: Include header files.
13666 * builtins.c: Likewise.
13667 * calls.c: Likewise.
13668 * cfgexpand.c: Likewise.
13669 * cgraph.c: Likewise.
13670 * cgraphunit.c: Likewise.
13671 * convert.c: Likewise.
13672 * dwarf2out.c: Likewise.
13673 * final.c: Likewise.
13674 * fold-const.c: Likewise.
13675 * function.c: Likewise.
13676 * gimple-expr.c: Likewise.
13677 * gimple-fold.c: Likewise.
13678 * gimple-pretty-print.c: Likewise.
13679 * gimple.c: Likewise.
13680 * gimplify.c: Likewise.
13681 * hsa-common.c: Likewise.
13682 * hsa-gen.c: Likewise.
13683 * internal-fn.c: Likewise.
13684 * ipa-chkp.c: Likewise.
13685 * ipa-cp.c: Likewise.
13686 * ipa-devirt.c: Likewise.
13687 * ipa-fnsummary.c: Likewise.
13688 * ipa-inline.c: Likewise.
13689 * ipa-visibility.c: Likewise.
13691 * lto-cgraph.c: Likewise.
13692 * omp-expand.c: Likewise.
13693 * omp-general.c: Likewise.
13694 * omp-low.c: Likewise.
13695 * omp-offload.c: Likewise.
13696 * omp-simd-clone.c: Likewise.
13697 * opts-global.c: Likewise.
13698 * passes.c: Likewise.
13699 * predict.c: Likewise.
13700 * sancov.c: Likewise.
13701 * sanopt.c: Likewise.
13702 * symtab.c: Likewise.
13703 * toplev.c: Likewise.
13704 * trans-mem.c: Likewise.
13705 * tree-chkp.c: Likewise.
13706 * tree-eh.c: Likewise.
13707 * tree-into-ssa.c: Likewise.
13708 * tree-object-size.c: Likewise.
13709 * tree-parloops.c: Likewise.
13710 * tree-profile.c: Likewise.
13711 * tree-ssa-ccp.c: Likewise.
13712 * tree-ssa-live.c: Likewise.
13713 * tree-ssa-loop.c: Likewise.
13714 * tree-ssa-sccvn.c: Likewise.
13715 * tree-ssa-structalias.c: Likewise.
13716 * tree-ssa.c: Likewise.
13717 * tree-streamer-in.c: Likewise.
13718 * tree-vectorizer.c: Likewise.
13719 * tree-vrp.c: Likewise.
13720 * tsan.c: Likewise.
13721 * ubsan.c: Likewise.
13722 * varasm.c: Likewise.
13723 * varpool.c: Likewise.
13724 * tree.c: Remove functions moved to attribs.[ch].
13725 * tree.h: Likewise.
13726 * config/aarch64/aarch64.c: Add attrs.h header file.
13727 * config/alpha/alpha.c: Likewise.
13728 * config/arc/arc.c: Likewise.
13729 * config/arm/arm.c: Likewise.
13730 * config/avr/avr.c: Likewise.
13731 * config/bfin/bfin.c: Likewise.
13732 * config/c6x/c6x.c: Likewise.
13733 * config/cr16/cr16.c: Likewise.
13734 * config/cris/cris.c: Likewise.
13735 * config/darwin.c: Likewise.
13736 * config/epiphany/epiphany.c: Likewise.
13737 * config/fr30/fr30.c: Likewise.
13738 * config/frv/frv.c: Likewise.
13739 * config/ft32/ft32.c: Likewise.
13740 * config/h8300/h8300.c: Likewise.
13741 * config/i386/winnt.c: Likewise.
13742 * config/ia64/ia64.c: Likewise.
13743 * config/iq2000/iq2000.c: Likewise.
13744 * config/lm32/lm32.c: Likewise.
13745 * config/m32c/m32c.c: Likewise.
13746 * config/m32r/m32r.c: Likewise.
13747 * config/m68k/m68k.c: Likewise.
13748 * config/mcore/mcore.c: Likewise.
13749 * config/microblaze/microblaze.c: Likewise.
13750 * config/mips/mips.c: Likewise.
13751 * config/mmix/mmix.c: Likewise.
13752 * config/mn10300/mn10300.c: Likewise.
13753 * config/moxie/moxie.c: Likewise.
13754 * config/msp430/msp430.c: Likewise.
13755 * config/nds32/nds32-isr.c: Likewise.
13756 * config/nds32/nds32.c: Likewise.
13757 * config/nios2/nios2.c: Likewise.
13758 * config/nvptx/nvptx.c: Likewise.
13759 * config/pa/pa.c: Likewise.
13760 * config/pdp11/pdp11.c: Likewise.
13761 * config/powerpcspe/powerpcspe.c: Likewise.
13762 * config/riscv/riscv.c: Likewise.
13763 * config/rl78/rl78.c: Likewise.
13764 * config/rx/rx.c: Likewise.
13765 * config/s390/s390.c: Likewise.
13766 * config/sh/sh.c: Likewise.
13767 * config/sol2.c: Likewise.
13768 * config/sparc/sparc.c: Likewise.
13769 * config/spu/spu.c: Likewise.
13770 * config/stormy16/stormy16.c: Likewise.
13771 * config/tilegx/tilegx.c: Likewise.
13772 * config/tilepro/tilepro.c: Likewise.
13773 * config/v850/v850.c: Likewise.
13774 * config/vax/vax.c: Likewise.
13775 * config/visium/visium.c: Likewise.
13776 * config/xtensa/xtensa.c: Likewise.
13778 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13781 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13782 constraints since the -mupper-regs-* switches have been
13784 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13785 into a vector from a double word element that was extracted from
13786 another vector, and eliminate extra XXPERMDI instructions.
13787 (vsx_concat_<mode>_2): Likewise.
13788 (vsx_concat_<mode>_3): Likewise.
13789 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13790 concat to allow optimizing inserts from previous extracts.
13792 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13794 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13795 memory reference to a SSP offset in TLS address space.
13796 (ix86_print_operand) <case '@'>: Remove.
13797 (ix86_print_operand_punct_valid_p): Remove '@' code.
13798 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13799 UNSPEC_SP_TLS_TEST.
13800 (stack_tls_protect_set_<mode>): Remove.
13801 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13802 (stack_tls_protect_test_<mode>): Remove.
13803 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13805 2017-08-07 Olivier Hainque <hainque@adacore.com>
13808 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13810 2017-08-07 Douglas Rupp <rupp@adacore.com>
13812 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13813 variable was referenced as multidir in command.
13815 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13818 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13821 2017-08-07 Martin Liska <mliska@suse.cz>
13823 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13824 * config/rl78/rl78.c: Add include of attribs.h.
13825 * config/sh/sh.c: Likewise.
13826 * config/v850/v850.c: Likewise.
13828 2017-08-07 Tom de Vries <tom@codesourcery.com>
13830 PR middle-end/78266
13831 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13833 2017-08-07 Martin Liska <mliska@suse.cz>
13835 * config/mips/mips.c: Include attribs.h.
13837 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13840 * doc/invoke.texi: Document change in behvaior for -Ofast for
13843 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13845 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13847 (aarch64_pop_regs): Likewise.
13848 (aarch64_gen_load_pair): Likewise.
13849 (aarch64_save_callee_saves): Likewise.
13850 (aarch64_restore_callee_saves): Likewise.
13852 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13854 * config/i386/i386.c: Revert the last change.
13856 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13859 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13861 (ix86_finalize_stack_frame_flags): This. Also clear
13862 frame_pointer_needed if -fno-omit-frame-pointer is used without
13864 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13865 with ix86_finalize_stack_frame_flags.
13866 (ix86_expand_epilogue): Likewise.
13867 (ix86_expand_split_stack_prologue): Likewise.
13869 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13872 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13873 to P_AES for Westmere.
13875 2017-08-07 Jonathan Yong <10walls@gmail.com>
13877 * config/i386/mingw.opt (fset-stack-executable): Removed.
13878 * config/i386/cygming.opt (fset-stack-executable): Moved
13880 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13882 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13884 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13886 2017-08-07 Marek Polacek <polacek@redhat.com>
13888 PR middle-end/81737
13889 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13891 2017-08-07 Martin Liska <mliska@suse.cz>
13893 * attribs.h (canonicalize_attr_name): New function.
13894 (cmp_attribs): Move from c-format.c and adjusted.
13895 (is_attribute_p): Moved from tree.h.
13896 * tree-inline.c: Add new includes.
13897 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13898 (private_is_attribute_p): Remove.
13899 (private_lookup_attribute): Likewise.
13900 (private_lookup_attribute_by_prefix): Simplify.
13901 (remove_attribute): Use is_attribute_p.
13902 * tree.h: Remove removed declarations.
13904 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13906 PR middle-end/81698
13907 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13908 instead of computing it in the function. Formatting fix.
13909 (expand_case): Don't rely on default_edge being the first edge,
13910 clear it if removing it, pass default_edge to
13911 emit_case_dispatch_table.
13912 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13915 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13917 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13918 insn in the function, emit NOP after the insn.
13920 2017-08-06 Tom de Vries <tom@codesourcery.com>
13922 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13925 2017-08-06 Tom de Vries <tom@codesourcery.com>
13927 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13930 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13932 PR tree-optimization/57371
13933 * match.pd: New pattern.
13935 2017-08-04 Marek Polacek <polacek@redhat.com>
13937 PR middle-end/81695
13938 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13939 perform the computation in offset_int.
13941 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13943 PR tree-optimization/81136
13944 * tree-vectorizer.h: Include tree-hash-traits.h.
13945 (vec_base_alignments): New typedef.
13946 (vec_info): Add a base_alignments field.
13947 (vect_record_base_alignments): Declare.
13948 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13950 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13951 (create_data_ref): Add an is_conditional_in_stmt argument.
13952 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13953 the is_conditional_in_stmt field.
13954 (data_ref_loc): Add an is_conditional_in_stmt field.
13955 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13956 (find_data_references_in_stmt): Update call to create_data_ref.
13957 (graphite_find_data_references_in_stmt): Likewise.
13958 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13959 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13960 (vect_record_base_alignment): New function.
13961 (vect_record_base_alignments): Likewise.
13962 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13963 for nested statements even if we fail to compute a misalignment.
13964 Use pooled base alignments for unconditional references.
13965 (vect_find_same_alignment_drs): Compare base addresses instead
13967 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13968 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13970 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13972 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13973 Add an explicit name for the enum. Use auto_vec for slp_instances
13974 and grouped_stores.
13975 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13977 (_bb_vec_info): Add a constructor and destructor.
13978 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13979 (destroy_loop_vec_info): Delete.
13980 (vect_destroy_datarefs): Likewise.
13981 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13982 (vec_info::vec_info): New function.
13983 (vec_info::~vec_info): Likewise.
13984 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13985 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13986 destroy_loop_vec_info.
13987 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13988 (_loop_vec_info::_loop_vec_info): ...this.
13989 (destroy_loop_vec_info): Replace with...
13990 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13991 the stmt_vec_infos. Leave handling of vec_info information to its
13992 destructor. Remove explicit vector releases.
13993 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13994 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13995 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13996 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13997 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13998 (destroy_bb_vec_info): Replace with...
13999 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14000 information to its destructor.
14001 (vect_slp_analyze_bb_1): Use new and delete instead of
14002 new_bb_vec_info and destroy_bb_vec_info.
14003 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14006 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14008 * tree-data-ref.h (subscript): Add access_fn field.
14009 (data_dependence_relation): Add could_be_independent_p.
14010 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14011 (same_access_functions): Move to tree-data-ref.c.
14012 * tree-data-ref.c (ref_contains_union_access_p): New function.
14013 (access_fn_component_p): Likewise.
14014 (access_fn_components_comparable_p): Likewise.
14015 (dr_analyze_indices): Add a reference to access_fn_component_p.
14016 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14018 (constant_access_functions): Likewise.
14019 (add_other_self_distances): Likewise.
14020 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14021 (initialize_data_dependence_relation): Use XCNEW and remove
14022 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14023 of access functions that have the same type. Allow the
14024 subsequence to end with different bases in some circumstances.
14025 Record the chosen access functions in SUB_ACCESS_FN.
14026 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14027 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14028 (subscript_dependence_tester_1): Likewise dra and drb.
14029 (build_classic_dist_vector): Update calls accordingly.
14030 (subscript_dependence_tester): Likewise.
14031 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14032 DDR_COULD_BE_INDEPENDENT_P.
14033 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14034 comp_alias_ddrs instead of may_alias_ddrs.
14035 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14037 (vect_analyze_data_ref_dependence): Use it if
14038 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14039 distance vectors if that fails.
14040 (dependence_distance_ge_vf): New function.
14041 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14042 LOOP_VINFO_MAY_ALIAS_DDRS.
14044 2017-08-04 Richard Biener <rguenther@suse.de>
14046 PR middle-end/81705
14047 * fold-const.c (fold_binary_loc): Properly restrict
14048 minus_var0 && minus_var1 case when associating undefined overflow
14051 2017-08-04 Tom de Vries <tom@codesourcery.com>
14053 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14055 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14057 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14058 Don't start diagnostic messages with a capital letter.
14059 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14061 (rs6000_invalid_builtin): Likewise.
14062 (rs6000_trampoline_init): Likewise.
14064 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14067 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14068 after setting changeable df flags.
14070 2017-08-03 Richard Biener <rguenther@suse.de>
14072 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14073 up if the use is in USE - X.
14075 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14077 * toplev.c (dumpfile.h): New include.
14078 (internal_error_reentered): New static function. Use it...
14079 (internal_error_function): ...here to handle reentered internal_error.
14081 2017-08-03 Richard Biener <rguenther@suse.de>
14083 PR middle-end/81148
14084 * fold-const.c (split_tree): Add minus_var and minus_con
14085 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14086 here but always use minus_*.
14087 (associate_trees): Assert we never associate with MINUS_EXPR
14088 and NULL first operand. Do not recurse for PLUS_EXPR operands
14089 when associating as MINUS_EXPR either.
14090 (fold_binary_loc): Track minus_var and minus_con.
14092 2017-08-03 Tom de Vries <tom@codesourcery.com>
14095 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14096 ACCEL_COMPILER, apply finish_options on
14097 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14099 2017-08-03 Tom de Vries <tom@codesourcery.com>
14102 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14103 function_entry_patch_area_size > 0.
14105 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14108 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14109 instead of 10??LU, perform unit multiplication in wide_int,
14110 don't change alloc_object_size_limit if the limit is larger
14113 PR tree-optimization/81655
14114 PR tree-optimization/81588
14115 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14116 the case when ranges[i].low and high are 1 for unsigned type with
14119 PR middle-end/81052
14120 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14121 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14123 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14125 * tree-vrp.h: Add include guard.
14127 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14130 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14131 (ud2): New insn pattern.
14132 * config/i386/i386.c (ix86_expand_epilogue):
14133 For naked functions, generate ud2 instead of trap insn.
14135 2017-08-02 Marek Polacek <polacek@redhat.com>
14138 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14140 2017-08-02 Tom de Vries <tom@codesourcery.com>
14141 Cesar Philippidis <cesar@codesourcery.com>
14143 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14144 Add missing edge probabilities.
14146 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14148 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14149 Correct endianness.
14151 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14153 PR middle-end/79499
14154 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14155 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14156 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14157 prologue_seq sequences - if any.
14159 2017-08-02 Richard Biener <rguenther@suse.de>
14161 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14162 via vectors if supported, integer extracts via punning if supported
14163 or otherwise vector extracts.
14165 2017-08-02 Richard Biener <rguenther@suse.de>
14167 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14169 (bitmap_insert_into_set): ... this.
14171 2017-08-02 Richard Biener <rguenther@suse.de>
14173 PR tree-optimization/81633
14175 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14177 PR tree-optimization/71752
14178 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14180 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14182 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14183 (machine_function::call_ms2sysv_pad_out): Remove field.
14184 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14185 (ix86_compute_frame_layout): Likewise.
14187 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14190 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14191 attribute with interrupt attribute.
14193 2017-08-01 Andrew Pinski <apinski@cavium.com>
14195 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14196 BIT_INSERT_EXPR's operand 1
14197 to see if the types precision matches.
14199 2017-08-01 Martin Liska <mliska@suse.cz>
14201 PR middle-end/70140
14202 * builtins.c (expand_builtin_memcpy_args): Remove.
14203 (expand_builtin_memcpy): Call newly added function
14204 expand_builtin_memory_copy_args.
14205 (expand_builtin_memcpy_with_bounds): Likewise.
14206 (expand_builtin_mempcpy): Remove last argument.
14207 (expand_builtin_mempcpy_with_bounds): Likewise.
14208 (expand_builtin_memory_copy_args): New function created from
14209 expand_builtin_mempcpy_args with small modifications.
14210 (expand_builtin_mempcpy_args): Remove.
14211 (expand_builtin_stpcpy): Remove unused argument.
14212 (expand_builtin): Likewise.
14213 (expand_builtin_with_bounds): Likewise.
14215 2017-08-01 Martin Liska <mliska@suse.cz>
14218 Make mempcpy more optimal (PR middle-end/70140).
14220 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14223 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14224 __builtin_vec_cmpne verify both arguments are compatible vectors
14225 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14226 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14227 move computation of aligned to after checking the argument types.
14231 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14232 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14235 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14236 Doug Rupp <rupp@adacore.com>
14237 Olivier Hainque <hainque@adacore.com>
14239 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14240 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14242 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14243 for TARGET_OS_CPP_BUILTIN.
14244 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14245 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14247 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14248 passing required abi options to the assembler for EABI configurations.
14249 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14250 of .text.hot and .text.unlikely sections for kernel modules when
14251 using ARM style exceptions.
14252 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14253 options. Add EXTRA_CC1_SPEC.
14254 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14256 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14258 (ARM_TARGET2_DWARF_FORMAT): Define.
14259 * config/arm/t-vxworks: Adjust multilib control to removal of the
14260 Diab command line options.
14262 2017-08-01 Martin Liska <mliska@suse.cz>
14264 PR gcov-profile/81561
14265 * gcov.c (unblock): Make unblocking safe as we need to preserve
14266 index correspondence of blocks and block_lists.
14268 2017-08-01 Richard Biener <rguenther@suse.de>
14270 PR tree-optimization/81181
14271 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14272 (compute_antic): ... end of iteration here.
14274 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14276 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14277 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14278 (ftree-slp-vectorize): Likewise.
14279 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14280 can no longer be set independent of flag_tree_loop_vectorize.
14281 * omp-general.c (emp_max_vf): Likewise.
14282 * opts.c (enable_fdo_optimizations): Remove references to
14283 flag_tree_vectorize, these are now implicit.
14284 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14285 and leave it for the options machinery.
14287 2017-08-01 Martin Liska <mliska@suse.cz>
14289 PR middle-end/70140
14290 * builtins.c (expand_builtin_memcpy_args): Remove.
14291 (expand_builtin_memcpy): Call newly added function
14292 expand_builtin_memory_copy_args.
14293 (expand_builtin_memcpy_with_bounds): Likewise.
14294 (expand_builtin_mempcpy): Remove last argument.
14295 (expand_builtin_mempcpy_with_bounds): Likewise.
14296 (expand_builtin_memory_copy_args): New function created from
14297 expand_builtin_mempcpy_args with small modifications.
14298 (expand_builtin_mempcpy_args): Remove.
14299 (expand_builtin_stpcpy): Remove unused argument.
14300 (expand_builtin): Likewise.
14301 (expand_builtin_with_bounds): Likewise.
14303 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14306 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14307 print "ds:" only for immediates in generic address space.
14309 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14312 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14313 (ix86_function_ok_for_sibcall): Return false for naked functions.
14315 2017-08-01 Richard Biener <rguenther@suse.de>
14317 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14318 (compute_antic): Seed worklist with exit block predecessors.
14319 * cfganal.c (dfs_find_deadend): For a cycle return the source
14320 of the edge closing it.
14322 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14324 * config/aarch64/aarch64.c
14325 (aarch64_can_const_movi_rtx_p): Move 0 check.
14327 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14329 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14330 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14332 * match.pd: Ditto in address comparison pattern.
14334 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14336 PR tree-optimization/81627
14337 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14338 closed ssa form for store-store chain.
14340 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14342 PR tree-optimization/81620
14343 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14344 for store-store chain.
14346 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14348 PR tree-optimization/81588
14349 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14350 ranges[i].in_p, invert comparison code ccode. For >/>=,
14351 swap rhs1 and rhs2 and comparison code unconditionally,
14352 for </<= don't do that. Don't swap rhs1/rhs2 again if
14353 ranges[i].in_p, instead invert comparison code ccode if
14354 opcode or oe->rank is BIT_IOR_EXPR.
14357 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14358 a direct optab to conversion optab.
14359 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14360 with GET_MODE_INNER as last argument instead of optab_handler.
14361 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14362 vector extraction if possible and optab is available.
14363 * expr.c (store_constructor): Use convert_optab_handler instead
14364 of optab_handler. Use vector initialization from smaller
14365 vectors if possible and optab is available.
14366 * tree-vect-stmts.c (vectorizable_load): Likewise.
14367 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14368 now have two modes.
14369 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14370 of vec_init from half-sized vectors with the same element mode.
14371 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14372 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14373 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14374 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14375 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14376 after mode in gen_vec_extract* calls.
14377 (vec_extract<mode>): Renamed to ...
14378 (vec_extract<mode><ssescalarmodelower>): ... this.
14379 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14380 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14381 element mode after mode in gen_vec_init* calls.
14382 (VEC_INIT_HALF_MODE): New mode iterator.
14383 (vec_init<mode>): Renamed to ...
14384 (vec_init<mode><ssescalarmodelower>): ... this.
14385 (vec_init<mode><ssehalfvecmodelower>): New expander.
14386 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14387 (vec_extractv2sfsf): ... this.
14388 (vec_initv2sf): Renamed to ...
14389 (vec_initv2sfsf): ... this.
14390 (vec_extractv2si): Renamed to ...
14391 (vec_extractv2sisi): ... this.
14392 (vec_initv2si): Renamed to ...
14393 (vec_initv2sisi): ... this.
14394 (vec_extractv4hi): Renamed to ...
14395 (vec_extractv4hihi): ... this.
14396 (vec_initv4hi): Renamed to ...
14397 (vec_initv4hihi): ... this.
14398 (vec_extractv8qi): Renamed to ...
14399 (vec_extractv8qiqi): ... this.
14400 (vec_initv8qi): Renamed to ...
14401 (vec_initv8qiqi): ... this.
14402 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14403 (vec_init<mode>): Renamed to ...
14404 (vec_init<mode><VEC_base_l>): ... this.
14405 (vec_extract<mode>): Renamed to ...
14406 (vec_extract<mode><VEC_base_l>): ... this.
14407 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14408 (vec_initv2sfsf): ... this.
14409 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14410 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14411 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14412 element mode after mode in gen_vec_init* calls.
14413 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14414 (vec_init<mode><Vel>): ... this.
14415 (vec_extract<mode>): Renamed to ...
14416 (vec_extract<mode><Vel>): ... this.
14417 * config/aarch64/iterators.md (Vel): New mode attribute.
14418 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14419 Add element mode after mode in gen_vec_extract* calls.
14420 * config/s390/vector.md (non_vec_l): New mode attribute.
14421 (vec_extract<mode>): Renamed to ...
14422 (vec_extract<mode><non_vec_l>): ... this.
14423 (vec_init<mode>): Renamed to ...
14424 (vec_init<mode><non_vec_l>): ... this.
14425 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14426 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14428 * config/arm/iterators.md (V_elem_l): New mode attribute.
14429 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14430 (vec_extract<mode><V_elem_l>): ... this.
14431 (vec_extractv2di): Renamed to ...
14432 (vec_extractv2didi): ... this.
14433 (vec_init<mode>): Renamed to ...
14434 (vec_init<mode><V_elem_l>): ... this.
14435 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14436 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14437 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14438 Add element mode after gen_vec_extract* calls.
14439 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14440 (vec_init<mode><unitmode>): ... this.
14441 (vec_extract<mode>): Renamed to ...
14442 (vec_extract<mode><unitmode>): ... this.
14443 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14444 (vec_init<mode><unitmode>): ... this.
14445 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14446 (vec_initv2sfsf): ... this.
14447 (vec_extractv2sf): Renamed to ...
14448 (vec_extractv2sfsf): ... this.
14449 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14450 Add element mode after gen_vec_extract* calls.
14451 * config/mips/mips.md (unitmode): New mode iterator.
14452 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14453 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14454 * config/spu/spu.md (inner_l): New mode attribute.
14455 (vec_init<mode>): Renamed to ...
14456 (vec_init<mode><inner_l>): ... this.
14457 (vec_extract<mode>): Renamed to ...
14458 (vec_extract<mode><inner_l>): ... this.
14459 * config/sparc/sparc.md (veltmode): New mode iterator.
14460 (vec_init<VMALL:mode>): Renamed to ...
14461 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14462 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14463 (vec_initv2sisi): ... this.
14464 (vec_initv2sf): Renamed to ...
14465 (vec_initv2sfsf): ... this.
14466 (vec_extractv2sf): Renamed to ...
14467 (vec_extractv2sfsf): ... this.
14468 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14469 (vec_init<mode>): Renamed to ...
14470 (vec_init<mode><VEC_base_l>): ... this.
14471 (vec_extract<mode>): Renamed to ...
14472 (vec_extract<mode><VEC_base_l>): ... this.
14473 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14474 (vec_initv2sfsf): ... this.
14475 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14476 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14477 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14478 gen_vec_init* calls.
14480 2017-08-01 Richard Biener <rguenther@suse.de>
14482 PR tree-optimization/81297
14483 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14484 TREE_OVERFLOW from INTEGER_CSTs.
14486 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14488 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14490 2017-07-31 Carl Love <cel@us.ibm.com>
14492 * config/rs6000/rs6000-c: Add support for built-in functions
14493 vector signed char vec_xl_be (signed long long, signed char *);
14494 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14495 vector signed int vec_xl_be (signed long long, signed int *);
14496 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14497 vector signed long long vec_xl_be (signed long long, signed long long *);
14498 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14499 vector signed short vec_xl_be (signed long long, signed short *);
14500 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14501 vector double vec_xl_be (signed long long, double *);
14502 vector float vec_xl_be (signed long long, float *);
14503 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14504 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14505 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14507 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14508 (altivec_expand_builtin): Add switch statement to call
14509 altivec_expand_xl_be for each builtin.
14510 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14511 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14512 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14513 __builtin_vsx_le_be_v16qi.
14514 * doc/extend.texi: Update the built-in documentation file for the
14515 new built-in functions.
14517 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14520 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14522 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14524 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14526 * config.gcc: Add z14.
14527 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14528 CPU model numbers for z13s and z14.
14529 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14531 * config/s390/s390-opts.h (enum processor_type): Rename
14532 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14533 * config/s390/s390.c (processor_table): Add field for CPU name to
14534 be passed to Binutils.
14535 (s390_asm_output_machine_for_arch): Use the new field in
14536 processor_table for Binutils.
14537 (s390_expand_builtin): Replace arch12 with z14.
14538 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14539 (s390_get_sched_attrmask): Likewise.
14540 (s390_get_unit_mask): Likewise.
14541 * config/s390/s390.opt: Add z14 to processor_type enum.
14543 2017-07-31 Martin Jambor <mjambor@suse.cz>
14546 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14547 regardless of optimization level.
14549 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14550 Martin Liska <mliska@suse.cz>
14552 * predict.def: Remove old comment and adjust probability.
14553 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14554 PREDICT statements.
14556 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14559 * config/i386/i386.c (ix86_function_naked): New function.
14560 (ix86_can_use_return_insn_p): Return false for naked functions.
14561 (ix86_expand_prologue): Skip prologue for naked functions.
14562 (ix86_expand_epilogue): Skip epilogue for naked functions
14563 and emit trap instruction.
14564 (ix86_warn_func_return): New function.
14565 (ix86_attribute_table): Add "naked" attribute specification.
14566 (TARGET_WARN_FUNC_RETURN): Define.
14567 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14569 2017-07-31 Martin Liska <mliska@suse.cz>
14571 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14572 (dump_gimple_bb_header): Always dump BB info.
14573 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14575 2017-07-31 Martin Liska <mliska@suse.cz>
14578 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14579 also with current_function_decl non-null equality.
14581 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14584 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14585 change type to the element type, instead add eltype variable and
14586 use it where we are interested in the element type.
14588 PR tree-optimization/81603
14589 * ipa-polymorphic-call.c
14590 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14591 offset arithmetic in offset_int, bail out if the resulting bit offset
14592 doesn't fit into shwi.
14594 2017-07-31 Martin Liska <mliska@suse.cz>
14596 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14597 (gimplify_save_expr): Fix comment.
14599 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14602 * config/i386/i386.c (ix86_function_arg): Update arguments for
14604 (ix86_compute_frame_layout): Set the initial stack offset to
14605 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14606 INCOMING_FRAME_SP_OFFSET.
14607 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14608 stack before exception handler returns.
14609 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14610 the 'ERROR_CODE' for exception handler.
14612 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14614 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14615 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14616 (ASM_OUTPUT_REG_POP): Ditto.
14617 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14618 instead of asm_fprintf to output pure string.
14620 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14622 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14623 to imported_module_or_decl hook.
14624 (debug_nothing_tree_tree_tree_bool): Remove.
14625 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14626 * debug.c (do_nothing_debug_hooks): Use
14627 debug_nothing_tree_tree_tree_bool_bool instead of
14628 debug_nothing_tree_tree_tree_bool.
14629 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14630 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14631 * sdbout.c (sdb_debug_hooks): Likewise.
14632 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14633 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14635 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14636 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14637 attribute, don't add anything.
14639 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14641 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14642 (fold_build2_stat_loc): Likewise.
14643 (fold_build3_stat_loc): Likewise.
14644 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14645 (fold_build1_loc): Remove macro.
14646 (fold_build2_loc): Likewise.
14647 (fold_build3_loc): Likewise.
14649 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14651 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14652 (gimple_build_debug_bind_source_stat): Likewise.
14653 * gimple.h (gimple_build_debug_bind): Remove macro.
14654 (gimple_build_debug_bind_source): Likewise.
14656 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14658 * bitmap.c (bitmap_alloc): Adjust.
14659 (bitmap_gc_alloc): Likewise.
14660 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14662 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14664 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14665 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14666 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14667 (bitmap_gc_alloc_stat): Likewise.
14668 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14670 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14672 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14673 * rtl.h (shallow_copy_rtx): Remove macro.
14675 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14677 * emit-rtl.c (gen_raw_REG): Adjust.
14678 * gengenrtl.c (gendef): Likewise.
14679 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14680 * rtl.h (rtx_alloc): Remove macro.
14682 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14684 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14685 (build_tree_list_stat): Likewise.
14686 * tree.h (build_tree_list): Remove macro.
14687 (build_tree_list_vec): Likewise.
14689 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14691 * tree.c (make_vector_stat): Remove _stat from name.
14692 (build_vector_stat): Likewise.
14693 * tree.h (make_vector_stat): Remove macro.
14694 (build_vector_stat): Likewise.
14696 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14698 * tree.h (build_var_debug_value): Remove prototype.
14700 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14702 * tree.c (tree_cons_stat): Remove _stat from name.
14703 * tree.h (tree_cons): Remove macro.
14705 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14707 * tree.c (build_vl_exp_stat): Remove _stat from name.
14708 * tree.h (build_vl_exp): Remove macro.
14710 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14712 * tree.c (build_decl_stat): Remove _stat from name.
14713 * tree.h (build_decl): Remove macro.
14715 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14717 * gimple.c (gimple_build_with_ops_stat): Adjust.
14718 (gimple_alloc_stat): Remove _stat from name.
14719 * gimple.h (gimple_alloc): Remove macro.
14721 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14723 * tree.c (make_tree_vec_stat): Remove _stat from name.
14724 (grow_tree_vec_stat): Likewise.
14725 * tree.h (make_tree_vec_stat): Adjust prototype.
14726 (grow_tree_vec_stat): Likewise.
14727 (make_tree_vec): Remove macro.
14728 (grow_tree_vec): Likewise.
14730 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14732 * fold-const.c (fold_build1_stat_loc): Adjust.
14733 (fold_build2_stat_loc): Likewise.
14734 (fold_build3_stat_loc): Likewise.
14735 * tree.c (build0_stat): Remove _stat from name.
14736 (build1_stat): Likewise.
14737 (build2_stat): Likewise.
14738 (build3_stat): Likewise.
14739 (build4_stat): Likewise.
14740 (build5_stat): Likewise.
14741 * tree.h (build1_loc): Remove macro, and rename _stat function
14743 (build2_loc): Likewise.
14744 (build3_loc): Likewise.
14745 (build4_loc): Likewise.
14746 (build5_loc): Likewise.
14748 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14750 * tree.c (make_int_cst_stat): Remove _stat from name.
14751 * tree.h (make_int_cst_stat): Adjust prototype.
14752 (make_int_cst): Remove macro.
14754 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14756 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14757 * tree.h (make_tree_binfo_stat): Adjust prototype.
14758 (make_tree_binfo): Remove.
14760 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14762 * tree.c (copy_node_stat): Rename to copy_node.
14763 (build_distinct_type_copy): Adjust.
14764 * tree.h (copy_node_stat): Adjust prototype.
14765 (copy_node): Remove macro.
14767 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14769 * tree.c (make_node_stat): rename to make_node.
14770 (build_tree_list_stat): Adjust.
14771 (build0_stat): Likewise.
14772 (build2_stat): Likewise.
14773 (build3_stat): Likewise.
14774 (build4_stat): Likewise.
14775 (build5_stat): Likewise.
14776 (build_decl_stat): Likewise.
14777 * tree.h (make_node_stat): Adjust prototype.
14778 (make_node): remove macro.
14780 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14782 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14783 (PPC_FEATURE2_SCV): Likewise.
14784 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14786 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14788 * config/aarch64/aarch64.c
14789 (aarch64_internal_mov_immediate): Add new special pattern.
14790 * config/aarch64/aarch64.md (*movdi_aarch64):
14791 Add reg/32bit const mov case.
14793 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14794 Richard Sandiford <richard.sandiford@linaro.org>
14796 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14797 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14798 Add integer and movi cases.
14799 (movi-split-hf-df-sf split, fp16): New.
14800 (enabled): Added TARGET_FP_F16INST.
14801 * config/aarch64/iterators.md (GPF_HF): New.
14802 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14804 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14806 * config/aarch64/aarch64.c
14807 (aarch64_simd_container_mode): Add prototype.
14808 (aarch64_expand_mov_immediate): Add HI support.
14809 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14810 (aarch64_can_const_movi_rtx_p): New.
14811 (aarch64_preferred_reload_class):
14812 Remove restrictions of using FP registers for certain SIMD operations.
14813 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14814 (aarch64_valid_floating_const): Add integer move validation.
14815 (aarch64_simd_imm_scalar_p): Remove.
14816 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14817 (aarch64_legitimate_constant_p): Expand list of supported cases.
14818 * config/aarch64/aarch64-protos.h
14819 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14820 (aarch64_reinterpret_float_as_int): New.
14821 (aarch64_simd_imm_scalar_p): Remove.
14822 * config/aarch64/constraints.md (Uvi): New.
14823 (Dd): Split into Ds and new Dd.
14824 * config/aarch64/aarch64.md (*movsi_aarch64):
14826 (*movdi_aarch64): Add SIMD mov case.
14828 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14830 * tree-predcom.c: (struct chain): Handle store-store chain in which
14831 stores for elimination only store loop invariant values.
14832 (execute_pred_commoning_chain): Ditto.
14833 (prepare_initializers_chain_store_elim): Ditto.
14834 (prepare_finalizers): Ditto.
14835 (is_inv_store_elimination_chain): New function.
14836 (initialize_root_vars_store_elim_1): New function.
14838 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14840 * tree-predcom.c: Revise general description of the pass.
14841 (enum chain_type): New enum type for store elimination.
14842 (struct chain): New field supporting store elimination.
14843 (struct component): Ditto.
14844 (dump_chain): Dump store-stores chain.
14845 (release_chain): Release resources.
14846 (split_data_refs_to_components): Compute and create component
14847 contains only stores for elimination.
14848 (get_chain_last_ref_at): New function.
14849 (make_invariant_chain): Initialization.
14850 (make_rooted_chain): Specify chain type in parameter and record it.
14851 (add_looparound_copies): Skip for store-stores chain.
14852 (determine_roots_comp): Compute type of chain and pass it to
14854 (initialize_root_vars_store_elim_2): New function.
14855 (finalize_eliminated_stores): New function.
14856 (remove_stmt): Handle store for elimination.
14857 (execute_pred_commoning_chain): Execute predictive commoning on
14858 store-store chains.
14859 (determine_unroll_factor): Skip unroll for store-stores chain.
14860 (prepare_initializers_chain_store_elim): New function.
14861 (prepare_initializers_chain): Hanlde store-store chain.
14862 (prepare_finalizers_chain, prepare_finalizers): New function.
14863 (tree_predictive_commoning_loop): Return integer value indicating
14864 if loop is unrolled or lcssa form is corrupted.
14865 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14867 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14869 * tree-predcom.c (initialize_root): Delete.
14870 (execute_pred_commoning_chain): Initialize root vars and replace
14871 reference of non-combined chain directly, rather than call above
14874 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14876 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14877 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14879 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14881 * tree-predcom.c (struct chain): New field init_seq.
14882 (release_chain): Release init_seq.
14883 (prepare_initializers_chain): Record intialization stmts in above
14885 (insert_init_seqs): New function.
14886 (tree_predictive_commoning_loop): Call insert_init_seqs.
14888 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14890 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14892 2017-07-28 Richard Biener <rguenther@suse.de>
14894 * match.pd: Remove superfluous :c.
14895 * genmatch.c (simplify::id): Add member.
14896 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14898 (current_id): New global.
14899 (dt_node::parent): Move from ...
14900 (dt_operand::parent): ... here. Add for_id member.
14901 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14902 (decision_tree::find_node): Relax order requirement when
14903 merging DT_TRUE nodes to ones inbetween the current simplify
14904 and the one we try to merge with. Add diagnostic whenever
14905 we need to enforce pattern order by not merging.
14906 (decision_tree::insert): Set current_id.
14907 (decision_tree::print_node): Dump parent node and for_id.
14908 (parser::last_id): Add member.
14909 (parser::push_simplify): Assign unique id.
14910 (parser::parser): Initialize last_id.
14912 2017-07-28 Martin Liska <mliska@suse.cz>
14915 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14916 gimple_build_debug_bind.
14918 2017-07-28 Richard Biener <rguenther@suse.de>
14920 PR tree-optimization/81502
14921 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14923 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14925 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14926 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14927 for BIT_FIELD_REF args.
14928 * fold-const.c (make_bit_field_ref): Likewise.
14929 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14931 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14934 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14935 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14936 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14937 Or it into SANITIZER_UNDEFINED.
14938 * ubsan.c: Include gimple-fold.h and varasm.h.
14939 (ubsan_expand_ptr_ifn): New function.
14940 (instrument_pointer_overflow): New function.
14941 (maybe_instrument_pointer_overflow): New function.
14942 (instrument_object_size): Formatting fix.
14943 (pass_ubsan::execute): Call instrument_pointer_overflow
14944 and maybe_instrument_pointer_overflow.
14945 * internal-fn.c (expand_UBSAN_PTR): New function.
14946 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14947 * sanitizer.def (__ubsan_handle_pointer_overflow,
14948 __ubsan_handle_pointer_overflow_abort): New builtins.
14949 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14950 * internal-fn.def (UBSAN_PTR): New internal function.
14951 * opts.c (sanitizer_opts): Add pointer-overflow.
14952 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14953 * fold-const.c (build_range_check): Compute pointer range check in
14954 integral type if pointer arithmetics would be needed. Formatting
14957 2017-07-28 Martin Liska <mliska@suse.cz>
14960 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14961 parameters that are of a variable-length.
14963 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14965 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14967 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14968 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14969 (CRT_CALL_STATIC_FUNCTION): Likewise.
14970 (ASM_DEFAULT_SPEC): New define.
14971 (ASM_SPEC32): Likewise.
14972 (ASM_SPEC64): Likewise.
14973 (ASM_SPEC_COMMON): Likewise.
14974 (ASM_SPEC): Likewise.
14975 (INVALID_64BIT): Likewise.
14976 (LINK_OS_DEFAULT_SPEC): Likewise.
14977 (LINK_OS_SPEC32): Likewise.
14978 (LINK_OS_SPEC64): Likewise.
14979 (POWERPC_LINUX): Likewise.
14980 (PTRDIFF_TYPE): Likewise.
14981 (RESTORE_FP_PREFIX): Likewise.
14982 (RESTORE_FP_SUFFIX): Likewise.
14983 (SAVE_FP_PREFIX): Likewise.
14984 (SAVE_FP_SUFFIX): Likewise.
14985 (SIZE_TYPE): Likewise.
14986 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14987 (TARGET_64BIT): Likewise.
14988 (TARGET_64BIT): Likewise.
14989 (TARGET_AIX): Likewise.
14990 (WCHAR_TYPE_SIZE): Likewise.
14991 (WCHAR_TYPE): Undefine.
14992 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14993 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14994 (CPP_OS_RTEMS_SPEC): Delete.
14995 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14996 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14998 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15000 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15002 PR tree-optimization/81578
15003 * tree-parloops.c (build_new_reduction): Bail out if
15004 reduction_code isn't one of the standard OpenMP reductions.
15005 Move the details printing after that decision.
15007 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15009 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15010 related to reload_in_progress.
15011 (splat_input_operand): Likewise.
15012 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15014 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15016 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15017 (TARGET_INSTANTIATE_DECLS): Likewise.
15018 (legitimate_indexed_address_p): Delete reload_in_progress code.
15019 (rs6000_debug_legitimate_address_p): Likewise.
15020 (rs6000_eliminate_indexed_memrefs): Likewise.
15021 (rs6000_emit_le_vsx_store): Likewise.
15022 (rs6000_emit_move_si_sf_subreg): Likewise.
15023 (rs6000_emit_move): Likewise.
15024 (register_to_reg_type): Likewise.
15025 (rs6000_pre_atomic_barrier): Likewise.
15026 (rs6000_machopic_legitimize_pic_address): Likewise.
15027 (rs6000_allocate_stack_temp): Likewise.
15028 (rs6000_address_for_fpconvert): Likewise.
15029 (rs6000_address_for_altivec): Likewise.
15030 (rs6000_secondary_memory_needed_rtx): Delete function.
15031 (rs6000_check_sdmode): Likewise.
15032 (rs6000_alloc_sdmode_stack_slot): Likewise.
15033 (rs6000_instantiate_decls): Likewise.
15034 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15035 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15036 Delete reload_in_progress.
15037 (*vec_reload_and_plus_<mptrsize>): Likewise.
15038 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15039 (vsx_div_v2di): Likewise.
15040 (vsx_udiv_v2di): Likewise.
15042 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15044 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15045 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15046 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15047 (rs6000_debug_reg_global): Delete print of LRA status.
15048 (rs6000_option_override_internal): Delete dead LRA related code.
15049 (rs6000_lra_p): Delete function.
15050 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15052 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15054 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15055 * config/riscv/rtems.h: New file.
15057 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15058 Sudakshina Das <sudi.das@arm.com>
15060 * config/aarch64/aarch64.md
15061 (define_split for and<mode>3nr_compare): Move
15062 non aarch64_logical_operand to a register.
15063 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15064 register immediate operand to a register.
15065 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15067 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15069 PR middle-end/81564
15070 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15072 2017-07-27 Richard Biener <rguenther@suse.de>
15074 PR tree-optimization/81573
15075 PR tree-optimization/81494
15076 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15077 multi defuse cycle case.
15079 2017-07-27 Richard Biener <rguenther@suse.de>
15081 PR tree-optimization/81571
15082 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15085 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15087 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15088 earlier and only if MASK_FPU is set. Adjust formatting.
15090 2017-07-27 Martin Liska <mliska@suse.cz>
15092 * opt-functions.awk: Add validation of value of Init.
15093 * optc-gen.awk: Pass new argument.
15095 2017-07-27 Martin Liska <mliska@suse.cz>
15097 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15098 Fix wrong condition.
15100 2017-07-27 Martin Liska <mliska@suse.cz>
15102 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15103 BBs and edges seen by autoFDO.
15105 2017-07-27 Richard Biener <rguenther@suse.de>
15107 PR tree-optimization/81502
15108 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15109 with incompatible but same sized type.
15110 (execute_update_addresses_taken): Likewise.
15112 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15114 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15115 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15117 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15120 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15121 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15122 Change s_operand to memory_operand.
15124 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15126 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15127 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15128 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15129 Emit instructions rather than returning an expression. Handle TFmode
15130 and KFmode by casting to TImode.
15131 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15132 (rs6000_emit_le_vsx_store): Likewise.
15133 * config/rs6000/vsx.md (VSX_TI): New iterator.
15134 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15135 (*vsx_le_undo_permute_<mode>): Likewise.
15136 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15137 emit the split sequence.
15138 (*vsx_le_perm_store_<mode>): Likewise.
15140 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15142 PR tree-optimization/81555
15143 PR tree-optimization/81556
15144 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15145 if true, force CHANGED for the recursive invocation.
15146 (reassociate_bb): Remember original length of ops array, pass
15147 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15149 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15150 attributes for noipa attribute. For naked attribute use
15151 lookup_attribute first before lookup_attribute_spec.
15152 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15154 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15155 for functions with noipa attribute.
15156 (cgraph_externally_visible_p): Return true for functions with noipa
15158 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15159 for functions with noipa attribute.
15160 * doc/extend.texi: Document noipa function attribute.
15161 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15162 also for functions with noipa attribute.
15163 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15165 2017-07-26 Andrew Pinski <apinski@cavium.com>
15167 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15168 vec_unalign_load_cost and vec_unalign_store_cost.
15170 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15172 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15173 -mvsx-small-integer option.
15174 (ISA_3_0_MASKS_IEEE): Likewise.
15175 (OTHER_VSX_VECTOR_MASKS): Likewise.
15176 (POWERPC_MASKS): Likewise.
15177 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15178 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15179 code, only testing for DImode being allowed in non-VSX floating
15181 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15182 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15184 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15185 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15186 TARGET_P8_VECTOR test.
15187 (rs6000_secondary_reload_simple_move): Likewise.
15188 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15189 since TARGET_P9_VECTOR was already tested.
15190 (rs6000_opt_masks): Remove -mvsx-small-integer.
15191 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15192 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15194 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15195 test for TARGET_VEXTRACTUB was used, and that uses
15197 (p9 extract splitter): Likewise.
15198 (vsx_extract_<mode>_di_p9): Likewise.
15199 (vsx_extract_<mode>_store_p9): Likewise.
15200 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15201 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15202 the elimination of TARGET_VSX_SMALL_INTEGER.
15203 (vsx_extract_<mode>_p8): Likewise.
15204 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15205 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15206 (vsx_set_<mode>_p9): Likewise.
15207 (vsx_set_v4sf_p9): Likewise.
15208 (vsx_set_v4sf_p9_zero): Likewise.
15209 (vsx_insert_extract_v4sf_p9): Likewise.
15210 (vsx_insert_extract_v4sf_p9_2): Likewise.
15211 * config/rs6000/rs6000.md (sign extend splitter): Change
15212 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15213 (floatsi<mode>2_lfiwax_mem): Likewise.
15214 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15215 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15216 since a test for TARGET_P9_VECTOR was used.
15217 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15218 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15219 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15220 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15221 TARGET_P8_VECTOR test.
15222 (fix_trunc<mode>si2_stfiwx): Likewise.
15223 (fix_trunc<mode>si2_internal): Likewise.
15224 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15225 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15227 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15228 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15229 TARGET_P8_VECTOR test.
15230 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15231 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15232 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15234 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15235 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15236 since a test for TARGET_P9_VECTOR was used.
15237 (splitter for loading small constants): Likewise.
15239 2017-07-26 Andrew Pinski <apinski@cavium.com>
15241 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15244 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15247 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15248 (fp_valid_at): Likewise.
15250 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15252 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15253 (qdf24xx_addrcost_table): Likewise.
15254 (cortexa57_tunings): Update to use generic_branch_cost.
15255 (cortexa72_tunings): Likewise.
15256 (cortexa73_tunings): Likewise.
15257 (qdf24xx_tunings): Likewise.
15259 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15261 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15262 (thunderx2t99_branch_cost): Likewise.
15263 (cortexa35_tunings): Update to use generic_branch_cost.
15264 (cortexa53_tunings): Likewise.
15265 (cortexa57_tunings): Likewise.
15266 (cortexa72_tunings): Likewise.
15267 (cortexa73_tunings): Likewise.
15268 (thunderx2t99_tunings): Likewise.
15270 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15272 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15273 (sparc_option_override): Honour MASK_FSMULD.
15274 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15275 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15276 * config/sparc/sparc.opt (mfsmuld): New option.
15277 * doc/invoke.texi (mfsmuld): Document option.
15279 2017-07-26 Marek Polacek <polacek@redhat.com>
15281 PR middle-end/70992
15282 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15284 2017-07-26 Richard Biener <rguenther@suse.de>
15286 * gimple-match-head.c (do_valueize): Return OP if valueize
15288 (get_def): New helper to get at the def stmt of a SSA name
15289 if valueize allows.
15290 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15291 do_valueize to get at the def stmt.
15292 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15294 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15296 PR middle-end/46932
15297 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15299 2017-07-26 Martin Liska <mliska@suse.cz>
15302 * function.c (expand_function_start): Make expansion of
15303 nonlocal_goto_save_area after parm_birth_insn.
15305 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15307 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15308 from all CPU target flags enable members.
15310 2017-07-26 Richard Biener <rguenther@suse.de>
15312 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15313 (decision_tree::gen): Make 'type' const.
15314 (write_predicate): Likewise.
15316 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15318 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15319 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15320 (rs6000_option_override_internal): Likewise.
15321 (rs6000_expand_vector_set): Likewise.
15322 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15323 (TARGET_UPPER_REGS_SF): Likewise.
15324 (TARGET_UPPER_REGS_DI): Likewise.
15325 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15326 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15327 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15328 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15329 (Splitters for DI constants in Altivec registers): Likewise.
15330 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15331 (vsx_set_v4sf_p9): Likewise.
15332 (vsx_set_v4sf_p9_zero): Likewise.
15333 (vsx_insert_extract_v4sf_p9): Likewise.
15334 (vsx_insert_extract_v4sf_p9_2): Likewise.
15336 2017-07-25 Carl Love <cel@us.ibm.com>
15338 * doc/extend.texi: Update the built-in documentation file for the
15339 existing built-in functions
15340 vector signed char vec_cnttz (vector signed char);
15341 vector unsigned char vec_cnttz (vector unsigned char);
15342 vector signed short vec_cnttz (vector signed short);
15343 vector unsigned short vec_cnttz (vector unsigned short);
15344 vector signed int vec_cnttz (vector signed int);
15345 vector unsigned int vec_cnttz (vector unsigned int);
15346 vector signed long long vec_cnttz (vector signed long long);
15347 vector unsigned long long vec_cnttz (vector unsigned long long);
15349 2017-07-25 Andrew Pinski <apinski@cavium.com>
15351 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15352 accesses where the use is for the first operand of a BIT_INSERT.
15354 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15357 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15358 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15360 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15362 * config/i386/gstabs.h: Delete.
15363 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15365 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15367 * config/i386/i386.c (ix86_decompose_address): Do not check for
15368 register RTX when looking at index_reg or base_reg.
15369 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15371 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15373 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15374 to update EH info here.
15376 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15378 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15380 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15382 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15384 2017-07-25 Torsten Duwe <duwe@suse.de>
15386 * common.opt: Introduce -fpatchable-function-entry
15387 command line option, and its variables function_entry_patch_area_size
15388 and function_entry_patch_area_start.
15389 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15390 including a two-value parser.
15391 * target.def (print_patchable_function_entry): New target hook.
15392 * targhooks.h (default_print_patchable_function_entry): New function.
15393 * targhooks.c (default_print_patchable_function_entry): Likewise.
15394 * toplev.c (process_options): Switch off IPA-RA if
15395 patchable function entries are being generated.
15396 * varasm.c (assemble_start_function): Look at the
15397 patchable-function-entry command line switch and current
15398 function attributes and maybe generate NOP instructions by
15399 calling the print_patchable_function_entry hook.
15400 * doc/extend.texi: Document patchable_function_entry attribute.
15401 * doc/invoke.texi: Document -fpatchable_function_entry
15402 command line option.
15403 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15405 * doc/tm.texi: Re-generate.
15407 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15410 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15411 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15413 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15415 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15417 2017-07-25 Richard Biener <rguenther@suse.de>
15419 PR tree-optimization/81455
15420 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15421 not walk in cycles when looking for guards.
15423 2017-07-25 Richard Biener <rguenther@suse.de>
15425 PR tree-optimization/81529
15426 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15427 when optimizing backedge uses.
15429 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15431 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15433 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15434 to dl_section_ref. On AIX, append an expression to subtract
15435 the size of the section length to dl_section_ref.
15437 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15439 * configure.ac: If any of the config.* scripts fail, exit 1.
15440 * configure: Regenerate.
15442 2017-07-25 Richard Biener <rguenther@suse.de>
15444 PR middle-end/81546
15445 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15446 of immediate uses, be more verbose on errors.
15448 2017-07-25 Richard Biener <rguenther@suse.de>
15450 PR tree-optimization/81510
15451 * tree-vect-loop.c (vect_is_simple_reduction): When the
15452 reduction stmt is not inside the loop bail out.
15454 2017-07-25 Richard Biener <rguenther@suse.de>
15456 PR tree-optimization/81303
15457 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15458 profitability check against LOOP_VINFO_NITERSM1.
15460 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15462 * domwalk.c (cmp_bb_postorder): Simplify.
15463 (sort_bbs_postorder): New function. Use it...
15464 (dom_walker::walk): ...here to optimize common cases.
15466 2017-07-25 Martin Liska <mliska@suse.cz>
15469 * ipa-visibility.c (function_and_variable_visibility): Make the
15470 redirection just on target that supports aliasing.
15471 Fix GNU coding style.
15473 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15476 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15478 * config/arm/rtems.h: Likewise.
15479 * config/bfin/rtems.h: Likewise.
15480 * config/i386/rtemself.h: Likewise.
15481 * config/lm32/rtems.h: Likewise.
15482 * config/m32c/rtems.h: Likewise.
15483 * config/m68k/rtemself.h: Likewise.
15484 * config/microblaze/rtems.h: Likewise.
15485 * config/mips/rtems.h: Likewise.
15486 * config/moxie/rtems.h: Likewise.
15487 * config/nios2/rtems.h: Likewise.
15488 * config/powerpcspe/rtems.h: Likewise.
15489 * config/rs6000/rtems.h: Likewise.
15490 * config/rtems.h: Likewise.
15491 * config/sh/rtems.h: Likewise.
15492 * config/sh/rtemself.h: Likewise.
15493 * config/sparc/rtemself.h: Likewise.
15495 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15498 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15499 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15500 * tree-ssa-structalias.c (alias_get_name): Same.
15502 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15505 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15506 instructions if no du chain is found.
15508 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15510 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15512 2017-07-25 Richard Biener <rguenther@suse.de>
15514 PR middle-end/81505
15515 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15518 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15520 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15521 upper-regs options.
15522 (ISA_2_7_MASKS_SERVER): Likewise.
15523 (ISA_3_0_MASKS_IEEE): Likewise.
15524 (OTHER_P8_VECTOR_MASKS): Likewise.
15525 (OTHER_VSX_VECTOR_MASKS): Likewise.
15526 (POWERPC_MASKS): Likewise.
15527 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15528 duplicate list of options.
15529 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15530 explicit -mupper-regs options.
15531 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15532 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15533 alias for -mupper-regs-df.
15534 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15535 (rs6000_init_hard_regno_mode_ok): Likewise.
15536 (rs6000_option_override_internal): Likewise.
15537 (rs6000_opt_masks): Likewise.
15538 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15539 options in terms of whether -mvsx or -mpower8-vector was used.
15540 (TARGET_UPPER_REGS_DI): Likewise.
15541 (TARGET_UPPER_REGS_SF): Likewise.
15542 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15543 -mupper-regs-* options.
15545 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15547 * passes.c (emergency_dump_function): Print some empty lines and a
15548 header before the RTL dump.
15550 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15552 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15554 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15557 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15558 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15560 2017-07-24 Carl Love <cel@us.ibm.com>
15562 * config/rs6000/rs6000-c.c: Add support for built-in functions
15563 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15564 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15565 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15566 vec_extract_fp_from_shortl): Add defines for the two builtins.
15567 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15568 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15570 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15571 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15572 * doc/extend.texi: Update the built-in documentation file for the
15573 new built-in function.
15575 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15578 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15580 * doc/generic.texi: Likewise.
15581 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15582 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15584 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15586 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15587 (aarch64_mls_elt_merge<mode>): Likewise.
15589 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15591 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15592 having __cxa_atexit.
15594 2017-07-23 Michael Collison <michael.collison@arm.com>
15596 * config/arm/arm.c (arm_option_override): Deprecate
15597 use of -mstructure-size-boundary.
15598 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15599 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15601 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15604 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15605 Reduce cost estimate for direct moves.
15607 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15610 * config/i386/i386.c (ix86_option_override_internal): Disable
15611 BMI, BMI2 and TBM instructions for -m16.
15613 2017-07-21 Carl Love <cel@us.ibm.com>
15615 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15616 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15617 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15618 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15619 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15620 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15621 VMULOSW): New enum "unspec" values.
15622 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15623 altivec_vmulosw): New patterns.
15624 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15625 VMULOSW): Add definitions.
15627 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15629 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15630 (qdf24xx): Likewise.
15631 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15632 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15633 (AARCH64_FL_V8_1): Renumber.
15634 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15635 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15636 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15637 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15638 rdma to feature modifiers list.
15640 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15642 PR middle-end/56727
15643 * ipa-visibility (function_and_variable_visibility): Convert
15644 recursive PLT call to direct call if appropriate.
15646 2017-07-21 Andrew Pinski <apinski@cavium.com>
15648 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15649 operand 1 to see if the types precision matches.
15650 * fold-const.c (operand_equal_p): Likewise.
15652 2017-07-21 Richard Biener <rguenther@suse.de>
15654 PR tree-optimization/81303
15655 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15656 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15657 (vect_peeling_hash_get_lowest_cost): Adjust.
15658 (vect_enhance_data_refs_alignment): Likewise. Use
15659 vect_get_peeling_costs_all_drs to compute the penalty for no
15660 peeling to match up costs.
15662 2017-07-21 Richard Biener <rguenther@suse.de>
15664 PR tree-optimization/81500
15665 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15666 we didn't identify a reduction path.
15668 2017-07-21 Tom de Vries <tom@codesourcery.com>
15669 Cesar Philippidis <cesar@codesourcery.com>
15671 PR gcov-profile/81442
15672 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15675 2017-07-21 Tom de Vries <tom@codesourcery.com>
15678 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15680 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15681 nvptx_override_options_after_change.
15683 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15685 * dwarf2out.c (output_file_names): Avoid double testing for
15686 dwarf_version >= 5.
15688 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15690 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15692 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15694 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15696 (try_crossjump_to_edge): Do not punt on partitioned functions.
15698 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15700 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15701 Put all BBs reachable only via paths crossing cold region to cold
15703 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15705 2016-07-21 Richard Biener <rguenther@suse.de>
15707 PR tree-optimization/81303
15708 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15709 into account prologue and epilogue iterations when raising
15710 min_profitable_iters to sth at least covering one vector iteration.
15712 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15714 * config/arm/arm.c (arm_test_cpu_arch_dat):
15717 2017-07-20 Nathan Sidwell <nathan@acm.org>
15719 Remove TYPE_METHODS.
15720 * tree.h (TYPE_METHODS): Delete.
15721 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15722 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15723 (dbxout_type_methods): Scan TYPE_FIELDS.
15724 (dbxout_type): Don't check TYPE_METHODS here.
15725 * function.c (use_register_for_decl): Always ignore register for
15726 class types when not optimizing.
15727 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15728 * tree.c (free_lang_data_in_type): Stitch out member functions and
15729 templates from TYPE_FIELDS.
15730 (build_distinct_type_copy, verify_type_variant,
15731 verify_type): Member fns are on TYPE_FIELDS.
15732 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15733 * tree-pretty-print.c (dump_generic_node): Likewise.
15735 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15738 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15739 V2TImode and V4TImode.
15740 (ix86_expand_vector_extract): Likewise.
15741 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15742 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15743 (ssescalarmode): Handle V4TImode and V2TImode.
15744 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15745 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15746 (VEXTRACTI128_MODE): New mode iterator.
15747 (splitter for *vec_extractv?ti first element): New.
15748 (VEC_INIT_MODE): New mode iterator.
15749 (vec_init<mode>): Consolidate 3 expanders into one using
15750 VEC_INIT_MODE mode iterator.
15752 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15754 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15755 non_spilled_static_chain_regno_p.
15757 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15759 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15761 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15763 * bb-reorder.c (connect_traces): Allow copying of blocks within
15766 2017-07-20 Richard Biener <rguenther@suse.de>
15768 * gimple.h (gimple_phi_result): Add gphi * overload.
15769 (gimple_phi_result_ptr): Likewise.
15770 (gimple_phi_arg): Likewise. Adjust index assert to only
15771 allow actual argument accesses rather than all slots available
15773 (gimple_phi_arg_def): Add gphi * overload.
15774 * tree-phinodes.c (make_phi_node): Initialize only actual
15776 (resize_phi_node): Clear memory not covered by old node,
15777 do not initialize excess argument slots.
15778 (reserve_phi_args_for_new_edge): Initialize new argument slot
15781 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15783 PR tree-optimization/81388
15785 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15787 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15788 by removing computation of may_be_zero.
15790 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15791 Tom de Vries <tom@codesourcery.com>
15793 PR middle-end/81030
15794 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15795 when gimple level profile disagrees with what RTL expander did.
15797 2017-07-20 Richard Biener <rguenther@suse.de>
15799 PR tree-optimization/61171
15800 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15801 (vect_analyze_stmt): Add slp instance parameter.
15802 (vectorizable_reduction): Likewise.
15803 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15804 (vect_is_simple_reduction): Deal with chains not detected
15805 as SLP reduction chain, specifically not properly associated
15806 chains containing a mix of plus/minus.
15807 (get_reduction_op): Remove.
15808 (get_initial_defs_for_reduction): Simplify, pass in whether
15809 this is a reduction chain, pass in the SLP node for the PHIs.
15810 (vect_create_epilog_for_reduction): Get the SLP instance as
15812 (vectorizable_reduction): Get the SLP instance as arg.
15813 During analysis remember the SLP node with the PHIs in the
15814 instance. Simplify getting at the vectorized reduction PHIs.
15815 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15816 through SLP instance.
15817 (vect_slp_analyze_operations): Likewise.
15818 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15819 (vect_transform_stmt): Likewise.
15821 2017-07-20 Tom de Vries <tom@codesourcery.com>
15823 PR tree-optimization/81489
15824 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15825 read of phi arg location to before loop that modifies phi.
15827 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15829 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15832 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15834 PR middle-end/81331
15835 * except.c (execute): Fix ordering issue.
15837 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15839 PR rtl-optimization/81423
15840 * combine.c (make_compound_operation_int): Don't try to optimize
15841 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15843 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15845 PR rtl-optimization/81423
15846 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15847 with a constant that is -1 in the truncated to mode.
15849 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15851 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15852 (determine_unlikely_bbs): ... here.
15853 * predict.h (propagate_unlikely_bbs_forward): Declare.
15854 * cfgexpand.c (pass_expand::execute): Use it.
15855 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15857 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15858 propagate_unlikely_bbs_forward.
15860 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15862 PR middle-end/81331
15863 * except.c (maybe_add_nop_after_section_switch): New function.
15866 2017-07-19 Tom de Vries <tom@codesourcery.com>
15868 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15870 2017-07-19 Tom de Vries <tom@codesourcery.com>
15872 * gimple.h (gimple_phi_arg): Make assert more strict.
15874 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15876 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15877 * config/rs6000/mmintrin.h: New file.
15878 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15880 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15882 PR tree-optimization/81346
15883 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15885 2017-07-19 Tom de Vries <tom@codesourcery.com>
15887 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15889 2017-07-19 Tom de Vries <tom@codesourcery.com>
15891 * config/nvptx/nvptx-modes.def: Add V2DImode.
15892 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15893 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15894 (nvptx_output_mov_insn): Handle lack of mov.b128.
15895 (nvptx_print_operand): Handle 'H' and 'L' codes.
15896 (nvptx_vector_mode_supported): Allow V2DImode.
15897 (nvptx_preferred_simd_mode): New function.
15898 (nvptx_data_alignment): New function.
15899 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15900 nvptx_preferred_simd_mode.
15901 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15903 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15905 2017-07-19 Tom de Vries <tom@codesourcery.com>
15907 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15908 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15909 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15910 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15911 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15912 (mov<VECIM>_insn): New define_insn.
15913 (define_expand "mov<VECIM>): New define_expand.
15915 2017-07-19 Tom de Vries <tom@codesourcery.com>
15917 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15919 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15921 PR tree-optimization/81346
15922 * fold-const.h (fold_div_compare, range_check_type): Declare.
15923 * fold-const.c (range_check_type): New function.
15924 (build_range_check): Use range_check_type.
15925 (fold_div_compare): No longer static, rewritten into
15926 a match.pd helper function.
15927 (fold_comparison): Don't call fold_div_compare here.
15928 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15929 as helper function.
15931 2017-07-19 Nathan Sidwell <nathan@acm.org>
15933 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15934 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15935 * tree.c (find_decls_types_r, verify_type): Use
15936 TYPE_{MIN,MAX}_VALUE_RAW.
15937 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15938 (hash_tree): Likewise.
15939 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15941 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15944 2017-07-18 Tom de Vries <tom@codesourcery.com>
15946 PR middle-end/81464
15947 * omp-expand.c (expand_omp_for_static_chunk): Handle
15948 equal-argument loop exit phi.
15950 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15953 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15954 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15955 operand 2 predicate.
15956 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15957 operand 2 predicate.
15958 (ror,rol -> rorx splitters): Use const_int_operand as
15959 operand 2 predicate.
15961 2017-06-18 Richard Biener <rguenther@suse.de>
15963 PR tree-optimization/81410
15964 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15965 the gap in the ! slp_perm SLP case after each group.
15967 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15969 PR middle-end/81463
15970 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15973 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15975 PR middle-end/81462
15976 * predict.c (set_even_probabilities): Cleanup; do not affect
15977 probabilities that are already known.
15978 (combine_predictions_for_bb): Call even when count is set.
15980 2017-07-18 Nathan Sidwell <nathan@acm.org>
15982 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15985 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15988 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15989 optimization for loop niter analysis.
15991 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15994 * config/avr/avr.c (avr_optimize_casesi): Don't use
15995 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15997 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15999 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16000 body_cost_vec from _vect_peel_extended_info.
16001 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16002 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16005 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16007 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16009 2017-07-18 Richard Biener <rguenther@suse.de>
16011 PR tree-optimization/80620
16012 PR tree-optimization/81403
16013 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16014 info when re-using a VN table entry.
16016 2017-07-18 Richard Biener <rguenther@suse.de>
16018 PR tree-optimization/81418
16019 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16020 vectype_in. Verify that with lane-reducing reduction operations
16021 we have a single def-use cycle.
16023 2017-07-17 Carl Love <cel@us.ibm.com>
16025 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16027 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16028 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16029 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16030 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16031 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16032 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16033 VMULOSW): New enum "unspec" values.
16034 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16035 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16036 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16037 altivec_vmulosw): New patterns.
16038 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16039 VMULOSW): Add definitions.
16040 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16042 * config/alpha/alpha.c: Include predict.h.
16044 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16046 * tree-vrp.c (compare_assert_loc): Fix comparison function
16047 to return predictable results.
16049 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16051 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16053 (subdi3): Likewise.
16054 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16055 * doc/invoke.texi (mexpand-adddi): Update text.
16057 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16059 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16060 that also clobbers the CC register. The old expand code is moved
16062 (*arc_clzsi2): ... here.
16063 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16064 the CC register. The old expand code is moved to ...
16065 (arc_ctzsi2): ... here.
16067 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16069 * config/arc/arc.opt (mindexed-loads): Use initial value
16070 TARGET_INDEXED_LOADS_DEFAULT.
16071 (mauto-modify-reg): Use initial value
16072 TARGET_AUTO_MODIFY_REG_DEFAULT.
16073 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16074 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16075 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16076 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16078 2017-07-17 Martin Liska <mliska@suse.cz>
16081 * opts.c (finish_options): Do not allow -fgnu-tm
16082 w/ -fsanitize={kernel-,}address. Say sorry.
16084 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16087 * tree-loop-distribution.c (classify_partition): Only assert on
16088 numer of iterations.
16089 (merge_dep_scc_partitions): Delete prameter. Update function call.
16090 (distribute_loop): Remove code handling loop with unknown niters.
16091 (pass_loop_distribution::execute): Skip loop with unknown niters.
16093 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16096 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16097 function sort_partitions_by_post_order.
16099 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16101 PR tree-optimization/81374
16102 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16103 the max index of basic blocks, rather than number of basic blocks.
16105 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16107 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16109 (arc_legitimate_pic_operand_p): Likewise.
16110 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16112 (arc_needs_pcl_p): Likewise.
16113 (arc_legitimate_pc_offset_p): Likewise.
16114 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16115 function is also used in constrains.md.
16116 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16117 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16118 PLUS. Only return true/false in known cases, otherwise assert.
16119 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16120 is already called in arc_legitimate_constant_p.
16121 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16123 (LEGITIMATE_PIC_OPERAND_P): Use
16124 arc_raw_symbolic_reference_mentioned_p function.
16125 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16130 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16131 Andrew Burgess <andrew.burgess@embecosm.com>
16133 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16134 (arc_return_address_register): New function.
16135 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16136 (arc_handle_fndecl_attribute): Add naked attribute.
16137 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16138 (TARGET_WARN_FUNC_RETURN): Likewise.
16139 (arc_allocate_stack_slots_for_args): New function.
16140 (arc_warn_func_return): Likewise.
16141 (machine_function): Change type fn_type.
16142 (arc_compute_function_type): Consider new naked function type,
16143 change function return type.
16144 (arc_must_save_register): Adapt to handle new
16145 arc_compute_function_type's return type.
16146 (arc_expand_prologue): Likewise.
16147 (arc_expand_epilogue): Likewise.
16148 (arc_return_address_regs): Delete.
16149 (arc_return_address_register): New function.
16150 (arc_epilogue_uses): Use above function.
16151 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16152 (arc_function_type): Change encoding, add naked type.
16153 (ARC_INTERRUPT_P): Change to handle the new encoding.
16154 (ARC_FAST_INTERRUPT_P): Likewise.
16155 (ARC_NORMAL_P): Define.
16156 (ARC_NAKED_P): Likewise.
16157 (arc_compute_function_type): Delete prototype.
16158 * config/arc/arc.md (in_ret_delay_slot): Use
16159 arc_return_address_register function.
16160 (simple_return): Likewise.
16161 (p_return_i): Likewise.
16163 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16165 PR tree-optimization/81428
16166 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16167 can't be built for those types.
16169 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16171 Remove stuff dead since r239246.
16173 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16174 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16175 (avr_inform_devices): Remove dead stuff.
16177 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16179 * config/arm/arm_neon.h: Fix softp typo.
16181 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16183 PR tree-optimization/81365
16184 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16185 aggregate moves onto bb predecessor edges, make sure there are no
16186 loads that could alias the lhs in between the start of bb and the
16189 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16192 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16193 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16194 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16196 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16198 PR tree-optimization/81396
16199 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16200 (init_symbolic_number): Initialize it to 1.
16201 (perform_symbolic_merge): Add n_ops from both operands into the new
16203 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16204 without base_addr as useless if they need more than one operation.
16205 (bswap_replace): Handle !bswap case for NULL base_addr.
16207 2017-07-17 Tom de Vries <tom@codesourcery.com>
16210 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16213 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16215 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16216 conditional builtin define __FIX_LEON3FT_B2BST.
16218 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16220 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16221 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16224 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16226 PR rtl-optimization/81424
16227 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16228 to remove potential trapping from operands if -fnon-call-exceptions.
16230 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16232 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16233 profile_proability for scalling.
16234 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16236 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16238 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16240 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16242 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16243 fixpoint arithmetics.
16245 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16247 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16248 fixpoint arithmetics.
16250 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16252 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16253 fixpoint arithmetics.
16255 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16257 * profile-count.h (profile_probability::from_reg_br_prob_note,
16258 profile_probability::to_reg_br_prob_note): New functions.
16259 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16260 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16261 * predict.c (probability_reliable_p): Update.
16262 (edge_probability_reliable_p): Update.
16263 (br_prob_note_reliable_p): Update.
16264 (invert_br_probabilities): Update.
16265 (add_reg_br_prob_note): New function.
16266 (combine_predictions_for_insn): Update.
16267 * asan.c (asan_clear_shadow): Update.
16268 * cfgbuild.c (compute_outgoing_frequencies): Update.
16269 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16270 (update_br_prob_note): Update.
16271 (rtl_verify_edges): Update.
16272 (purge_dead_edges): Update.
16273 (fixup_reorder_chain): Update.
16274 * emit-rtl.c (try_split): Update.
16275 * ifcvt.c (cond_exec_process_insns): Update.
16276 (cond_exec_process_if_block): Update.
16277 (dead_or_predicable): Update.
16278 * internal-fn.c (expand_addsub_overflow): Update.
16279 (expand_neg_overflow): Update.
16280 (expand_mul_overflow): Update.
16281 * loop-doloop.c (doloop_modify): Update.
16282 * loop-unroll.c (compare_and_jump_seq): Update.
16283 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16284 * predict.h: Update.
16285 * reorg.c (mostly_true_jump): Update.
16287 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16288 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16289 * config/arc/arc.c: (emit_unlikely_jump): Update.
16290 * config/arm/arm.c: (emit_unlikely_jump): Update.
16291 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16292 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16293 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16294 (ix86_print_operand): Update.
16295 (ix86_split_fp_branch): Update.
16296 (predict_jump): Update.
16297 * config/ia64/ia64.c (ia64_print_operand): Update.
16298 * config/mmix/mmix.c (mmix_print_operand): Update.
16299 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16300 (rs6000_expand_split_stack_prologue): Update.
16301 * config/rs6000/rs6000.c: Update.
16302 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16303 (s390_expand_vec_movstr): Update.
16304 (s390_expand_cs_tdsi): Update.
16305 (s390_expand_split_stack_prologue): Update.
16306 * config/sh/sh.c (sh_print_operand): Update.
16307 (expand_cbranchsi4): Update.
16308 (expand_cbranchdi4): Update.
16309 * config/sparc/sparc.c (output_v9branch): Update.
16310 * config/spu/spu.c (get_branch_target): Update.
16311 (ea_load_store_inline): Update.
16312 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16313 * config/tilepro/tilepro.c: Update.
16315 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16317 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16318 (gimplify_save_expr): Likewise.
16320 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16322 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16324 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16326 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16328 * timevar.def (TV_IPA_FNSUMMARY): Define.
16330 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16332 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16333 to back store errata sensitive sequence from being generated.
16334 (sqrtdf2_fix): Likewise.
16336 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16338 * tree-ssa-threadupdate.c (compute_path_counts,
16339 update_joiner_offpath_counts): Use profile_probability.
16341 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16344 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16346 * config/arm/arm-c.c (arm_cpu_builtins): Define
16347 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16349 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16351 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16352 array entries to represent __ieee128 versions of the
16353 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16354 scalar_extract_sig, and scalar_insert_exp built-in functions.
16355 (altivec_resolve_overloaded_builtin): Add special case handling
16356 for the __builtin_scalar_insert_exp function, as represented by
16357 the P9V_BUILTIN_VEC_VSIEDP constant.
16358 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16359 exponent support for __ieee128 argument.
16360 (VSESQP): Add scalar extract signature support for __ieee128
16362 (VSTDCNQP): Add scalar test negative support for __ieee128
16364 (VSIEQP): Add scalar insert exponent support for __int128 argument
16365 with __ieee128 result.
16366 (VSIEQPF): Add scalar insert exponent support for __ieee128
16367 argument with __ieee128 result.
16368 (VSTDCQP): Add scalar test data class support for __ieee128
16370 (VSTDCNQP): Add overload support for scalar test negative with
16371 __ieee128 argument.
16372 (VSTDCQP): Add overload support for scalar test data class
16373 __ieee128 argument.
16374 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16375 UNSPEC_VSX_SXSIGDP.
16376 (UNSPEC_VSX_SIEXPQP): New constant.
16377 (xsxexpqp): New insn for VSX scalar extract exponent quad
16379 (xsxsigqp): New insn for VSX scalar extract significand quad
16381 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16382 precision with floating point argument.
16383 (xststdcqp): New expand for VSX scalar test data class quad
16385 (xststdcnegqp): New expand for VSX scalar test negative quad
16387 (xststdcqp): New insn to match expansions for VSX scalar test data
16388 class quad precision and VSX scalar test negative quad precision.
16389 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16390 special case operand checking to enforce that second operand of
16391 VSX scalar test data class with quad precision argument is a 7-bit
16393 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16394 prototypes and descriptions of __ieee128 versions of
16395 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16396 scalar_test_data_class, and scalar_test_neg built-in functions.
16398 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16400 PR tree-optimization/81162
16401 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16402 replace a negate with an add.
16404 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16406 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16408 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16410 * config/arm/arm-c.c (arm_cpu_builtins): Define
16411 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16413 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16415 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16416 (armv8-r): Set ARM Cortex-R52 as default CPU.
16417 * config/arm/arm-tables.opt: Regenerate.
16418 * config/arm/arm-tune.md: Regenerate.
16419 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16421 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16422 extension for -mcpu=cortex-r52.
16424 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16426 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16427 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16428 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16429 (fp-armv8): Define it as FP_ARMv8 only.
16430 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16431 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16433 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16434 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16435 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16436 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16437 than TARGET_FPU_ARMV8.
16438 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16439 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16440 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16442 * config/arm/neon.md (neon_vrint): Likewise.
16443 (neon_vcvt): Likewise.
16444 (neon_<fmaxmin_op><mode>): Likewise.
16445 (<fmaxmin><mode>3): Likewise.
16446 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16447 * config/arm/predicates.md (arm_cond_move_operator): Check against
16448 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16450 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16452 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16453 to top of function.
16455 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16457 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16458 loop in comment with memset.
16460 2017-07-14 Martin Liska <mliska@suse.cz>
16462 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16463 * dwarf2out.c (is_java): Remove the function.
16464 (output_pubname): Remove usage of the function.
16465 (lower_bound_default): Remove usage of DW_LANG_Java.
16466 (gen_compile_unit_die): Likewise.
16467 * gcc.c: Remove compiler defaults for .java and .zip files.
16468 * gimple-expr.c (remove_suffix): Change as there's no longer
16469 extension than 4-letter one.
16470 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16471 (gimplify_save_expr): Likewise.
16472 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16473 as it's possible even for other languages than Java.
16474 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16475 * lto-opts.c (lto_write_options): Remove reference to Java.
16476 * opts.c (strip_off_ending): Update file extension handling.
16477 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16478 * tree-eh.c (lower_resx): Likewise.
16479 * tree.c (free_lang_data_in_type): Remove dead code.
16480 (find_decls_types_r): Likewise.
16481 (build_common_builtin_nodes): Remove Java from a comment.
16482 (verify_type): Remove dead code.
16483 * varasm.c (assemble_external): Remove Java from a comment.
16485 2017-07-14 Martin Liska <mliska@suse.cz>
16487 * opts.c (finish_options): Add quotes.
16488 (common_handle_option): Likewise.
16490 2017-07-14 Martin Liska <mliska@suse.cz>
16492 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16493 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16494 Remove N_SO_PASCAL.
16495 * dwarf2out.c (lower_bound_default): Do not handle
16497 (gen_compile_unit_die): Likewise.
16498 * gcc.c: Remove default extension binding for GNU Pascal.
16499 * stmt.c: Remove Pascal language from a comment.
16500 * xcoffout.c: Likewise.
16502 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16505 * diagnostic-show-locus.c (fixit_cmp): New function.
16506 (layout::layout): Sort m_fixit_hints.
16507 (column_range::column_range): Assert that the values are valid.
16508 (struct char_span): New struct.
16509 (correction::overwrite): New method.
16510 (struct source_line): New struct.
16511 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16512 calls in terms of classes source_line and char_span, and
16513 correction::overwrite.
16514 (selftest::test_overlapped_fixit_printing_2): New function.
16515 (selftest::diagnostic_show_locus_c_tests): Call it.
16517 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16519 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16520 early if there is no lhs.
16522 2017-07-13 Martin Liska <mliska@suse.cz>
16524 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16525 (gen_reference_type_die): Likewise.
16526 * stor-layout.c: Remove Pascal-related comment.
16528 2017-07-13 Martin Liska <mliska@suse.cz>
16530 * opts.c (finish_options): Add quotes to error messages.
16531 (parse_sanitizer_options): Likewise.
16533 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16535 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16537 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16539 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16541 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16543 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16545 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16547 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16550 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16551 provides the hardware capability bits, define the macro
16552 __BUILTIN_CPU_SUPPORTS__.
16553 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16554 if GLIBC does not provide the hardware capability bits. Add a
16555 gcc_unreachable call if the built-in cpu function is neither
16556 __builtin_cpu_is nor __builtin_cpu_supports.
16557 (rs6000_get_function_versions_dispatcher): Change the warning
16558 that an old GLIBC is used which does not export the capability
16559 bits to be an error.
16560 * doc/extend.texi (target_clones attribute): Document the
16561 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16562 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16563 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16564 the macros defined by GCC if the newer GLIBC is available.
16566 2017-07-12 Jeff Law <law@redhat.com>
16568 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16569 remaining includes slightly.
16570 * config/riscv/riscv-builtins.c: Include profile-count.h.
16572 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16575 * config/avr/avr.c (avr_set_current_function): In diagnostic
16576 messages: Quote keywords and (parts of) identifiers.
16577 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16580 2017-07-12 Carl Love <cel@us.ibm.com>
16582 * config/rs6000/rs6000-c.c: Add support for built-in functions
16583 vector bool char vec_revb (vector bool char);
16584 vector bool short vec_revb (vector short char);
16585 vector bool int vec_revb (vector bool int);
16586 vector bool long long vec_revb (vector bool long long);
16587 * doc/extend.texi: Update the built-in documentation file for the
16588 new built-in functions.
16590 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16592 * config/s390/s390.md: Remove movcc splitter.
16594 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16596 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16597 load/store on condition.
16599 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16602 * config/avr/avr.c (avr_encode_section_info)
16603 [progmem && !TREE_READONLY]: Error if progmem object needs
16606 2017-07-11 Michael Collison <michael.collison@arm.com>
16608 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16611 2017-07-11 Carl Love <cel@us.ibm.com>
16613 * config/rs6000/rs6000-c.c: Add support for builtins
16614 vector unsigned int vec_parity_lsbb (vector signed int);
16615 vector unsigned int vec_parity_lsbb (vector unsigned int);
16616 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16617 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16618 vector unsigned long long vec_parity_lsbb (vector signed long long);
16619 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16620 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16621 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16622 * doc/extend.texi: Update the built-in documentation file for the
16623 new built-in functions.
16625 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16627 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16628 (layout::m_primary_loc): New field.
16629 (layout::layout): Initialize new field. Move location filtering
16630 logic from here to...
16631 (layout::maybe_add_location_range): ...this new method. Add
16632 support for filtering to just the lines already specified by other
16634 (layout::will_show_line_p): New method.
16635 (gcc_rich_location::add_location_if_nearby): New method.
16636 (selftest::test_add_location_if_nearby): New test function.
16637 (selftest::diagnostic_show_locus_c_tests): Call it.
16638 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16641 2017-07-11 Tom de Vries <tom@codesourcery.com>
16643 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16644 (bb_first_real_insn): New function.
16645 (nvptx_single): Add extra initialization of broadcasted condition
16648 2017-07-11 Nathan Sidwell <nathan@acm.org>
16650 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16652 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16654 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16655 Binutils doc as TEXI will mess them up.
16656 * doc/invoke.texi (AVR Options): Same here.
16658 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16660 * config/sparc/sparc.opt (mfix-ut700): New option.
16661 (mfix-gr712rc): Likewise.
16662 (sparc_fix_b2bst): New variable.
16663 * doc/invoke.texi (SPARC options): Document them.
16664 (ARM options): Fix warnings.
16665 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16666 instructions to prevent sequences that can trigger the store-store
16667 errata for certain LEON3FT processors.
16668 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16669 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16670 * config/sparc/sparc.md (fix_b2bst): New attribute.
16671 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16673 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16676 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16678 (*rsqrtsf2_sse): Ditto.
16680 (div<mode>3): Macroize insn from divdf3 and divsf3
16681 using MODEF mode iterator.
16683 2017-07-10 Martin Sebor <msebor@redhat.com>
16685 PR tree-optimization/80397
16686 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16687 instead of testing for equality to INTEGER_TYPE.
16689 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16691 * config.gcc: Remove uclibc from arc target spec.
16693 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16695 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16697 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16700 * lto-wrapper.c (remove_option): New function.
16701 (merge_and_complain): Merge PIC/PIE options more realistically.
16703 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16705 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16709 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16710 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16711 * config.in: Regenerate.
16712 * configure: Regenerate.
16713 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16714 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16715 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16716 (TARGET_GASISR_PROLOGUES): ...target mask.
16717 * common/config/avr/avr-common.c
16718 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16719 Set -mgas-isr-prologues.
16720 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16721 INSERT_PASS_BEFORE for it.
16722 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16723 * config/avr/avr.c (avr_option_override)
16724 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16725 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16726 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16727 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16728 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16730 (make_avr_pass_pre_proep): New function.
16731 (emit_push_sfr) <treg>: Add argument to function and use it
16732 instead of TMP_REG.
16733 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16734 and set machine->gasisr.yes.
16735 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16736 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16737 __gcc_isr.n_pushed to .L__stack_usage.
16738 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16739 (avr_asm_final_postscan_insn): ...this new static function.
16740 * config/avr/avr.h (machine_function)
16741 <is_no_gccisr, use_L__stack_usage>: New fields.
16742 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16743 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16744 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16745 (gasisr, *gasisr): New expander and insn.
16746 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16747 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16748 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16750 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16752 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16755 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16757 Move jump-tables out of .text again.
16760 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16761 (ASM_OUTPUT_ADDR_VEC): New function.
16762 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16763 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16764 INSN_ADDRESSes as asm comment.
16765 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16766 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16767 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16768 * config/avr/avr.md (*tablejump): Adjust comment.
16769 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16770 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16772 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16773 (avr_output_addr_vec): New proto.
16774 (avr_log_t) <insn_addresses>: New field.
16776 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16779 * config/i386/i386.c (ix86_function_arg_advance): Set
16780 outgoing_args_on_stack to true if there are outgoing arguments
16782 (ix86_function_arg): Likewise.
16783 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16784 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16785 * config/i386/i386.h (machine_function): Add
16786 outgoing_args_on_stack.
16788 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16790 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16791 supporting pthreds.
16792 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16794 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16796 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16797 (REAL_H): Remove $(MACHMODE_H).
16798 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16800 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16801 $(MACHMODE_H) and double-int.h.
16802 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16804 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16807 2017-07-07 Andrew Pinski <apinski@cavium.com>
16809 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16810 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16812 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16814 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16815 Add warning if GCC was not configured to link against a GLIBC that
16816 exports the hardware capability bits.
16817 (make_resolver_func): Make resolver function private and not a
16818 COMDAT function. Create the name with clone_function_name instead
16819 of make_unique_name.
16822 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16823 correct operand in doing the split.
16825 2017-07-07 Carl Love <cel@us.ibm.com>
16827 * config/rs6000/rs6000-c: Add support for built-in function
16828 vector unsigned short vec_pack_to_short_fp32 (vector float,
16830 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16831 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16832 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16833 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16834 (convert_4f32_8i16): Add define_expand.
16835 * doc/extend.texi: Update the built-in documentation file for the
16836 new built-in function.
16838 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16840 * config/sparc/m8.md: New file.
16841 * config/sparc/sparc.md: Include m8.md.
16843 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16845 * config/sparc/sparc.opt: New option -mvis4b.
16846 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16847 (sparc_option_override): Handle VIS4B.
16848 (enum sparc_builtins): Define
16849 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16850 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16851 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16852 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16853 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16854 (check_constant_argument): New function.
16855 (sparc_vis_init_builtins): Define builtins
16856 __builtin_vis_dictunpack{8,16,32},
16857 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16858 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16859 __builtin_vis_fpcmpde{8,16,32}shl and
16860 __builtin_vis_fpcmpur{8,16,32}shl.
16861 (sparc_expand_builtin): Check that the constant operands to
16862 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16863 constant and in range.
16864 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16866 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16867 (SPARC_IMM5_P): Likewise.
16868 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16869 (enabled): Handle vis4b.
16870 (UNSPEC_DICTUNPACK): New unspec.
16871 (UNSPEC_FPCMPSHL): Likewise.
16872 (UNSPEC_FPUCMPSHL): Likewise.
16873 (UNSPEC_FPCMPDESHL): Likewise.
16874 (UNSPEC_FPCMPURSHL): Likewise.
16875 (cpu_feature): New CPU feature `vis4b'.
16876 (dictunpack{8,16,32}): New insns.
16877 (FPCSMODE): New mode iterator.
16878 (fpcscond): New code iterator.
16879 (fpcsucond): Likewise.
16880 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16881 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16882 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16883 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16884 * config/sparc/constraints.md: Define constraints `q' for unsigned
16885 2-bit integer constants and `t' for unsigned 5-bit integer
16887 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16889 (imm5_operand_dictunpack16): Likewise.
16890 (imm5_operand_dictunpack32): Likewise.
16891 (imm2_operand): Likewise.
16892 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16893 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16894 ditunpack* and fpcmp*shl builtins.
16896 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16898 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16899 * config.in: Add HAVE_AS_SPARC6 define.
16900 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16902 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16904 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16905 (CPP_CPU_SPEC): Handle m8.
16906 (ASM_CPU_SPEC): Likewise.
16907 * config/sparc/sparc-opts.h (enum processor_type): Add
16909 * config/sparc/sparc.c (m8_costs): New struct.
16910 (sparc_option_override): Handle TARGET_CPU_m8.
16911 (sparc32_initialize_trampoline): Likewise.
16912 (sparc64_initialize_trampoline): Likewise.
16913 (sparc_issue_rate): Likewise.
16914 (sparc_register_move_cost): Likewise.
16915 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16916 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16917 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16918 (CPP_CPU_SPEC): Handle M8.
16919 (ASM_CPU_SPEC): Likewise.
16920 (AS_M8_FLAG): Define.
16921 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16922 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16923 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16925 * configure: Regenerate.
16926 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16929 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16931 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16933 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16934 ("*movdi_insn_sp32"): Do not set v3pipe.
16935 ("*movsi_insn"): Likewise.
16936 ("*movdi_insn_sp64"): Likewise.
16937 ("*movsf_insn"): Likewise.
16938 ("*movdf_insn_sp32"): Likewise.
16939 ("*movdf_insn_sp64"): Likewise.
16940 ("*zero_extendsidi2_insn_sp64"): Likewise.
16941 ("*sign_extendsidi2_insn"): Likewise.
16942 ("*mov<VM32:mode>_insn"): Likewise.
16943 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16944 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16945 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16946 ("<vlop:code><VL:mode>3"): Likewise.
16947 ("*not_<vlop:code><VL:mode>3"): Likewise.
16948 ("*nand<VL:mode>_vis"): Likewise.
16949 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16950 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16951 ("one_cmpl<VL:mode>2"): Likewise.
16952 ("faligndata<VM64:mode>_vis"): Likewise.
16953 ("alignaddrsi_vis"): Likewise.
16954 ("alignaddrdi_vis"): Likweise.
16955 ("alignaddrlsi_vis"): Likewise.
16956 ("alignaddrldi_vis"): Likewise.
16957 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16958 ("bmaskdi_vis"): Likewise.
16959 ("bmasksi_vis"): Likewise.
16960 ("bshuffle<VM64:mode>_vis"): Likewise.
16961 ("cmask8<P:mode>_vis"): Likewise.
16962 ("cmask16<P:mode>_vis"): Likewise.
16963 ("cmask32<P:mode>_vis"): Likewise.
16964 ("pdistn<P:mode>_vis"): Likewise.
16965 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16967 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16969 * config/sparc/sparc.md ("subtype"): New insn attribute.
16970 ("*wrgsr_sp64"): Set insn subtype.
16971 ("*rdgsr_sp64"): Likewise.
16972 ("alignaddrsi_vis"): Likewise.
16973 ("alignaddrdi_vis"): Likewise.
16974 ("alignaddrlsi_vis"): Likewise.
16975 ("alignaddrldi_vis"): Likewise.
16976 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16977 ("fexpand_vis"): Likewise.
16978 ("fpmerge_vis"): Likewise.
16979 ("faligndata<VM64:mode>_vis"): Likewise.
16980 ("bshuffle<VM64:mode>_vis"): Likewise.
16981 ("cmask8<P:mode>_vis"): Likewise.
16982 ("cmask16<P:mode>_vis"): Likewise.
16983 ("cmask32<P:mode>_vis"): Likewise.
16984 ("fchksm16_vis"): Likewise.
16985 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16986 ("fmean16_vis"): Likewise.
16987 ("fp<plusminus_insn>64_vis"): Likewise.
16988 ("<plusminus_insn>v8qi3"): Likewise.
16989 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16990 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16991 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16992 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16993 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16994 ("*movqi_insn"): Likewise.
16995 ("*movhi_insn"): Likewise.
16996 ("*movsi_insn"): Likewise.
16997 ("movsi_pic_gotdata_op"): Likewise.
16998 ("*movdi_insn_sp32"): Likewise.
16999 ("*movdi_insn_sp64"): Likewise.
17000 ("movdi_pic_gotdata_op"): Likewise.
17001 ("*movsf_insn"): Likewise.
17002 ("*movdf_insn_sp32"): Likewise.
17003 ("*movdf_insn_sp64"): Likewise.
17004 ("*zero_extendhisi2_insn"): Likewise.
17005 ("*zero_extendqihi2_insn"): Likewise.
17006 ("*zero_extendqisi2_insn"): Likewise.
17007 ("*zero_extendqidi2_insn"): Likewise.
17008 ("*zero_extendhidi2_insn"): Likewise.
17009 ("*zero_extendsidi2_insn_sp64"): Likewise.
17010 ("ldfsr"): Likewise.
17011 ("prefetch_64"): Likewise.
17012 ("prefetch_32"): Likewise.
17013 ("tie_ld32"): Likewise.
17014 ("tie_ld64"): Likewise.
17015 ("*tldo_ldub_sp32"): Likewise.
17016 ("*tldo_ldub1_sp32"): Likewise.
17017 ("*tldo_ldub2_sp32"): Likewise.
17018 ("*tldo_ldub_sp64"): Likewise.
17019 ("*tldo_ldub1_sp64"): Likewise.
17020 ("*tldo_ldub2_sp64"): Likewise.
17021 ("*tldo_ldub3_sp64"): Likewise.
17022 ("*tldo_lduh_sp32"): Likewise.
17023 ("*tldo_lduh1_sp32"): Likewise.
17024 ("*tldo_lduh_sp64"): Likewise.
17025 ("*tldo_lduh1_sp64"): Likewise.
17026 ("*tldo_lduh2_sp64"): Likewise.
17027 ("*tldo_lduw_sp32"): Likewise.
17028 ("*tldo_lduw_sp64"): Likewise.
17029 ("*tldo_lduw1_sp64"): Likewise.
17030 ("*tldo_ldx_sp64"): Likewise.
17031 ("*mov<VM32:mode>_insn"): Likewise.
17032 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17033 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17035 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17037 * config/sparc/sparc.md ("type"): New insn type viscmp.
17038 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17040 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17041 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17042 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17043 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17045 ("n7_vis_logical_11cycle"): Likewise.
17046 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17047 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17048 * config/sparc/niagara.md ("niag_vis"): Likewise.
17049 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17050 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17052 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17054 * config/sparc/sparc.md: New instruction type `bmask'.
17055 (bmaskdi_vis): Use the `bmask' type.
17056 (bmasksi_vis): Likewise.
17057 * config/sparc/ultra3.md (us3_array): Likewise.
17058 * config/sparc/niagara7.md (n7_array): Likewise.
17059 * config/sparc/niagara4.md (n4_array): Likewise.
17060 * config/sparc/niagara2.md (niag2_vis): Likewise.
17061 (niag3_vis): Likewise.
17062 * config/sparc/niagara.md (niag_vis): Likewise.
17064 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17066 * ipa-comdats.c: Remove optimize check from gate.
17067 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17068 for functions not optimized.
17069 (ipa_fn_summary_read): Skip optimize check.
17070 (ipa_fn_summary_write): Likewise.
17071 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17073 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17075 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17076 (check_callers): Give up on caller which is not optimized.
17077 (inline_small_functions): Likewise.
17078 (ipa_inline): Do not give up when not optimizing.
17079 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17080 away unoptimizes cdtors.
17081 (whole_program_function_and_variable_visibility): Do
17082 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17083 * ipa.c (process_references): Do not check optimize.
17084 (symbol_table::remove_unreachable_nodes): Update optimize check.
17085 (set_writeonly_bit): Update optimize check.
17086 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17087 (pass_ipa_single_use::gate): Remove.
17089 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17091 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17092 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17093 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17094 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17095 permute_load, permute_store, adjust_extract, adjust_splat,
17096 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17097 replace_swap_with_copy, dump_swap_insn_table,
17098 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17099 recombine_lvx_pattern, recombine_stvx_pattern,
17100 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17101 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17102 to file rs6000-p8swap.c.
17103 * config/rs6000/rs6000-p8swap.c: New file.
17104 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17105 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17106 and rs6000*-*-* targets.
17108 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17110 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17112 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17114 * lto-wrapper.c (merge_and_complain): Do not merge
17115 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17116 fsigned_zeros, ftrapping_math, fwrapv.
17117 (append_compiler_options): Do not track these options.
17118 (append_linker_options): Likewie
17120 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17122 * cgraphunit.c (cgraph_node::finalize_function): When
17123 !flag_toplevel_reorde set no_reorder flag.
17124 (varpool_node::finalize_decl): Likewise.
17125 (symbol_table::compile): Drop no toplevel reorder path.
17127 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17129 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17130 edges; zero probability is not better than uninitialized.
17132 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17134 * asan.h (asan_sanitize_allocas_p): Declare.
17135 * asan.c (asan_sanitize_allocas_p): New function.
17136 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17137 (handle_builtin_alloca): Likewise.
17138 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17139 if !asan_sanitize_allocas_p.
17140 * params.def (asan-instrument-allocas): Add new option.
17141 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17142 * opts.c (common_handle_option): Disable allocas sanitization for
17145 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17147 * asan.c: Include gimple-fold.h.
17148 (get_last_alloca_addr): New function.
17149 (handle_builtin_stackrestore): Likewise.
17150 (handle_builtin_alloca): Likewise.
17151 (asan_emit_allocas_unpoison): Likewise.
17152 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17153 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17154 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17155 (instrument_builtin_call): Pass gimple iterator to
17156 get_mem_refs_of_builtin_call.
17157 (last_alloca_addr): New global.
17158 * asan.h (asan_emit_allocas_unpoison): Declare.
17159 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17160 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17161 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17162 if function calls alloca.
17163 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17164 * gimple-fold.h (replace_call_with_value): Declare.
17165 * internal-fn.c: Include asan.h.
17166 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17167 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17169 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17171 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17172 (C_SELFTEST_FLAGS): New.
17173 (CPP_SELFTEST_FLAGS): New.
17174 (SELFTEST_DEPS): New, from deps of s-selftest.
17175 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17176 (CPP_SELFTEST_DEPS): New.
17177 (selftest): Add dependency on s-selftest-c++.
17178 (s-selftest): Rename to...
17179 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17180 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17181 than SELFTEST_FLAGS.
17182 (selftest-gdb): Rename to...
17183 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17185 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17186 (selftest-valgrind): Rename to...
17187 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17189 (selftest-valgrind): Reintroduce as an alias for
17190 selftest-c-valgrind.
17191 (s-selftest-c++): New.
17192 (selftest-c++-gdb): New.
17193 (selftest-c++-valgrind): New.
17195 2017-07-06 Olivier Hainque <hainque@adacore.com>
17197 * gcc.c (process_command): When deciding if undefined variables
17198 should be ignored when processing specs, accept "gcc -v" as well.
17200 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17202 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17203 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17205 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17207 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17208 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17209 * config/arm/arm-tables.opt: Regenerate.
17210 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17212 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17214 2017-07-06 Carl Love <cel@us.ibm.com>
17216 * ChangeLog: Clean up from mid air collision
17218 2017-07-06 Carl Love <cel@us.ibm.com>
17220 * config/rs6000/rs6000-c.c: Add support for built-in functions
17221 vector signed int vec_subc (vector signed int, vector signed int);
17222 vector signed __int128 vec_subc (vector signed __int128,
17223 vector signed __int128);
17224 vector unsigned __int128 vec_subc (vector unsigned __int128,
17225 vector unsigned __int128);
17226 vector signed int vec_sube (vector signed int, vector signed int,
17227 vector signed int);
17228 vector unsigned int vec_sube (vector unsigned int,
17229 vector unsigned int,
17230 vector unsigned int);
17231 vector signed __int128 vec_sube (vector signed __int128,
17232 vector signed __int128,
17233 vector signed__int128);
17234 vector unsigned __int128 vec_sube (vector unsigned __int128,
17235 vector unsigned __int128,
17236 vector unsigned __int128);
17237 vector signed int vec_subec (vector signed int, vector signed int,
17238 vector signed int);
17239 vector unsigned int vec_subec (vector unsigned int,
17240 vector unsigned int,
17241 vector unsigned int);
17242 vector signed __int128 vec_subec (vector signed __int128,
17243 vector signed __int128,
17244 vector signed__int128);
17245 vector unsigned __int128 vec_subec (vector unsigned __int128,
17246 vector unsigned __int128,
17247 vector unsigned __int128);
17248 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17249 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17250 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17251 BU_ALTIVEC_OVERLOAD_X definitions.
17252 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17253 * doc/extend.texi: Update the built-in documentation file for the new
17254 built-in functions.
17256 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17259 * diagnostic-show-locus.c (layout::layout): Use start and finish
17260 spelling location for the start and finish of each range.
17261 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17262 Add unused aspect param.
17263 * input.c (expand_location_1): Add "aspect" param, and use it
17264 to access the correct part of the location.
17265 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17267 (expand_location_to_spelling_point): Likewise.
17268 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17269 param, and pass it to expand_location_1.
17271 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17273 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17274 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17275 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17276 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17277 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17278 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17279 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17280 _mm_maskz_getmant_ss): New intrinsics.
17281 (__builtin_ia32_getexpss128_mask): Changed to ...
17282 __builtin_ia32_getexpss128_round ... this.
17283 (__builtin_ia32_getexpsd128_mask): Changed to ...
17284 __builtin_ia32_getexpsd128_round ... this.
17285 * config/i386/i386-builtin-types.def
17286 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17287 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17288 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17289 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17290 __builtin_ia32_getmantss_mask_round): New builtins.
17291 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17292 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17293 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17294 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17295 * config/i386/sse.md
17296 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17297 avx512f_sgetexp<mode><mask_scalar_name>
17298 <round_saeonly_scalar_name> ... this.
17299 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17300 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17301 vgetexp<ssescalarmodesuffix>
17302 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17303 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17304 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17305 avx512f_vgetmant<mode><mask_scalar_name>
17306 <round_saeonly_scalar_name> ... this.
17307 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17308 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17309 vgetmant<ssescalarmodesuffix>
17310 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17311 %0<mask_scalar_operand4>, %1, %2
17312 <round_saeonly_scalar_mask_op4>, %3} ... this.
17313 * config/i386/subst.md (mask_scalar_operand4,
17314 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17315 round_saeonly_scalar_nimm_predicate): New subst attributes.
17317 2017-07-06 Julia Koval <julia.koval@intel.com>
17319 * config/i386/i386.c (ix86_erase_embedded_rounding):
17320 Remove code for old rounding pattern.
17322 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17324 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17326 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17328 * doc/sourcebuild.texi (Test Directives, Variants of
17329 dg-require-support): Add documentation for dg-require-stack-check.
17331 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17333 * config/i386/subst.md (mask_scalar, round_scalar,
17334 round_saeonly_scalar): New meta-templates.
17335 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17336 round_scalar_mask_operand3, round_scalar_mask_op3,
17337 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17338 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17339 round_saeonly_scalar_constraint,
17340 round_saeonly_scalar_prefix): New subst attribute.
17341 * config/i386/sse.md
17342 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17343 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17344 <round_scalar_name> ... this.
17345 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17346 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17347 <round_scalar_name> ... this.
17348 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17349 <sse>_vm<code><mode>3<mask_scalar_name>
17350 <round_saeonly_scalar_name> ... this.
17351 (v<plusminus_mnemonic><ssescalarmodesuffix>
17352 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17353 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17354 v<plusminus_mnemonic><ssescalarmodesuffix>
17355 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17356 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17357 (v<multdiv_mnemonic><ssescalarmodesuffix>
17358 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17359 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17360 v<multdiv_mnemonic><ssescalarmodesuffix>
17361 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17362 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17363 (v<maxmin_float><ssescalarmodesuffix>
17364 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17365 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17366 v<maxmin_float><ssescalarmodesuffix>
17367 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17368 %0<mask_scalar_operand3>, %1, %<iptr>2
17369 <round_saeonly_scalar_mask_op3>} ... this.
17371 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17373 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17374 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17376 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17377 Alan Hayward <alan.hayward@arm.com>
17378 David Sherwood <david.sherwood@arm.com>
17380 * combine.c (simplify_if_then_else): Remove "enum" before
17382 * compare-elim.c (can_eliminate_compare): Likewise.
17383 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17385 (aarch64_lookup_simd_builtin_type): Likewise.
17386 (aarch64_simd_builtin_type): Likewise.
17387 (aarch64_init_simd_builtin_types): Likewise.
17388 (aarch64_simd_expand_args): Likewise.
17389 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17391 (aarch64_reverse_mask): Likewise.
17392 (aarch64_simd_emit_reg_reg_move): Likewise.
17393 (aarch64_gen_adjusted_ldpstp): Likewise.
17394 (aarch64_ccmp_mode_to_code): Likewise.
17395 (aarch64_operands_ok_for_ldpstp): Likewise.
17396 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17397 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17399 (aarch64_min_divisions_for_recip_mul): Likewise.
17400 (aarch64_reassociation_width): Likewise.
17401 (aarch64_get_condition_code_1): Likewise.
17402 (aarch64_simd_emit_reg_reg_move): Likewise.
17403 (aarch64_simd_attr_length_rglist): Likewise.
17404 (aarch64_reverse_mask): Likewise.
17405 (aarch64_operands_ok_for_ldpstp): Likewise.
17406 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17407 (aarch64_gen_adjusted_ldpstp): Likewise.
17408 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17410 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17411 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17412 (arm_lookup_simd_builtin_type): Likewise.
17413 (arm_simd_builtin_type): Likewise.
17414 (arm_init_simd_builtin_types): Likewise.
17415 (arm_expand_builtin_args): Likewise.
17416 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17417 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17418 (ft32_setup_incoming_varargs): Likewise.
17419 (ft32_function_arg): Likewise.
17420 (ft32_function_arg_advance): Likewise.
17421 (ft32_pass_by_reference): Likewise.
17422 (ft32_arg_partial_bytes): Likewise.
17423 (ft32_valid_pointer_mode): Likewise.
17424 (ft32_addr_space_pointer_mode): Likewise.
17425 (ft32_addr_space_legitimate_address_p): Likewise.
17426 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17428 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17429 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17430 (iamcu_alignment): Likewise.
17431 (canonicalize_vector_int_perm): Likewise.
17432 (ix86_noce_conversion_profitable_p): Likewise.
17433 (ix86_mpx_bound_mode): Likewise.
17434 (ix86_operands_ok_for_move_multiple): Likewise.
17435 * config/microblaze/microblaze-protos.h
17436 (microblaze_expand_conditional_branch_reg): Likewise.
17437 * config/microblaze/microblaze.c
17438 (microblaze_expand_conditional_branch_reg): Likewise.
17439 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17441 (rs6000_reassociation_width): Likewise.
17442 (rs6000_invalid_binary_op): Likewise.
17443 (fusion_p9_p): Likewise.
17444 (emit_fusion_p9_load): Likewise.
17445 (emit_fusion_p9_store): Likewise.
17446 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17448 (riscv_hard_regno_mode_ok_p): Likewise.
17449 (riscv_address_insns): Likewise.
17450 (riscv_split_symbol): Likewise.
17451 (riscv_legitimize_move): Likewise.
17452 (riscv_function_value): Likewise.
17453 (riscv_hard_regno_nregs): Likewise.
17454 (riscv_expand_builtin): Likewise.
17455 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17456 (riscv_build_integer): Likewise.
17457 (riscv_split_integer): Likewise.
17458 (riscv_legitimate_constant_p): Likewise.
17459 (riscv_cannot_force_const_mem): Likewise.
17460 (riscv_regno_mode_ok_for_base_p): Likewise.
17461 (riscv_valid_base_register_p): Likewise.
17462 (riscv_valid_offset_p): Likewise.
17463 (riscv_valid_lo_sum_p): Likewise.
17464 (riscv_classify_address): Likewise.
17465 (riscv_legitimate_address_p): Likewise.
17466 (riscv_address_insns): Likewise.
17467 (riscv_load_store_insns): Likewise.
17468 (riscv_force_binary): Likewise.
17469 (riscv_split_symbol): Likewise.
17470 (riscv_force_address): Likewise.
17471 (riscv_legitimize_address): Likewise.
17472 (riscv_move_integer): Likewise.
17473 (riscv_legitimize_const_move): Likewise.
17474 (riscv_legitimize_move): Likewise.
17475 (riscv_address_cost): Likewise.
17476 (riscv_subword): Likewise.
17477 (riscv_output_move): Likewise.
17478 (riscv_canonicalize_int_order_test): Likewise.
17479 (riscv_emit_int_order_test): Likewise.
17480 (riscv_function_arg_boundary): Likewise.
17481 (riscv_pass_mode_in_fpr_p): Likewise.
17482 (riscv_pass_fpr_single): Likewise.
17483 (riscv_pass_fpr_pair): Likewise.
17484 (riscv_get_arg_info): Likewise.
17485 (riscv_function_arg): Likewise.
17486 (riscv_function_arg_advance): Likewise.
17487 (riscv_arg_partial_bytes): Likewise.
17488 (riscv_function_value): Likewise.
17489 (riscv_pass_by_reference): Likewise.
17490 (riscv_setup_incoming_varargs): Likewise.
17491 (riscv_print_operand): Likewise.
17492 (riscv_elf_select_rtx_section): Likewise.
17493 (riscv_save_restore_reg): Likewise.
17494 (riscv_for_each_saved_reg): Likewise.
17495 (riscv_register_move_cost): Likewise.
17496 (riscv_hard_regno_mode_ok_p): Likewise.
17497 (riscv_hard_regno_nregs): Likewise.
17498 (riscv_class_max_nregs): Likewise.
17499 (riscv_memory_move_cost): Likewise.
17500 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17501 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17502 (rl78_addr_space_address_mode): Likewise.
17503 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17505 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17506 (rs6000_reassociation_width): Likewise.
17507 (rs6000_invalid_binary_op): Likewise.
17508 (fusion_p9_p): Likewise.
17509 (emit_fusion_p9_load): Likewise.
17510 (emit_fusion_p9_store): Likewise.
17511 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17512 (ok_for_simple_move_operands): Likewise.
17513 (ok_for_simple_move_strict_operands): Likewise.
17514 (ok_for_simple_arith_logic_operands): Likewise.
17515 (visium_legitimize_reload_address): Likewise.
17516 (visium_select_cc_mode): Likewise.
17517 (output_cbranch): Likewise.
17518 (visium_split_double_move): Likewise.
17519 (visium_expand_copysign): Likewise.
17520 (visium_expand_int_cstore): Likewise.
17521 (visium_expand_fp_cstore): Likewise.
17522 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17523 (visium_function_arg): Likewise.
17524 (visium_function_arg_advance): Likewise.
17525 (visium_libcall_value): Likewise.
17526 (visium_setup_incoming_varargs): Likewise.
17527 (visium_legitimate_constant_p): Likewise.
17528 (visium_legitimate_address_p): Likewise.
17529 (visium_legitimize_address): Likewise.
17530 (visium_secondary_reload): Likewise.
17531 (visium_register_move_cost): Likewise.
17532 (visium_memory_move_cost): Likewise.
17533 (prepare_move_operands): Likewise.
17534 (ok_for_simple_move_operands): Likewise.
17535 (ok_for_simple_move_strict_operands): Likewise.
17536 (ok_for_simple_arith_logic_operands): Likewise.
17537 (visium_function_value_1): Likewise.
17538 (rtx_ok_for_offset_p): Likewise.
17539 (visium_legitimize_reload_address): Likewise.
17540 (visium_split_double_move): Likewise.
17541 (visium_expand_copysign): Likewise.
17542 (visium_expand_int_cstore): Likewise.
17543 (visium_expand_fp_cstore): Likewise.
17544 (visium_split_cstore): Likewise.
17545 (visium_select_cc_mode): Likewise.
17546 (visium_split_cbranch): Likewise.
17547 (output_cbranch): Likewise.
17548 (visium_print_operand_address): Likewise.
17549 * expmed.c (flip_storage_order): Likewise.
17550 * expmed.h (emit_cstore): Likewise.
17551 (flip_storage_order): Likewise.
17552 * genrecog.c (validate_pattern): Likewise.
17553 * hsa-gen.c (gen_hsa_addr): Likewise.
17554 * internal-fn.c (expand_arith_overflow): Likewise.
17555 * ira-color.c (allocno_copy_cost_saving): Likewise.
17556 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17557 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17558 (process_invariant_for_inheritance): Likewise.
17559 * lra-eliminations.c (move_plus_up): Likewise.
17560 * omp-low.c (lower_oacc_reductions): Likewise.
17561 * simplify-rtx.c (simplify_subreg): Likewise.
17562 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17563 (TARGET_CHKP_BOUND_MODE): Likewise..
17564 * targhooks.c (default_chkp_bound_mode): Likewise.
17565 (default_setup_incoming_vararg_bounds): Likewise.
17566 * targhooks.h (default_chkp_bound_mode): Likewise.
17567 (default_setup_incoming_vararg_bounds): Likewise.
17568 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17569 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17570 (have_whole_vector_shift): Likewise.
17571 * tree-vect-stmts.c (vectorizable_load): Likewise.
17572 * doc/tm.texi: Regenerate.
17574 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17576 Graceful degrade if Binutils PR21472 is not available.
17579 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17580 .rodata in flash test fails.
17581 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17582 * confgure: Regenerate.
17583 * config.in: Regenerate.
17584 * config/avr/avr.c (avr_asm_named_section)
17585 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17586 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17587 (avr_asm_init_sections): Same.
17589 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17591 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17592 (fma<VH:mode>4_intrinsic): Likewise.
17593 (*fmsub<VCVTF:mode>4): Likewise.
17594 (*fmsub<VH:mode>4_intrinsic): Likewise.
17596 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17599 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17600 Don't depend on "optimize > 0".
17601 (out_movhi_r_mr, out_movqi_mr_r): Same.
17602 (out_movhi_mr_r, out_movqi_r_mr): Same.
17603 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17604 io_address_operand on "optimize > 0".
17606 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17608 * tree-loop-distribution.c: Add general explanantion on the pass.
17609 (generate_loops_for_partition): Mark distributed loop.
17610 (pg_add_dependence_edges): New parameter. Handle alias data
17611 dependence specially and record it in the parameter if asked.
17612 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17613 (init_partition_graph_vertices, add_partition_graph_edge): New.
17614 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17615 (free_partition_graph_vdata, build_partition_graph): New.
17616 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17617 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17618 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17619 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17620 (version_for_distribution_p, finalize_partitions): New.
17621 (distribute_loop): Handle alias data dependence specially. Factor
17622 out loop fusion code as functions and call these functions.
17624 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17626 * tree-loop-distribution.c (classify_partition): New parameter and
17627 better handle reduction statement.
17628 (rdg_build_partitions): Revise comment.
17629 (distribute_loop): Compute statements in all partitions and pass it
17630 to classify_partition.
17632 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17634 * tree-loop-distribution.c (enum partition_type): New.
17635 (struct partition): New field type.
17636 (partition_merge_into): Add parameter. Update partition type.
17637 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17638 (build_rdg_partition_for_vertex): Compute partition type.
17639 (rdg_build_partitions): Dump partition type.
17640 (distribute_loop): Update calls to partition_merge_into.
17642 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17644 * tree-loop-distribution.c (struct ddr_hasher): New.
17645 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17647 (classify_partition): Call get_data_dependence.
17648 (pg_add_dependence_edges): Ditto.
17649 (distribute_loop): Release data dependence hash table.
17651 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17653 * tree-loop-distribution.c (ref_base_address): Delete.
17654 (similar_memory_accesses): Rename ...
17655 (share_memory_accesses): ... to this. Check if partitions access
17656 the same memory reference.
17657 (distribute_loop): Call share_memory_accesses.
17659 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17661 * tree-loop-distribution.c (struct partition): New field recording
17662 its data reference.
17663 (partition_alloc, partition_free): Init and release data refs.
17664 (partition_merge_into): Merge data refs.
17665 (build_rdg_partition_for_vertex): Collect data refs for partition.
17666 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17668 (distribute_loop): Remve data refs from vertice data of partition
17671 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17673 * tree-loop-distribution.c (params.h): Include header file.
17674 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17675 (datarefs_vec): New global var.
17676 (create_rdg_vertices): Use datarefs_vec directly.
17677 (free_rdg): Don't free data references.
17678 (build_rdg): Update use. Don't free data references.
17679 (distribute_loop): Compute global variable for data references.
17680 Bail out if there are too many data references.
17682 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17684 * tree-loop-distribution.c (loop_nest): New global var.
17685 (build_rdg): Use loop directly, rather than loop nest.
17686 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17688 (distribute_loop): Compute global variable loop nest. Update use.
17690 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17692 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17693 (partition_merge_into): New parameter. Dump reason for fusion.
17694 (distribute_loop): Update use of partition_merge_into.
17696 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17698 * tree-loop-distribution.c (bb_top_order_index): New.
17699 (bb_top_order_index_size, bb_top_order_cmp): New.
17700 (stmts_from_loop): Use topological order.
17701 (pass_loop_distribution::execute): Compute and release topological
17702 order for basic blocks.
17704 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17706 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17709 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17711 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17712 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17713 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17714 * internal-fn.def (LOOP_DIST_ALIAS): New.
17715 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17716 (fold_loop_internal_call): ... this.
17717 (vect_loop_dist_alias_call): New function.
17718 (set_uid_loop_bbs): Call fold_loop_internal_call.
17719 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17722 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17725 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17726 Require dead FLAGS_REG at the beginning of a peephole.
17728 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17731 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17732 arguments in the call to __builtin_ia32_sbb_u32.
17733 (_subborrow_u64): Swap _X and _Y arguments in the call to
17734 __builtin_ia32_sbb_u64.
17736 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17739 * tree-vrp.c (compare_assert_loc): Turn into a function template
17740 with stable template parameter. Only test if a->e is NULL,
17741 !a->e == !b->e has been verified already. Use e == NULL or
17742 e != NULL instead of e or ! e tests. If stable is true, don't use
17743 iterative_hash_expr, on the other side allow a or b or both NULL
17744 and sort the NULLs last.
17745 (process_assert_insertions): Sort using compare_assert_loc<false>
17746 instead of compare_assert_loc, later sort using
17747 compare_assert_loc<true> before calling process_assert_insertions_for
17748 in a loop. Use break instead of continue once seen NULL pointer.
17750 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17752 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17753 Cortex-R7 and Cortex-R8 processors.
17755 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17757 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17758 uninitialized while src is not.
17760 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17762 * common/config/arm/arm-common.c: Adjust include path for
17764 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17765 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17766 (arm-cpu-data.h): Likewise.
17767 (arm-cpu-cdata.h): Likewise.
17768 * config/arm/arm-cpu.h: Delete.
17769 * config/arm/arm-cpu-cdata.h: Delete.
17770 * config/arm/arm-cpu-data.h: Delete.
17772 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17774 * config/arm/arm-cpus.in (cortex-a55): New.
17775 (cortex-a75): Likewise.
17776 (cortex-a75.cortex-a55): Likewise.
17777 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17779 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17780 * config/arm/arm-cpu-cdata.h: Regenerate.
17781 * config/arm/arm-cpu-data.h: Regenerate.
17782 * config/arm/arm-cpu.h: Regenerate.
17783 * config/arm/arm-tables.opt: Regenerate.
17784 * config/arm/arm-tune.md: Regenerate.
17786 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17788 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17790 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17792 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17795 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17797 PR tree-optimization/81292
17798 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17799 full_string_p, also call adjust_related_strinfos if the adjustment
17800 is simple, otherwise invalidate related strinfos.
17802 2017-07-04 Martin Liska <mliska@suse.cz>
17805 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17806 newly created variable as DECL_IGNORED_P.
17808 2017-07-04 Martin Liska <mliska@suse.cz>
17811 * ipa-inline.c (inline_small_functions):
17812 Use xstrdup_for_dump.
17814 2017-07-04 Tom de Vries <tom@codesourcery.com>
17816 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17818 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17821 * config/darwin.c (darwin_function_switched_text_sections):
17824 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17826 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17828 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17830 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17832 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17834 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17835 min_profitable_iters, and th as inclusive lower bounds.
17836 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17837 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17838 for min_profitable_iters and min_profitable_estimate.
17839 (vect_transform_loop): Treat th as an inclusive lower bound.
17840 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17842 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17845 * config/darwin.c (darwin_function_switched_text_sections):
17846 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17847 in two pieces, and suppress the use of buf.
17849 2017-07-03 Nathan Sidwell <nathan@acm.org>
17851 * hash-table.h (hash_table_mod1): Fix indentation.
17853 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17855 PR middle-end/81290
17856 * predict.c (force_edge_cold): Be more careful about propagation
17858 * profile-count.h (profile_probability::guessed,
17859 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17861 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17863 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17865 * doc/invoke.texi (rcpc architecture extension): Document it.
17867 2017-07-03 Richard Biener <rguenther@suse.de>
17869 PR tree-optimization/60510
17870 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17871 the scalar reduction PHI and use it.
17872 (vectorizable_reduction): Properly guard the single_defuse_cycle
17873 path for non-SLP reduction chains where we cannot use it.
17874 Rework reduc_def/index and vector type deduction. Rework
17875 vector operand gathering during reduction op code-gen.
17876 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17877 chains dissolve the chain and leave it to non-SLP reduction
17880 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17882 * tree-data-ref.h (dr_alignment): Declare.
17883 * tree-data-ref.c (dr_alignment): New function.
17884 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17885 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17887 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17889 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17891 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17892 and base_misalignment fields.
17893 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17894 * tree-data-ref.c: Include builtins.h.
17895 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17896 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17897 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17898 * tree-vect-data-refs.c: Include tree-cfg.h.
17899 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17900 fields instead of calculating an alignment here.
17901 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17902 innermost_loop_behavior fields.
17904 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17906 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17908 (DR_STEP_ALIGNMENT): New macro.
17909 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17910 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17911 (create_data_ref): Print it.
17912 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17913 to tell whether the step preserves vector (mis)alignment.
17914 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17915 Move the check for an integer step and generalise to all INTEGER_CST.
17916 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17917 Print the outer step alignment.
17919 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17921 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17922 with offset_alignment.
17923 (DR_ALIGNED_TO): Delete.
17924 (DR_OFFSET_ALIGNMENT): New macro.
17925 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17926 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17927 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17928 (create_data_ref): Likewise.
17929 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17930 (vect_analyze_data_refs): Likewise.
17931 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17932 creating dummy innermost behavior.
17934 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17936 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17937 with a "innermost_loop_behavior *" and refeence tree.
17938 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17939 (create_data_ref): Update call accordingly.
17940 * tree-predcom.c (find_looparound_phi): Likewise.
17942 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17944 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17945 fields with dr_wrt_vec_loop.
17946 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17947 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17948 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17949 (vect_dr_behavior): New function.
17950 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17951 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17952 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17953 track whether the step preserves the misalignment.
17954 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17955 Use vect_dr_behavior.
17956 (vect_setup_realignment): Update call accordingly.
17957 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17958 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17959 call to vect_create_addr_base_for_vector_ref.
17960 (vect_create_cond_for_align_checks): Likewise.
17961 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17962 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17963 (vect_recog_mask_conversion_pattern): Likewise.
17964 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17965 (new_stmt_vec_info): Remove redundant zeroing.
17967 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17969 * common/config/arm/arm-common.c (arm_be8_option): New function.
17970 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17971 (ISA_ARMv6): Add isa_bit_be8.
17972 * config/arm/arm.h (arm_be8_option): Add prototype.
17973 (BE8_SPEC_FUNCTION): New define.
17974 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17975 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17976 (mlittle-endian): Similarly.
17977 (mbe8, mbe32): New options.
17978 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17979 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17981 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17983 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17985 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17987 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17988 (cleanup_tree_cfg_bb): Use it.
17989 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17991 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17993 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17996 * loop-doloop.c (add_test): Update profile.
17998 2017-07-03 Martin Liska <mliska@suse.cz>
18001 * sanopt.c (rewrite_usage_of_param): New function.
18002 (sanitize_rewrite_addressable_params): Likewise.
18003 (pass_sanopt::execute): Call rewrite_usage_of_param.
18005 2017-07-03 Richard Biener <rguenther@suse.de>
18007 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18008 back to using VIEW_CONVERT_EXPR.
18010 2017-07-03 Martin Liska <mliska@suse.cz>
18013 * doc/extend.texi: Document when a resolver function is
18014 generated for target_clones.
18016 2017-07-03 Martin Liska <mliska@suse.cz>
18018 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18019 and shadow memory of auto variables which are subject of
18020 use-after-scope sanitization.
18021 (asan_expand_mark_ifn): Add do set only when is_poison.
18023 2016-07-03 Richard Biener <rguenther@suse.de>
18025 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18027 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18028 (vectorizable_reduction): Perform reduction PHI creation when
18029 visiting a reduction PHI and adjust and simplify code generation
18030 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18031 (vect_transform_loop): Visit reduction PHIs.
18032 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18033 defs into the SLP tree.
18034 (vect_build_slp_tree): Reduction defs terminate the recursion.
18035 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18037 (vect_get_vec_defs_for_stmt_copy): Export.
18038 (vect_get_vec_defs): Likewise.
18039 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18041 (vect_get_vec_defs_for_stmt_copy): Declare.
18042 (vect_get_vec_defs): Likewise.
18044 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18046 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18047 parameter with a "loop" parameter and use it instead of the
18048 loop containing DR_STMT. Don't check simple_iv when doing
18049 BB analysis. Describe the two analysis modes in the comment.
18051 2017-07-03 Tom de Vries <tom@codesourcery.com>
18053 PR tree-optimization/69468
18054 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18055 (find_same_succ_bb): Handle ignore_edge_flags.
18057 2017-07-03 Tom de Vries <tom@codesourcery.com>
18059 PR tree-optimization/81192
18060 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18062 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18064 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18066 2017-07-03 Tom de Vries <tom@codesourcery.com>
18068 PR tree-optimization/81192
18069 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18070 BB_SAME_SUCC (bb) == NULL.
18072 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18074 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18077 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18079 * dumpfile.c: Include profile-count.h
18080 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18082 (insert_cond_bb): Update profile.
18083 * tree-cfg.h (insert_cond_bb): Update prototype.
18084 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18085 * tree-dump.c: Do not include tree-cfg.
18087 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18089 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18091 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18093 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18096 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18098 * tree-complex.c (expand_complex_div_wide): update profile.
18100 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18101 Alan Hayward <alan.hayward@arm.com>
18102 David Sherwood <david.sherwood@arm.com>
18104 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18105 (CORETYPES_H): New define.
18106 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18107 (insn-modes-inline.h, s-modes-inline-h): New rules.
18108 (generated_files): Add insn-modes-inline.h.
18109 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18110 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18111 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18112 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18113 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18114 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18115 (build/gencodes.o, build/genconditions.o): Likewise.
18116 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18117 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18118 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18119 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18120 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18121 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18122 * coretypes.h: Include everything up to real.h for generators.
18123 Include insn-modes.h first. Include wide-int-print.h after
18124 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18125 * machmode.h: Don't include insn-modes.h here.
18126 * function-tests.c: Remove includes of signop.h, machmode.h,
18127 double-int.h and wide-int.h.
18129 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18131 * optc-save-gen.awk: Likewise.
18132 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18133 * godump.c: Remove include of wide-int-print.h.
18134 * pretty-print.h: Likewise.
18135 * wide-int-print.cc: Likewise.
18136 * wide-int.cc: Likewise.
18137 * hash-map-tests.c: Remove include of signop.h.
18138 * hash-set-tests.c: Likewise.
18139 * rtl-tests.c: Likewise.
18140 * mkconfig.sh: Remove include of machmode.h.
18141 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18143 (emit_insn_modes_inline_h): ...this new function. Emit the code
18144 into an insn-modes-inline.h header file, adding appropriate
18145 include guards and end comments.
18146 (emit_insn_modes_c_header): Remove include of machmode.h.
18147 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18149 (main): Handle -i flag and call emit_insn_modes_inline_h when
18152 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18154 * tree-ssa-strlen.c (strinfo): Rename the length field to
18155 nonzero_chars. Add a full_string_p field.
18156 (compare_nonzero_chars, zero_length_string_p): New functions.
18157 (get_addr_stridx): Add an offset_out parameter.
18158 Use compare_nonzero_chars.
18159 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18160 (new_strinfo): Update after above changes to strinfo.
18161 (set_endptr_and_length): Set full_string_p.
18162 (get_string_length): Update after above changes to strinfo.
18163 (unshare_strinfo): Update call to new_strinfo.
18164 (maybe_invalidate): Likewise.
18165 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18166 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18167 as a uhwi instead of an shwi. Update after above changes to
18168 strinfo and new_strinfo.
18169 (zero_length_string): Assert that chainsi contains full strings.
18170 Use zero_length_string_p. Update call to new_strinfo.
18171 (adjust_related_strinfos): Update after above changes to strinfo.
18172 Copy full_string_p from origsi.
18173 (adjust_last_stmt): Use zero_length_string_p.
18174 (handle_builtin_strlen): Update after above changes to strinfo and
18175 new_strinfo. Install the lhs as the string length if the previous
18176 entry didn't describe a full string.
18177 (handle_builtin_strchr): Update after above changes to strinfo
18179 (handle_builtin_strcpy): Likewise.
18180 (handle_builtin_strcat): Likewise.
18181 (handle_builtin_malloc): Likewise.
18182 (handle_pointer_plus): Likewise.
18183 (handle_builtin_memcpy): Likewise. Track nonzero characters
18184 that aren't necessarily followed by a nul terminator.
18185 (handle_char_store): Likewise.
18187 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18189 PR tree-optimization/80769
18190 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18191 for malloc and calloc. Document the new invariant that all related
18192 strinfos have delayed lengths or none do.
18193 (verify_related_strinfos): Move earlier in file.
18194 (set_endptr_and_length): New function, split out from...
18195 (get_string_length): ...here. Also set the lengths of related
18197 (zero_length_string): Assert that chainsi has known (rather than
18199 (adjust_related_strinfos): Likewise.
18201 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18203 PR tree-optimization/81136
18204 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18205 assert that two references with the same misalignment have the same
18206 compile-time misalignment if those compile-time misalignments
18209 2017-07-01 Andi Kleen <ak@linux.intel.com>
18211 * print-tree.c (print_node): Print all attributes.
18213 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18215 * cfg.c (scale_bbs_frequencies): New function.
18216 * cfg.h (scale_bbs_frequencies): Declare it.
18217 * cfgloopanal.c (single_likely_exit): Cleanup.
18218 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18220 (scale_loop_profile): Likewise.
18221 (loop_version): Likewise.
18222 (create_empty_loop_on_edge): Update.
18223 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18224 scale_loop_frequencies, scale_loop_profile, loopify,
18225 loop_version): Update prototypes.
18226 * modulo-sched.c (sms_schedule): Update.
18227 * predict.c (unlikely_executed_edge_p): Also check probability.
18228 (probably_never_executed_edge_p): Fix typo.
18229 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18230 * tree-parloops.c (gen_parallel_loop): Update.
18231 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18232 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18233 * tree-ssa-loop-split.c (split_loop): Update.
18234 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18235 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18236 (vect_loop_versioning): Update.
18237 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18239 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18241 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18243 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18245 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18246 to keep profile consistent.
18248 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18250 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18251 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18252 * profile-count.h (max_safe_multiplier): Make unsigned.
18253 (profile_count::guessed_zero): New.
18255 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18257 * bb-reorder.c (fix_up_crossing_landing_pad,
18258 fix_crossing_conditional_branches): Use make_single_succ_edge
18259 to keep profile consistent.
18261 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18263 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18266 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18269 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18270 the right scopes, make sure cond_jump isn't preserved between multiple
18271 iterations. Search for fallthru edge whenever there are 3+ edges and
18272 use find_fallthru_edge for it.
18274 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18276 Patch by Alexander Monakov <amonakov@ispras.ru>
18277 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18278 probabilities consistently.
18280 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18282 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18283 emit_cmp_and_jump_insns.
18285 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18288 * tree-inline.c (expand_call_inline): Combine profile statuses.
18290 2017-06-30 Andrew Pinski <apinski@cavium.com>
18292 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18293 fold_stmt returned true.
18295 2017-06-30 Nathan Sidwell <nathan@acm.org>
18297 * ggc.h (empty_string): Delete.
18298 * cfgexpand.c (expand_asm_stmt): Use plain "".
18299 * optabs.c (expand_asm_memory_barrier): Likewise.
18300 * stringpool.c (empty_string): Delete.
18301 (digit_vector, digit_string): Delete.
18302 (ggc_alloc_string): Use plain "", don't optimize single digit
18303 strings. Use ggc_alloc_atomic.
18305 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18307 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18308 comparison set and one other set, use the cost of the non-comparison
18311 2017-06-30 Nathan Sidwell <nathan@acm.org>
18313 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18316 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18318 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18319 loops. Remove now unneeded calls to gimple_switch_set_label() that
18320 just set removed labels to NULL_TREE.
18322 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18324 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18325 (set_range_info): ...here. Only call set_range_info_raw if domain
18327 (set_nonzero_bits): Call set_range_info_raw.
18328 * tree-ssanames.h (set_range_info_raw): New.
18330 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18333 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18334 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18335 of nonimmediate_operand and <store_mask_constraint> instead of m
18336 for the input operand. For V8FI iterator, always split if input
18337 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18338 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18339 <store_mask_predicate> instead of register_operand and
18340 <store_mask_constraint> instead of v for the input operand. Make
18341 sure both operands aren't MEMs for if not <mask_applied>.
18343 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18345 * lto-wrapper.c (copy_file) Close both file descriptors before
18348 2017-06-30 Martin Liska <mliska@suse.cz>
18351 * multiple_target.c (create_dispatcher_calls): Make ifunc
18352 also for function that don't have calls or are not referenced.
18354 2017-06-30 Richard Biener <rguenther@suse.de>
18356 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18357 analyze the first scalar stmt. Move vector type computation
18358 for the BB case here from ...
18359 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18360 live operation processing in the SLP case properly.
18362 2017-06-30 Richard Biener <rguenther@suse.de>
18364 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18366 2017-06-30 Martin Liska <mliska@suse.cz>
18369 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18370 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18372 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18374 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18377 2017-06-30 Richard Biener <rguenther@suse.de>
18379 PR tree-optimization/81249
18380 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18381 condition reduction result to original scalar type.
18383 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18385 * profile-count.h (enum profile_quality): Fix typos and whitespace
18388 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18390 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18391 type for branch probabilities.
18393 2017-06-29 Julian Brown <julian@codesourcery.com>
18394 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18396 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18397 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18398 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18399 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18401 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18403 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18404 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18405 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18406 CC usage from generic code to here.
18407 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18408 CC usage into the target macros.
18410 2017-06-29 Maya Rashish <coypu@sdf.org>
18412 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18415 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18417 * arm/arm-builtins.c: Include profile-count.h
18418 * except.c (sjlj_emit_function_enter): Use
18419 profile_probability::unlikely.
18421 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18423 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18424 and tocrel_offset be pointer args rather than implicitly using
18426 (legitimate_constant_pool_address_p, rs6000_emit_move,
18427 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18428 tocrel_offset and use in toc_relative_expr_p call.
18429 (print_operand, print_operand_address): Use static tocrel_base_oac
18430 and tocrel_offset_oac.
18431 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18434 2017-06-29 Maya Rashish <coypu@sdf.org>
18436 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18438 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18440 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18441 objects, take into account only the alignment of 'op0' and 'mode1' if
18444 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18446 * ccmp.c (ccmp_tree_comparison_p): New function.
18447 (ccmp_candidate_p): Update to use above function.
18448 (get_compare_parts): New function.
18449 (expand_ccmp_next): Update to use new functions.
18450 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18452 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18453 take mode as argument.
18454 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18455 * expr.c (expand_expr_real_1): Pass mode as argument.
18457 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18459 * combine.c (combine_instructions): Print insns to dump_file, together
18462 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18464 * asan.c (asan_emit_stack_protection): Update.
18465 (create_cond_insert_point): Update.
18466 * auto-profile.c (afdo_propagate_circuit): Update.
18467 * basic-block.h (struct edge_def): Turn probability to
18468 profile_probability.
18469 (EDGE_FREQUENCY): Update.
18470 * bb-reorder.c (find_traces_1_round): Update.
18471 (better_edge_p): Update.
18472 (sanitize_hot_paths): Update.
18473 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18474 (make_single_succ_edge): Update.
18475 (check_bb_profile): Update.
18476 (dump_edge_info): Update.
18477 (update_bb_profile_for_threading): Update.
18478 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18479 probabilitycount to 0.
18480 * cfgbuild.c (compute_outgoing_frequencies): Update.
18481 * cfgcleanup.c (try_forward_edges): Update.
18482 (outgoing_edges_match): Update.
18483 (try_crossjump_to_edge): Update.
18484 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18485 (expand_gimple_tailcall): Update.
18486 (construct_init_block): Use make_single_succ_edge.
18487 (construct_exit_block): Use make_single_succ_edge.
18488 * cfghooks.c (verify_flow_info): Update.
18489 (redirect_edge_succ_nodup): Update.
18490 (split_edge): Update.
18491 (account_profile_record): Update.
18492 * cfgloopanal.c (single_likely_exit): Update.
18493 * cfgloopmanip.c (scale_loop_profile): Update.
18494 (set_zero_probability): Remove.
18495 (duplicate_loop_to_header_edge): Update.
18496 * cfgloopmanip.h (loop_version): Update prototype.
18497 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18498 (force_nonfallthru_and_redirect): Update.
18499 (update_br_prob_note): Update.
18500 (rtl_verify_edges): Update.
18501 (purge_dead_edges): Update.
18502 (rtl_lv_add_condition_to_bb): Update.
18503 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18504 * cgraphunit.c (init_lowered_empty_function): Update.
18505 (cgraph_node::expand_thunk): Update.
18506 * cilk-common.c: Include profile-count.h
18507 * dojump.c (inv): Remove.
18508 (jumpifnot): Update.
18509 (jumpifnot_1): Update.
18510 (do_jump_1): Update.
18512 (do_jump_by_parts_greater_rtx): Update.
18513 (do_compare_rtx_and_jump): Update.
18514 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18515 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18516 * dwarf2cfi.c: Include profile-count.h
18517 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18518 (sjlj_emit_dispatch_table): Likewise.
18519 * explow.c: Include profile-count.h
18520 * expmed.c (emit_store_flag_force): Update.
18521 (do_cmp_and_jump): Update.
18522 * expr.c (compare_by_pieces_d::generate): Update.
18523 (compare_by_pieces_d::finish_mode): Update.
18524 (emit_block_move_via_loop): Update.
18525 (store_expr_with_bounds): Update.
18526 (store_constructor): Update.
18527 (expand_expr_real_2): Update.
18528 (expand_expr_real_1): Update.
18529 * expr.h (try_casesi, try_tablejump): Update prototypes.
18530 * gimple-pretty-print.c (dump_probability): Update.
18531 (dump_profile): New.
18532 (dump_gimple_label): Update.
18533 (dump_gimple_bb_header): Update.
18534 * graph.c (draw_cfg_node_succ_edges): Update.
18535 * hsa-gen.c (convert_switch_statements): Update.
18536 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18537 (find_if_case_1): Update.
18538 (find_if_case_2): Update.
18539 * internal-fn.c (expand_arith_overflow_result_store): Update.
18540 (expand_addsub_overflow): Update.
18541 (expand_neg_overflow): Update.
18542 (expand_mul_overflow): Update.
18543 (expand_vector_ubsan_overflow): Update.
18544 * ipa-cp.c (good_cloning_opportunity_p): Update.
18545 * ipa-split.c (split_function): Use make_single_succ_edge.
18546 * ipa-utils.c (ipa_merge_profiles): Update.
18547 * loop-doloop.c (add_test): Update.
18548 (doloop_modify): Update.
18549 * loop-unroll.c (compare_and_jump_seq): Update.
18550 (unroll_loop_runtime_iterations): Update.
18551 * lra-constraints.c (lra_inheritance): Update.
18552 * lto-streamer-in.c (input_cfg): Update.
18553 * lto-streamer-out.c (output_cfg): Update.
18554 * mcf.c (adjust_cfg_counts): Update.
18555 * modulo-sched.c (sms_schedule): Update.
18556 * omp-expand.c (expand_omp_for_init_counts): Update.
18557 (extract_omp_for_update_vars): Update.
18558 (expand_omp_ordered_sink): Update.
18559 (expand_omp_for_ordered_loops): Update.
18560 (expand_omp_for_generic): Update.
18561 (expand_omp_for_static_nochunk): Update.
18562 (expand_omp_for_static_chunk): Update.
18563 (expand_cilk_for): Update.
18564 (expand_omp_simd): Update.
18565 (expand_omp_taskloop_for_outer): Update.
18566 (expand_omp_taskloop_for_inner): Update.
18567 * omp-simd-clone.c (simd_clone_adjust): Update.
18568 * optabs.c (expand_doubleword_shift): Update.
18569 (expand_abs): Update.
18570 (emit_cmp_and_jump_insn_1): Update.
18571 (expand_compare_and_swap_loop): Update.
18572 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18573 * predict.c (predictable_edge_p): Update.
18574 (edge_probability_reliable_p): Update.
18575 (set_even_probabilities): Update.
18576 (combine_predictions_for_insn): Update.
18577 (combine_predictions_for_bb): Update.
18578 (propagate_freq): Update.
18579 (estimate_bb_frequencies): Update.
18580 (force_edge_cold): Update.
18581 * profile-count.c (profile_count::dump): Add missing space into dump.
18582 (profile_count::debug): Add newline.
18583 (profile_count::differs_from_p): Explicitly convert to unsigned.
18584 (profile_count::stream_in): Update.
18585 (profile_probability::dump): New member function.
18586 (profile_probability::debug): New member function.
18587 (profile_probability::differs_from_p): New member function.
18588 (profile_probability::differs_lot_from_p): New member function.
18589 (profile_probability::stream_in): New member function.
18590 (profile_probability::stream_out): New member function.
18591 * profile-count.h (profile_count_quality): Rename to ...
18592 (profile_quality): ... this one.
18593 (profile_probability): New.
18594 (profile_count): Update.
18595 * profile.c (compute_branch_probabilities): Update.
18596 * recog.c (peep2_attempt): Update.
18597 * sched-ebb.c (schedule_ebbs): Update.
18598 * sched-rgn.c (find_single_block_region): Update.
18599 (compute_dom_prob_ps): Update.
18600 (schedule_region): Update.
18601 * sel-sched-ir.c (compute_succs_info): Update.
18602 * stmt.c (struct case_node): Update.
18603 (do_jump_if_equal): Update.
18604 (get_outgoing_edge_probs): Update.
18605 (conditional_probability): Update.
18606 (emit_case_dispatch_table): Update.
18607 (expand_case): Update.
18608 (expand_sjlj_dispatch_table): Update.
18609 (emit_case_nodes): Update.
18610 * targhooks.c: Update.
18611 * tracer.c (better_p): Update.
18612 (find_best_successor): Update.
18613 * trans-mem.c (expand_transaction): Update.
18614 * tree-call-cdce.c: Update.
18615 * tree-cfg.c (gimple_split_edge): Upate.
18616 (move_sese_region_to_fn): Upate.
18617 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18618 * tree-eh.c (lower_resx): Upate.
18619 (cleanup_empty_eh_move_lp): Upate.
18620 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18621 * tree-inline.c (copy_edges_for_bb): Update.
18622 (copy_cfg_body): Update.
18623 * tree-parloops.c (gen_parallel_loop): Update.
18624 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18625 (gimple_gen_time_profiler): Update.
18626 * tree-ssa-dce.c (remove_dead_stmt): Update.
18627 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18628 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18629 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18630 (unloop_loops): Update.
18631 (try_peel_loop): Update.
18632 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18633 * tree-ssa-loop-split.c (connect_loops): Update.
18634 (split_loop): Update.
18635 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18636 (hoist_guard): Update.
18637 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18638 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18639 (value_replacement): Update.
18640 * tree-ssa-reassoc.c (branch_fixup): Update.
18641 * tree-ssa-tail-merge.c (replace_block_by): Update.
18642 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18643 (create_edge_and_update_destination_phis): Update.
18644 (compute_path_counts): Update.
18645 (recompute_probabilities): Update.
18646 (update_joiner_offpath_counts): Update.
18647 (freqs_to_counts_path): Update.
18648 (duplicate_thread_path): Update.
18649 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18650 (struct switch_conv_info): Update.
18651 (gen_inbound_check): Update.
18652 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18653 (vect_do_peeling): Update.
18654 (vect_loop_versioning): Update.
18655 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18656 (optimize_mask_stores): Update.
18657 * ubsan.c (ubsan_expand_null_ifn): Update.
18658 * value-prof.c (gimple_divmod_fixed_value): Update.
18659 (gimple_divmod_fixed_value_transform): Update.
18660 (gimple_mod_pow2): Update.
18661 (gimple_mod_pow2_value_transform): Update.
18662 (gimple_mod_subtract): Update.
18663 (gimple_mod_subtract_transform): Update.
18664 (gimple_ic): Update.
18665 (gimple_stringop_fixed_value): Update.
18666 (gimple_stringops_transform): Update.
18667 * value-prof.h: Update.
18669 2017-06-29 Carl Love <cel@us.ibm.com>
18671 * config/rs6000/rs6000-c.c: Add support for built-in functions
18672 vector signed int vec_signed (vector float);
18673 vector signed long long vec_signed (vector double);
18674 vector signed int vec_signed2 (vector double, vector double);
18675 vector signed int vec_signede (vector double);
18676 vector signed int vec_signedo (vector double);
18677 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18678 instruction generator.
18679 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18680 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18681 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18683 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18684 vunsignede_v2df): Add define_expands.
18685 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18686 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18687 VEC_UNSIGNEDO): Add definitions.
18688 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18689 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18690 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18691 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18692 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18693 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18694 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18695 vec_unsignede, vec_unsignedo): Add builtin defines.
18696 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18698 * doc/extend.texi: Update the built-in documentation file for the
18699 new built-in functions.
18701 2017-06-29 Richard Biener <rguenther@suse.de>
18703 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18704 reduction chains to LOOP_VINFO_REDUCTIONS.
18705 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18706 SLP reductions after processing reduction chains.
18708 2017-06-29 Nathan Sidwell <nathan@acm.org>
18710 * builtins.c (fold_builtin_FUNCTION): Use
18711 lang_hooks.decl_printable_name.
18713 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18715 PR middle-end/81194
18716 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18717 with only one label.
18718 * stmt.c (expand_case): Assert NCASES is greater than one.
18720 2017-06-29 Richard Biener <rguenther@suse.de>
18722 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18724 (group_case_labels): Likewise.
18725 (find_taken_edge): Push sanity checking on val to workers...
18726 (find_taken_edge_cond_expr): ... here
18727 (find_taken_edge_switch_expr): ... and here, handle cases
18728 with just a default label.
18729 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18730 (group_case_labels): Likewise.
18731 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18732 group_case_labels does anything cleanup the CFG again.
18734 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18736 PR tree-optimization/81196
18737 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18738 exit condition comparing two IVs.
18740 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18742 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18743 profile to the dummy entry at the end of the list of architectures.
18744 * config/arm/arm-cpu-cdata.h: Regenerated.
18746 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18747 Michael Collison <michael.collison@arm.com>
18750 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18752 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18753 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18754 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18755 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18756 of shift when the shift amount is masked with constant equal to
18757 the size of the mode.
18758 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18761 2017-06-29 Martin Liska <mliska@suse.cz>
18763 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18765 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18768 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18771 (edge_growth_cache_entry::edge_growth_cache_entry): New
18773 (reset_edge_growth_cache): Update to use constructor.
18775 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18777 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18778 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18779 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18781 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18783 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18784 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18786 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18788 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18789 (*-linux-uclibc*): Add t-uclibc tmake_file.
18790 * config/t-musl: New.
18791 * config/t-uclibc: New.
18793 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18795 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18797 (gen_comm_data): Emit architectural setting of arch_prof.
18798 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18800 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18801 (armv8-m.base, armv8-m.main): Likewise.
18802 * arm-protos.h (arm_build_target): Add profile field.
18803 (arch_option): Likewise.
18804 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18806 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18807 arm_active_target.profile.
18809 2017-06-28 Richard Biener <rguenther@suse.de>
18811 PR middle-end/81227
18812 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18813 TYPE_OVERFLOW_WRAPS.
18814 * match.pd (negate_expr_p): Likewise.
18815 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18816 fold_build2, not fold_binary.
18818 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18820 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18821 Convert memory address to Pmode.
18822 (aarch64_print_operand): Assert MEM operands are always Pmode.
18824 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18827 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18828 Remove redundant if.
18829 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18830 * config/arm/aarch-common-protos.h
18831 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18832 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18834 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18837 * multiple_target.c (create_dispatcher_calls): Set the default
18838 clone to be static, not public.
18840 2017-06-28 Richard Biener <rguenther@suse.de>
18842 * tree-vect-loop.c (vectorizable_reduction): Move special
18843 cond reduction IV var creation ...
18844 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18845 parameter. Use STMT_VINFO_VECTYPE.
18846 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18849 2017-06-28 Martin Liska <mliska@suse.cz>
18852 * ipa-visibility.c (non_local_p): Handle visibility.
18854 2017-06-28 Martin Liska <mliska@suse.cz>
18857 * common.opt: Add IntegerRange to various options.
18858 * opt-functions.awk (integer_range_info): New function.
18859 * optc-gen.awk: Add integer_range_info to cl_options struct.
18860 * opts-common.c (decode_cmdline_option): Handle
18861 CL_ERR_INT_RANGE_ARG.
18862 (cmdline_handle_error): Likewise.
18863 * opts.c (print_filtered_help): Show valid interval in
18864 when --help is provided.
18865 * opts.h (struct cl_option): Add range_min and range_max fields.
18866 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18868 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18870 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18871 (x * C EQ/NE y * C): New transformation.
18873 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18875 * genmultilib (combination_space): Accept '+' in option names.
18877 2017-06-28 Martin Liska <mliska@suse.cz>
18880 * asan.c (instrument_derefs): Bail out inner references
18881 that are hard register variables.
18883 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18886 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18887 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18889 2017-06-28 Richard Biener <rguenther@suse.de>
18891 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18892 (vect_get_slp_defs): Adjust.
18893 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18895 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18897 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18898 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18899 (vectorizable_reduction): Adjust.
18900 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18902 (vect_get_slp_defs): Likewise.
18903 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18904 (vectorizable_bswap): Adjust.
18905 (vectorizable_call): Likewise.
18906 (vectorizable_conversion): Likewise.
18907 (vectorizable_assignment): Likewise.
18908 (vectorizable_shift): Likewise.
18909 (vectorizable_operation): Likewise.
18910 (vectorizable_store): Likewise.
18911 (vectorizable_condition): Likewise.
18912 (vectorizable_comparison): Likewise.
18914 2017-06-28 Michael Collison <michael.collison@arm.com>
18917 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18919 (arm_gen_movmemqi): Removed unused variable 'i'.
18920 Convert 'for' loop into 'while' loop.
18921 (arm_expand_prologue): Remove last unnecessary set of insn.
18922 (thumb_pop): Remove unused variable 'pushed_words'.
18923 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18925 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18927 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18928 * config/s390/s390-protos.h: Add prototype of
18929 s390_rel_address_ok_p.
18930 * config/s390/s390.c (s390_got_symbol): New function.
18931 (s390_rel_address_ok_p): New function.
18932 (legitimize_pic_address): Use s390_rel_address_ok_p.
18933 (s390_load_got): Use s390_got_symbol.
18934 (s390_option_override): Issue error if
18935 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18936 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18938 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18940 2017-06-27 Andrew Pinski <apinski@cavium.com>
18942 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18943 (X * copysign (1.0, X)): New pattern.
18944 (X * copysign (1.0, -X)): New pattern.
18945 (copysign (-1.0, CST)): New pattern.
18947 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18949 * genmultilib (combination_space): Remove variable.
18950 Validate reuse rules against regular expression for any sequence
18951 of multilib options in any order.
18953 2017-06-27 Michael Collison <michael.collison@arm.com>
18955 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18956 call aarch64_split_simd_combine.
18957 * (aarch64_combine_internal<mode>): Delete pattern.
18958 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18959 Allow register and subreg operands.
18961 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18963 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18964 specific need, just fallback on defaults.
18965 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18967 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18968 Olivier Hainque <hainque@adacore.com>
18970 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18972 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18973 targets. Pick a default if no particular attempt applied.
18974 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18977 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18979 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18980 (x86_64-wrs-vxworks7): Likewise.
18982 2017-06-27 Marek Polacek <polacek@redhat.com>
18985 * ubsan.c (instrument_null): Check get_base_address's result for null.
18987 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18989 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18991 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18993 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18994 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18995 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18996 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18997 New function types.
18998 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18999 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19000 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19001 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19002 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19003 BUILT_IN_FEUPDATEENV): New builtins.
19004 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19005 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19006 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19007 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19009 (builtin_structptr_types): Adjust size.
19010 * tree.c (builtin_structptr_types): Add four entries.
19012 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19013 Olivier Hainque <hainque@adacore.com>
19015 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19016 (TLS_SYM): New local macro, forcing reference to __tls__ on
19017 link command lines for VxWorks 7 RTPs, triggering initialization
19019 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19020 OS features TLS support, true for RTPs on VxWorks 7.
19021 * config/vxworks.c (vxworks_override_options): Setup emutls
19024 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19026 * predict.c (test_prediction_value_range): Use -1U instead of -1
19027 to avoid narrowing conversion warning.
19028 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19029 to avoid narrowing conversion warning.
19030 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19032 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19034 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19036 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19037 64bit configurations.
19038 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19039 (SIZE_TYPE): Likewise.
19040 * config/vxworks.c (vxworks_emutls_var_fields): Use
19041 long_unsigned_type_node instead of unsigned_type_node as the offset
19042 field type, which is "pointer" mode in emutls.c.
19044 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19047 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19049 PR middle-end/81207
19050 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19051 gimple_vuse copying separately from gimple_vdef copying.
19053 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19055 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19058 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19059 Olivier Hainque <hainque@adacore.com>
19061 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19062 for all vxworks7 targets.
19063 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19064 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19065 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19066 variations for VX6/VX7 and 32/64bits later on in ...
19067 (VXWORKS_LIB_SPEC): Leverage new macros.
19068 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19069 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19071 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19073 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19074 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19076 2017-06-26 Carl Love <cel@us.ibm.com>
19078 * config/rs6000/rs6000-c.c: Add support for built-in functions
19079 vector bool char vec_reve (vector bool char);
19080 vector signed char vec_reve (vector signed char);
19081 vector unsigned char vec_reve (vector unsigned char);
19082 vector bool int vec_reve (vector bool int);
19083 vector signed int vec_reve (vector signed int);
19084 vector unsigned int vec_reve (vector unsigned int);
19085 vector bool long long vec_reve (vector bool long long);
19086 vector signed long long vec_reve (vector signed long long);
19087 vector unsigned long long vec_reve (vector unsigned long long);
19088 vector bool short vec_reve (vector bool short);
19089 vector signed short vec_reve (vector signed short);
19090 vector double vec_reve (vector double);
19091 vector float vec_reve (vector float);
19092 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19093 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19094 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19095 (altivec_vreve): New pattern.
19096 * config/rs6000/altivec.h (vec_reve): New define.
19097 * doc/extend.texi (vec_rev): Update the built-in documentation file
19098 for the new built-in functions.
19100 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19102 PR tree-optimization/71815
19103 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19105 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19107 (slsr_process_phi): Likewise.
19108 (replace_uncond_cands_and_profitable_phis): Don't replace a
19109 multiply candidate with a stride of 1 (copy or cast).
19110 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19112 (lowest_cost_path): Likewise.
19113 (total_savings): Likewise.
19115 2017-06-26 Richard Biener <rguenther@suse.de>
19118 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19119 Use def_builtin_pure for all gather builtins.
19121 2017-06-26 Richard Biener <rguenther@suse.de>
19123 PR tree-optimization/81203
19124 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19125 non-dominating BBs.
19127 2017-06-26 Marek Polacek <polacek@redhat.com>
19130 * doc/invoke.texi: Document -Wmultistatement-macros.
19132 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19134 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19135 arm_neon_ok_no_float_abi effective target.
19137 2017-06-26 Richard Biener <rguenther@suse.de>
19139 PR tree-optimization/80928
19140 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19141 (copy_bbs): Set BB_DUPLICATED flag early.
19142 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19144 (execute_on_shrinking_pred): Likewise.
19145 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19146 BB_DUPLICATED blocks.
19147 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19148 iterate over all PHIs considering removal of *gsi.
19150 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19152 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19155 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19157 * config/rs6000/rs6000-string.c: (expand_block_clear,
19158 do_load_for_compare, select_block_compare_mode,
19159 compute_current_alignment, expand_block_compare,
19160 expand_strncmp_align_check, expand_strn_compare,
19161 expand_block_move, rs6000_output_load_multiple)
19162 Move functions related to string/block move/compare
19163 to a separate file.
19164 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19165 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19166 for this function which is now used in two files.
19167 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19168 * config.gcc: Add rs6000-string.o to extra_objs for
19169 targets powerpc*-*-* and rs6000*-*-*.
19171 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19174 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19175 32-bit, since indexed is not valid for DImode.
19176 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19177 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19178 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19179 (define_peephole2 for Altivec d-form store): Likewise.
19182 * multiple_target.c (create_dispatcher_calls): Only create the
19183 dispatcher call if the function is the default clone of a
19184 versioned function.
19186 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19188 PR middle-end/80902
19189 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19190 a call, force the call to not be a tail call.
19192 2017-06-23 Jeff Law <law@redhat.com>
19194 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19197 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19199 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19200 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19201 handling for early expansion of vector shifts (sl,sr,sra,rl).
19202 (builtin_function_type): Add vector shift right instructions
19203 to the unsigned argument list.
19205 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19207 rtl-optimizatoin/79286
19208 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19209 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19210 trap. PIC register plus a const unspec without offset can never trap.
19212 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19214 * tree.h (builtin_structptr_type): New type.
19215 (builtin_structptr_types): Declare new array.
19216 * tree.c (builtin_structptr_types): New array.
19217 (free_lang_data, build_common_tree_nodes): Use it.
19219 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19222 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19225 2017-06-22 Matt Turner <mattst88@gmail.com>
19227 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19228 Lake models to skylake case. Assume skylake for unknown
19229 models with clflushopt.
19231 2017-06-22 Jeff Law <law@redhat.com>
19233 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19234 frame sizes that do not satisfy aarch64_uimm12_shift.
19236 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19238 * profile-count.h (apply_probability,
19239 apply_scale, probability_in): Fix checks for zero.
19241 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19243 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19244 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19246 2016-06-22 Richard Biener <rguenther@suse.de>
19248 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19249 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19250 REDUC_MAX_EXPR support.
19251 (vectorizable_reduction): Likewise.
19252 (vect_create_epilog_for_reduction): Likewise.
19254 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19256 * match.pd (A / (1 << B) -> A >> B): New.
19257 * generic-match-head.c: Include optabs-tree.h.
19258 * gimple-match-head.c: Likewise.
19259 * optabs-tree.h (target_supports_op_p): New.
19260 * optabs-tree.c (target_supports_op_p): New.
19262 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19264 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19265 $gcc_cv_ld --help output.
19266 (gcc_cv_ld_demangle): Likewise.
19267 (gcc_cv_ld_eh_frame_hdr): Likewise.
19268 (gcc_cv_ld_pie): Likewise.
19269 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19270 (gcc_cv_ld_buildid): Likewise.
19271 (gcc_cv_ld_sysroot): Likewise.
19272 (ld_bndplt_support): Likewise.
19273 (ld_pushpopstate_support): Likewise.
19274 * configure: Regenerate.
19275 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19277 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19280 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19281 operands indexes to avoid gap between operands and match_dups.
19283 2017-06-21 Andrew Pinski <apinski@cavium.com>
19285 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19286 Increment Arith_shift and Arith_shift_reg by 1.
19287 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19289 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19290 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19291 (aarch64_strip_extend): Add new argument and test for it.
19292 (aarch64_cheap_mult_shift_p): New function.
19293 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19294 add a cost if it is true.
19295 Update calls to aarch64_strip_extend.
19296 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19298 2017-06-21 Andrew Pinski <apinski@cavium.com>
19300 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19302 (thunderxt88): Likewise.
19303 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19304 (thunderx_prefetch_tune): New variable.
19305 (thunderx2t99_prefetch_tune): Update for the correct values.
19306 (thunderxt88_tunings): New variable.
19307 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19308 generic_prefetch_tune.
19309 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19311 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19313 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19314 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19315 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19316 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19317 (aarch64_atomic_cas<mode>, GPI): Likewise.
19319 2017-06-21 Martin Liska <mliska@suse.cz>
19321 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19322 statements on cold and hot labels.
19323 * predict.c (tree_estimate_probability_bb): Remove the
19324 prediction from this place.
19326 2017-06-21 Martin Liska <mliska@suse.cz>
19328 PR tree-optimization/79489
19329 * gimplify.c (maybe_add_early_return_predict_stmt): New
19331 (gimplify_return_expr): Call the function.
19332 * predict.c (tree_estimate_probability_bb): Remove handling
19334 * predict.def: Update comment about early return predictor.
19335 * gimple-predict.h (is_gimple_predict): New function.
19336 * predict.def: Change default value of early return to 66.
19337 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19339 * passes.def: Put pass_strip_predict_hints to the beginning of
19342 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19344 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19345 FUNCTION_DECL declarations.
19346 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19348 (dwaf2out_decl): Likewise.
19349 * godump.c (go_early_global_decl): Skip call to the real debug hook
19350 for FUNCTION_DECL declarations.
19351 * passes.c (rest_of_decl_compilation): Skip call to the
19352 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19353 -fdump-go-spec is passed.
19355 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19357 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19358 Reorder for compactness.
19359 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19360 (def_builtin_pure, def_builtin_pure2): New functions.
19361 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19363 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19365 * match.pd (nop_convert): New predicate.
19366 ((A +- CST1) +- CST2): Allow some NOP conversions.
19368 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19371 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19372 with ctors/dtors if GOVD_SHARED is set.
19374 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19376 * config/aarch64/aarch64.md (movti_aarch64):
19377 Emit mov rather than orr.
19378 (movtf_aarch64): Likewise.
19379 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19380 Emit mov rather than orr.
19382 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19384 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19385 Swap alternatives, make integer dup more expensive.
19387 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19389 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19390 Return true for non-tls symbols.
19392 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19394 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19395 (cortex-a75): Likewise.
19396 (cortex-a75.cortex-a55): Likewise.
19397 * config/aarch64/aarch64-tune.md: Regenerate.
19398 * doc/invoke.texi (-mtune): Document new values for -mtune.
19400 2017-06-21 Tom de Vries <tom@codesourcery.com>
19402 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19403 stack_size feature.
19404 (Effective-Target Keywords, Other attributes): Suggest using
19405 dg-add-options stack_size feature to get stack limit in stack_size
19406 effective target documentation.
19408 2017-06-21 Julian Brown <julian@codesourcery.com>
19409 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19411 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19412 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19413 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19415 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19416 attribute type list for neon_multiply.
19417 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19418 attribute type list for neon_multiply.
19419 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19420 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19421 attribute type list for neon_multiply.
19422 * config/arm/types.md (crypto_pmull): Add.
19423 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19424 attribute type list.
19426 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19428 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19431 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19433 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19434 to make sure not to dereference a NULL cost_classes_ptr pointer.
19436 2017-06-20 Carl Love <cel@us.ibm.com>
19438 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19439 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19440 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19441 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19442 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19443 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19444 VMULOSW): New enum "unspec" values.
19445 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19446 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19447 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19448 altivec_vmulosw): New patterns.
19449 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19450 VMULOSW): Add definitions.
19452 2017-06-20 Julia Koval <julia.koval@intel.com>
19454 * config/i386/i386.c: Fix rounding expand for new pattern.
19455 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19457 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19459 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19460 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19462 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19464 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19467 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19469 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19470 architecture, then by implementer ID.
19471 * config/aarch64/aarch64-tune.md: Regenerate.
19473 2017-06-20 Richard Biener <rguenther@suse.de>
19475 PR middle-end/81097
19476 * fold-const.c (split_tree): Fold to type before negating.
19478 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19480 * diagnostic-show-locus.c
19481 (selftest::test_fixit_deletion_affecting_newline): New function.
19482 (selftest::diagnostic_show_locus_c_tests): Call it.
19484 2017-06-20 Andreas Schwab <schwab@suse.de>
19487 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19490 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19492 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19493 __ARM_FEATURE_COPROC according to support.
19495 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19497 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19498 Rewritten to avoid overflow for > 32-bit pointers.
19501 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19502 by removing enum keyword.
19503 (ubsan_type_descriptor): Likewise. Formatting fix.
19506 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19507 splitter): Require TARGET_SSE2 in the condition.
19509 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19512 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19513 for doing vector set of SFmode on ISA 3.0.
19514 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19515 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19517 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19518 SFmode value into a V4SF variable that was extracted from another
19519 V4SF variable without converting the element to double precision
19520 and back to single precision vector format.
19521 (vsx_insert_extract_v4sf_p9_2): Likewise.
19523 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19525 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19526 in UWHI to avoid undefined overflow.
19529 * ubsan.h (enum ubsan_encode_value_phase): New.
19530 (ubsan_encode_value): Change second argument to
19531 enum ubsan_encode_value_phase with default value of
19532 UBSAN_ENCODE_VALUE_GENERIC.
19533 * ubsan.c (ubsan_encode_value): Change second argument to
19534 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19535 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19536 create_tmp_var_raw instead of create_tmp_var and use a
19538 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19539 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19540 ubsan_encode_value callers.
19543 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19544 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19545 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19547 2017-06-19 Richard Biener <rguenther@suse.de>
19549 PR middle-end/81118
19550 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19551 estimates if we changed anything.
19553 2017-06-19 Richard Biener <rguenther@suse.de>
19555 PR tree-optimization/80887
19556 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19557 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19558 simplified lookups, then reset mprts_hook.
19559 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19561 (try_to_simplify): Likewise.
19563 2017-06-19 Martin Liska <mliska@suse.cz>
19566 * gimplify.c (gimplify_switch_expr):
19567 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19569 2017-06-19 Martin Liska <mliska@suse.cz>
19571 * doc/install.texi: Document that PGO runs in 4 stages.
19573 2017-06-19 Martin Liska <mliska@suse.cz>
19576 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19577 to dispatcher function name.
19578 * multiple_target.c (replace_function_decl): New function.
19579 (create_dispatcher_calls): Redirect both edges and references.
19581 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19583 * profile-count.c (profile_count::dump): Dump quality.
19584 (profile_count::differs_from_p): Update for unsigned val.
19585 * profile-count.h (profile_count_quality): New enum.
19586 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19588 2017-06-19 Richard Biener <rguenther@suse.de>
19590 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19591 struct function as arg.
19592 (estimate_numbers_of_iterations): Export overload with loop arg.
19593 (free_numbers_of_iterations_estimates_loop): Use an overload of
19594 free_numbers_of_iterations_estimates instead.
19595 * tree-cfg.c (remove_bb): Adjust.
19596 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19597 * tree-parloops.c (gen_parallel_loop): Likewise.
19598 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19600 (tree_unroll_loops_completely): Likewise.
19601 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19602 Use an overload instead and export.
19603 (estimated_loop_iterations): Adjust.
19604 (max_loop_iterations): Likewise.
19605 (likely_max_loop_iterations): Likewise.
19606 (estimate_numbers_of_iterations): Take struct function as arg
19608 (loop_exits_before_overflow): Adjust.
19609 (free_numbers_of_iterations_estimates_loop): Use an overload.
19610 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19611 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19613 2017-06-19 Richard Biener <rguenther@suse.de>
19616 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19617 RANGE_EXPR conservatively.
19619 2017-06-16 Carl Love <cel@us.ibm.com>
19621 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19622 definitions for vec_float, vec_float2, vec_floato,
19623 vec_floate built-ins.
19624 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19625 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19627 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19628 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19629 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19630 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19631 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19632 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19633 vec_floato): Add builtin defines.
19634 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19635 Update the built-in documentation file for the new built-in
19638 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19640 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19641 (mthumb): Mark as the negative of -marm.
19643 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19645 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19647 (ARM Options, -mtune): Document that this accepts the same
19648 extension options as -mcpu.
19649 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19651 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19653 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19654 permitted extensions.
19656 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19658 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19659 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19660 (armv8-m.main): Add option +nodsp.
19661 * config/arm/arm-cpu-cdata.h: Regenerated.
19663 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19665 * config/arm/t-fuchsia: New file.
19666 * config.gcc (arm*-*-fuchsia*): Use it.
19668 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19670 * config/arm/t-symbian: Rewrite for new option infrastructure.
19672 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19674 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19675 (MULTILIB_REQUIRED): Likewise.
19677 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19679 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19680 (MULTILIB_RESUE): Likewise.
19681 (MULTILIB_MATCHES): Likewise.
19682 (MULTLIB_REQUIRED): Likewise.
19684 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19686 * config/arm/t-rtems: Rewrite for new option framework.
19688 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19690 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19691 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19692 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19693 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19694 * config/arm/t-multilib: ... here.
19695 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19696 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19697 armv7-a and armv8*-a when A-profile libraries have not been built.
19698 * config/arm/t-rmprofile: Rewrite.
19700 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19702 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19703 with a backslash. Remove the backslash after substituting unescaped
19705 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19707 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19709 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19710 the driver to pass through the default setting of -mfloat-abi.
19711 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19713 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19714 (all_feat_combs): New rule.
19715 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19717 * config/arm/t-aprofile: Rewrite.
19719 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19721 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19722 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19723 fpu is not specified by the user/command-line.
19724 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19725 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19726 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19727 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19728 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19729 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19731 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19733 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19734 * config/arm/t-arm-elf: Rewritten.
19736 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19738 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19739 have some floating-point instructions.
19740 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19741 (TARGET_MAYBE_HARD_FLOAT): New macro.
19742 * config/arm/arm-builtins.c (arm_init_builtins): Use
19743 TARGET_MAYBE_HARD_FLOAT.
19744 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19746 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19748 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19749 (configargs.h): Include it.
19750 (arm_print_hint_for_fpu_option): New function.
19751 (arm_parse_fpu_option): New function.
19752 (candidate_extension): New class.
19753 (arm_canon_for_multilib): New function.
19754 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19755 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19756 (ARCH_CANONICAL_SPECS): New macro.
19757 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19759 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19761 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19762 are set after handling multilib fragments. Set target_cpu_default2
19765 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19767 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19769 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19770 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19772 (check_arch): Likewise.
19773 * config/arm/arm.c (arm_configure_build_target): Handle
19774 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19775 options in the default.
19777 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19779 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19780 when an option is an alias of another.
19781 * config/arm/parsecpu.awk (optalias): New parser token.
19782 (gen_comm_data): Mark non-alias options as such. Emit entries
19783 for extension aliases.
19784 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19785 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19786 (armv6kz, armv6zk, armv6t2): Likewise.
19787 (armv7): Make vfpv3-d16 an alias.
19788 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19790 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19791 Sort in canonical order.
19792 (armv8-a): Sort in canonical order.
19793 (armv8.1-a, armv8.2-a): Likewise.
19794 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19796 (cortex-a9): Sort in canonical order.
19797 * config/arm/arm.c (selftests.h): Include it.
19798 (arm_test_cpu_arch_data): New function.
19799 (arm_run_self_tests): New function.
19800 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19801 (targetm): Move declaration to the end of the file.
19802 * arm-cpu-cdata.h: Regenerated.
19804 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19806 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19807 call to target_mode_check describing the type of option passed.
19808 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19809 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19810 arm_parse_cpu_option_name to match parameters against list of
19812 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19813 arm_arch_core_flags data structure.
19814 * config/arm/arm-cpu_cdata.h: Regenerated.
19816 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19818 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19820 (arm_print_hint_for_cpu_option): Likewise.
19821 (arm_print_hint_for_arch_option): Likewise.
19822 (arm_parse_cpu_option_name): Likewise.
19823 (arm_parse_arch_option_name): Likewise.
19824 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19825 of entries in the all_fpus list.
19826 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19827 (arm_parse_cpu_option_name): Declare.
19828 (arm_parse_arch_option_name): Declare.
19829 (arm_parse_option_features): Declare.
19830 (arm_intialize_isa): Declare.
19831 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19833 (gen_comm_data): ... here. Make definitions non-static.
19834 * config/arm/arm-cpu-data.h: Regenerated.
19835 * config/arm/arm-cpu-cdata.h: Regenerated.
19837 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19839 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19840 (cpu_arch_extension): New structure.
19841 (cpu_arch_option, arch_option, cpu_option): New structures.
19842 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19843 architecture types.
19844 (gen_data): Generate new format data tables.
19845 * config/arm/arm.c (cpu_tune): New structure.
19846 (cpu_option, processors): Delete.
19847 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19848 (arm_print_hint_for_cpu_option): ... this and ...
19849 (arm_print_hint_for_arch_option): ... this.
19850 (arm_parse_arch_cpu_name): Delete. Replace with ...
19851 (arm_parse_cpu_option_name): ... this and ...
19852 (arm_parse_arch_option_name): ... this.
19853 (arm_unrecognized_feature): Change type of target parameter to
19855 (arm_parse_arch_cpu_features): Delete. Replace with ...
19856 (arm_parse_option_features): ... this.
19857 (arm_configure_build_target): Rework to use new configuration data
19859 (arm_print_tune_info): Rework for new configuration data tables.
19860 * config/arm/arm-cpu-data.h: Regenerated.
19861 * config/arm/arm-cpu.h: Regenerated.
19863 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19865 * Makefile.in (OBJS): Move sbitmap.o from here ...
19866 (OBJS-libcommon): ... to here.
19868 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19870 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19871 (ISA_ALL_CRYPTO): New macro.
19872 (ISA_ALL_SIMD): New macro
19873 (ISA_ALL_FP): New macro.
19874 * config/arm/arm.c (fpu_bitlist): Update initializer.
19875 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19877 (arm9e): Add fpu. Add option for nofp
19878 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19879 (arm926ej-s, arm1026ej-s): Likewise.
19880 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19881 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19882 neon-fp16, neon-vfpv4, nofp and nosimd.
19883 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19884 (cortex-a8): Add fpu. Add option for nofp.
19885 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19886 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19887 (cortex-r4f): Add fpu.
19888 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19889 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19891 (cortex-r8): Likewise.
19892 (cortex-m4): Add fpu. Add option for nofp.
19893 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19894 (cortex-a17.cortex-a7): Likewise.
19895 (cortex-a32): Add fpu. Add options for crypto and nofp.
19896 (cortex-a35, cortex-a53): Likewise.
19897 (cortex-a57): Add fpu. Add option for crypto.
19898 (cortex-a72, cortex-a73): Likewise.
19899 (exynos-m1): Likewise.
19900 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19901 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19902 (cortex-m33): Add fpu. Add option for nofp.
19903 * config/arm/arm-cpu-cdata.h: Regenerated
19904 * config/arm/arm-cpu-data.h: Regenerated.
19906 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19908 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19909 (armv5te, armv5tej): Likewise.
19910 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19911 (armv7): Add options fp and vfpv3-d16.
19912 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19913 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19915 (armv7ve): Likewise.
19916 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19917 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19918 (armv8-a): Add nocrypto option.
19919 (armv8.1-a, armv8.2-a): Likewise.
19920 (armv8-m.main): add options fp, fp.dp and nofp.
19922 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19924 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19926 (armv8-a+crc): Delete.
19927 (armv8.1-a): Add options simd, crypto and nofp.
19928 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19929 (armv8.2-a+fp16): Delete.
19930 (armv8-m.main): Add option dsp.
19931 (armv8-m.main+dsp): Delete.
19932 (cortex-a8): Add fpu. Add nofp option.
19933 (cortex-a9): Add fpu. Add nofp and nosimd options.
19934 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19935 link to main cpu and architecture data structures.
19936 (gen_comm_data): Only put isa attributes from the main architecture
19938 (option): New statement for architecture and CPU entries.
19939 * arm.c (struct cpu_option): New structure.
19940 (struct processors): Add entry for options.
19941 (arm_unrecognized_feature): New function.
19942 (arm_parse_arch_cpu_name): Ignore any characters after the first
19944 (arm_parse_arch_cpu_feature): New function.
19945 (arm_configure_build_target): Separate out any CPU and architecture
19946 features and parse separately. Don't error out if -mfpu=auto is
19947 used with only an architecture string.
19948 (arm_print_asm_arch_directives): New function.
19949 (arm_file_start): Call it.
19950 * config/arm/arm-cpu-cdata.h: Regenerated.
19951 * config/arm/arm-cpu-data.h: Likewise.
19952 * config/arm/arm-tables.opt: Likewise.
19954 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19956 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19957 assembler when it is not -mfpu=auto.
19959 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19961 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19962 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19963 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19964 (ASM_CPU_SPEC): Rewrite.
19965 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19966 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19967 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19968 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19969 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19970 copied string is NUL-terminated. Also strip any characters prefixed
19972 (arm_rewrite_selected_arch): New function.
19973 (arm_rewrite_march): New function.
19975 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19977 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19978 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19979 (march, mcpu, mtune): Convert to string-based options.
19980 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19981 (arm_parse_arch_cpu_name): New function.
19982 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19983 identify selected architecture or CPU.
19984 (arm_option_save): New function.
19985 (TARGET_OPTION_SAVE): Redefine.
19986 (arm_option_restore): Restore string options.
19987 (arm_option_print): Print string options.
19989 2017-06-16 Martin Sebor <msebor@redhat.com>
19991 PR tree-optimization/80933
19992 PR tree-optimization/80934
19993 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19994 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19995 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19996 (gimple_fold_builtin): Call them.
19998 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20000 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20001 as unlikely; update profile.
20003 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20005 * predict.c (force_edge_cold): Handle declaring edges impossible
20008 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20010 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20012 (try_unroll_loop_completely): Fix reporting.
20014 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20016 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20018 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20021 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20022 if given a non-constant argument for an intrinsic which requires a
20025 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20027 * profile.c (compare_freqs): New function.
20028 (branch_prob): Sort edge list.
20029 (find_spanning_tree): Assume that the list is priority sorted.
20031 2017-06-16 Richard Biener <rguenther@suse.de>
20033 PR tree-optimization/81090
20034 * passes.def (pass_record_bounds): Remove.
20035 * tree-pass.h (make_pass_record_bounds): Likewise.
20036 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20037 make_pass_record_bounds): Likewise.
20038 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20039 not free niter estimates at the beginning but at the end.
20040 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20042 2017-06-16 Richard Biener <rguenther@suse.de>
20044 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20045 initializer to workaround ICE in host GCC 4.8.
20047 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20049 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20051 (clone_inlined_nodes): Update.
20053 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20055 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20056 prefetch settings, and enable prefetching by default at -O3.
20058 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20060 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20061 Set flag_prefetch_loop_arrays according to tuning data.
20063 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20065 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20066 New tune structure.
20067 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20068 [Unrelated to main purpose of the patch] Place the pointer field last
20069 to enable type checking errors when tune structure are wrongly merged.
20070 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20071 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20072 (thunderx2t99_prefetch_tune): New tune constants.
20073 (tune_params *_tunings): Update all tunings (no functional change).
20074 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20075 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20076 from tunings structures.
20078 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20081 * ubsan.c (instrument_null): Add T argument, use it instead
20082 of computing it based on IS_LHS.
20083 (instrument_object_size): Likewise.
20084 (pass_ubsan::execute): Adjust instrument_null and
20085 instrument_object_size callers to pass gimple_get_lhs or
20086 gimple_assign_rhs1 result to it. Use instrument_null instead of
20087 calling get_base_address and instrument_mem_ref. Handle
20088 aggregate call arguments for object-size sanitization.
20090 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20092 PR tree-optimization/81089
20093 * tree-vrp.c (is_masked_range_test): Validate operands of
20096 2017-06-15 Martin Sebor <msebor@redhat.com>
20099 * dumpfile.c (dump_register): Avoid calling memset to initialize
20100 a class with a default ctor.
20101 * gcc.c (struct compiler): Remove const qualification.
20102 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20103 * hash-table.h: Ditto.
20104 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20106 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20107 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20109 * params.h (struct param_info): Make struct members non-const.
20110 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20111 with default initialization.
20112 * vec.h (vec_copy_construct, vec_default_construct): New helper
20114 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20115 with vec_copy_construct.
20116 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20117 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20118 * doc/invoke.texi (-Wclass-memaccess): Document.
20120 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20122 * emit-rtl.h (is_leaf): Update comment about local
20123 register allocator.
20125 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20128 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20129 for a variable to have a section before checking if the section has a
20131 Set section to.persistent if persistent attribute is set.
20132 Warn if .persistent attribute is used on an automatic variable.
20134 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20136 PR rtl-optimization/80474
20137 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20139 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20141 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20144 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20146 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20147 (casesi): Emit bounds checking as RTL.
20148 (casesi_internal_mips16_<mode>): Remove bounds checking.
20150 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20152 * config/xtensa/xtensa.c (xtensa_option_override): Append
20153 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20154 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20155 xtensa_doloop_hooks): Define unconditionally.
20156 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20158 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20159 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20160 for it in xtensa_option_override.
20161 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20162 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20164 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20166 * doc/cppopts.texi: Document '-' special value to -MF.
20168 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20170 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20171 (cortex_a53_fconst): Likewise.
20172 (cortex_a53_fpmul): Likewise.
20173 (cortex_a53_f_load_64): Likewise.
20174 (cortex_a53_f_load_many): Likewise.
20175 (cortex_a53_advsimd_alu): Likewise.
20176 (cortex_a53_advsimd_alu_q): Likewise.
20177 (cortex_a53_advsimd_mul): Likewise.
20178 (cortex_a53_advsimd_mul_q): Likewise.
20179 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20180 Add missing fmul, r2f_cvt and fconst cases.
20182 2017-06-14 Richard Biener <rguenther@suse.de>
20184 PR middle-end/81088
20185 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20187 (fold_binary_loc): When associating do not treat pre-existing
20188 TREE_OVERFLOW on literal constants as a reason to allow
20189 TREE_OVERFLOW on associated literal constants.
20191 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20193 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20194 (MASK_FEATURES): New macro.
20195 * config/sparc/sparc.c (sparc_option_override): Remove the special
20196 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20198 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20200 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20201 a division of 0 if non-call exceptions are enabled.
20203 2017-06-14 Andrew Pinski <apinski@cavium.com>
20204 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20207 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20208 Improve vector initialization code gen for only variable case.
20210 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20212 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20214 2017-06-14 Richard Biener <rguenther@suse.de>
20216 PR tree-optimization/81083
20217 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20220 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20222 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20223 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20224 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20225 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20226 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20227 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20229 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20231 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20232 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20234 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20236 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20238 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20240 * config/rs6000/t-rtems: Don't handle SPE.
20242 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20244 * config/rs6000/t-linux: Don't handle SPE.
20246 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20248 * config/rs6000/eabispe.h: Delete file.
20250 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20252 * config/rs6000/t-spe: Delete file.
20254 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20256 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20257 (rs6000_legitimate_offset_address_p): Return false for anything in
20258 V2SImode or V2SFmode.
20260 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20262 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20263 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20264 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20266 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20267 (rs6000_legitimate_offset_address_p): Ditto.
20268 (rs6000_emit_move): Ditto.
20269 (rs6000_init_builtins): Remove V4HI_type_node.
20271 2017-06-13 Martin Liska <mliska@suse.cz>
20274 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20275 (gate_asan): Likewise.
20276 * asan.h (asan_no_sanitize_address_p): Remove the function.
20277 (sanitize_flags_p): New function.
20278 * builtins.def: Fix coding style.
20279 * common.opt: Use renamed enum value.
20280 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20281 * doc/extend.texi: Document no_sanitize attribute.
20282 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20283 to SANITIZE_UNDEFINED_NONDEFAULT.
20284 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20285 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20286 Use sanitize_flags_p.
20287 * gimplify.c (gimplify_function_tree): Likewise.
20288 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20289 * opts.c (parse_no_sanitize_attribute): New function.
20290 (common_handle_option): Use renamed enum value.
20291 * opts.h (parse_no_sanitize_attribute): Declare.
20292 * tree.c (sanitize_flags_p): New function.
20293 * tree.h: Declared here.
20294 * tsan.c: Use sanitize_flags_p.
20295 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20296 (instrument_mem_ref): Likewise.
20297 (instrument_bool_enum_load): Likewise.
20298 (do_ubsan_in_current_function): Remove the function.
20299 (pass_ubsan::execute): Use sanitize_flags_p.
20300 * ubsan.h: Remove do_ubsan_in_current_function
20301 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20302 (dump_function_to_file): Use it here.
20304 2017-06-13 Martin Jambor <mjambor@suse.cz>
20306 PR tree-optimization/80803
20307 PR tree-optimization/81063
20308 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20309 (propagate_subaccesses_across_link): Enqueue subtree whenever
20310 necessary instead of relying on the caller.
20312 2017-06-13 Martin Jambor <mjambor@suse.cz>
20314 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20315 that have a first_link.
20316 (sort_and_splice_var_accesses): Do not check first_link before
20318 (subtree_mark_written_and_enqueue): Likewise.
20319 (propagate_all_subaccesses): Likewise and do not stop at first
20320 parent with a first_link.
20322 2017-06-13 Martin Jambor <mjambor@suse.cz>
20324 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20327 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20329 * match.pd: New pattern.
20331 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20333 * tree-vrp.c (is_masked_range_test): New function.
20334 (register_edge_assert_for): Determine ranges for
20337 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20339 PR tree-optimization/67328
20340 * fold-const.c (maskable_range_p): New function.
20341 (build_range_check): Generate bittests if possible.
20343 2017-06-13 Martin Liska <mliska@suse.cz>
20345 * gimple-pretty-print.c (dump_probability): Add new argument.
20346 (dump_edge_probability): Dump both probability and count.
20347 (dump_gimple_label): Likewise.
20348 (dump_gimple_bb_header): Likewise.
20350 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20353 * config/avr/avr-devices.c: Fix indentation.
20354 * config/avr/gen-avr-mmcu-specs.c: Dito.
20356 2017-06-13 Richard Biener <rguenther@suse.de>
20358 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20359 instead get vector type from stmt_info.
20360 (vectorizable_reduction): Adjust. Remove dead code.
20362 2017-06-13 Richard Biener <rguenther@suse.de>
20364 PR middle-end/81065
20365 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20366 case of C * (x * C2 + C3).
20367 (fold_addr_of_array_ref_difference): Properly fold index difference.
20369 2017-06-12 David S. Miller <davem@davemloft.net>
20372 * config/sparc/sparc.md (return expander): Emit frame blockage if
20373 function uses alloca.
20375 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20377 * combine.c (make_field_assignment): Check len rather than the mode
20378 precision when calling force_to_mode.
20380 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20382 Support multilibs and devices that see flash in RAM address range.
20385 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20386 (avr_mcu_t) <flash_pm_offset>: New field.
20387 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20388 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20389 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20390 (AVR_TINY_PM_OFFSET): Remove macro.
20391 * config/avr/avr.opt (-mshort-calls): New option.
20392 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20393 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20394 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20395 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20396 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20397 instead of avr_arch->have_jmp_call.
20398 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20399 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20400 avr_arch->flash_pm_offset to define.
20401 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20402 new field flash_pm_offset. Add entry for avrxmega3.
20403 (avr_texinfo): Add entry for avrxmega3.
20404 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20405 attiny212, attiny214,
20406 attiny412, attiny414, attiny416, attiny417,
20407 attiny814, attiny816, attiny817,
20408 attiny1614, attiny1616, attiny1617,
20409 attiny3214, attiny3216, attiny3217.
20410 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20411 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20412 (avr_print_operand_address) [AVR_TINY]: Same.
20413 (avr_asm_init_sections) <readonly_data_section>: Only patch
20414 callback if avr_arch->flash_pm_offset = 0.
20415 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20416 for rodata if avr_arch->flash_pm_offset != 0.
20417 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20418 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20419 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20420 (m_options): Append opt_rcall.
20421 (m_dirnames): Append dir_rcall.
20422 * config/avr/t-multilib: Regenerate.
20424 * configure.ac [target=avr]: Check whether avrxmega3 default
20425 linker description file works as needed.
20426 * configure: Regenerate.
20427 * doc/avr-mmcu.texi: Regenerate.
20428 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20429 <__AVR_ARCH__>: Document avrxmega3 and 103.
20430 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20431 <__AVR_SHORT_CALLS__>: Document it.
20432 <__AVR_PM_BASE_ADDRESS__>: Document it.
20433 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20434 (AVR Variable Attributes) <progmem>: Document this is
20435 not needed for avrxmega3.
20436 (AVR Named Address Spaces) <__flash>: Dito.
20438 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20440 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20442 2017-06-12 Doug Rupp <rupp@adacore.com>
20444 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20445 Append vxworks-stdint.h to the tm_file list.
20446 * config/vxworks-stdint.h: New file.
20448 2017-06-12 Martin Liska <mliska@suse.cz>
20450 PR tree-optimization/81041
20451 * tree-profile.c (gimple_gen_ic_func_profiler):
20452 Create an extra BB in profile-generate
20453 (gimple_gen_time_profiler): Likewise.
20455 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20457 PR tree-optimization/81003
20458 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20459 (update_range_test): Use it instead of force_gimple_operand_gsi.
20461 2017-06-12 Richard Biener <rguenther@suse.de>
20463 PR tree-optimization/81053
20464 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20465 with backedge value not defined in loop. Simplify def stmt
20468 2017-06-11 Tom de Vries <tom@codesourcery.com>
20471 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20473 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20474 nvptx_cannot_force_const_mem.
20476 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20478 * opts.c (finish_options): Move test for flag_split_stack after
20479 it has been initialized.
20481 2017-06-11 Jason Merrill <jason@redhat.com>
20483 * tree.h (id_equal): New.
20484 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20485 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20486 instead of strcmp of IDENTIFIER_POINTER.
20488 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20490 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20491 (mark_all_inlined_calls_cdtor): Fix formating.
20492 (inline_transform): Rescale profile before inlining.
20494 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20496 * cgraph.h (cgraph_edge::clone): Update prototype.
20497 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20498 (cgraph_node::create_clone): Update.
20499 (cgraph_node::create_version_clone): Update.
20500 * tree-inline.c (copy_bb): Update.
20501 (expand_call_inline): Update.
20503 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20505 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20506 factored out from ...
20507 (rs6000_emit_prologue): ... here.
20509 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20511 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20512 factored out from ...
20513 (rs6000_emit_prologue): ... here.
20515 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20517 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20519 (handle_missing_profiles): Fix computation of tp_first_run.
20520 (counts_to_freqs): Do not touch freqs when count is 0.
20522 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20524 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20527 2017-06-10 Tom de Vries <tom@codesourcery.com>
20529 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20530 attributes): Document signal effective target.
20532 2017-06-10 Tom de Vries <tom@codesourcery.com>
20534 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20535 Document effective target stack_size.
20537 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20539 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20540 to the edit_context if they can be auto-applied.
20542 2017-06-9 Ian Lance Taylor <iant@golang.org>
20544 * opts.c (finish_options): If -fsplit-stack, disable implicit
20545 -forder-blocks-and-partition.
20546 * doc/invoke.texi (Optimize Options): Document that when using
20547 -fsplit-stack -forder-blocks-and-partition is not implicitly
20550 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20552 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20553 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20554 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20555 * builtins.def (abort, trap, unreachable): Declare cold.
20556 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20557 * tree-core.h (ECF_COLD): New.
20558 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20559 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20561 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20563 * predict.c (unlikely_executed_stmt_p): Cleanup.
20565 2017-06-09 Richard Biener <rguenther@suse.de>
20567 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20568 model if the ref is always written to.
20570 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20572 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20574 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20576 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20579 2017-06-09 Tom de Vries <tom@codesourcery.com>
20582 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20583 "target cannot support label values" when encountering LABEL_REF.
20585 2017-06-09 Martin Liska <mliska@suse.cz>
20587 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20588 (gimple_gen_ic_func_profiler): Emit direct comparison
20589 of __gcov_indirect_call_callee with NULL.
20590 (gimple_gen_time_profiler): Change probability from
20591 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20593 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20595 * profile.c (edge_gcov_counts): Turn to pointer.
20596 (compute_branch_probabilities, compute_branch_probabilities): Update.
20597 (branch_prob): Do not clear edge_gcov_count.
20598 * profile.h (edge_gcov_counts): Turn to pointer.
20599 (edge_gcov_count): Update.
20601 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20603 * gimple.h (gimple_check_failed): Mark cold.
20605 2017-06-09 Richard Biener <rguenther@suse.de>
20607 PR tree-optimization/66623
20608 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20609 refactor check_reduction into two parts, properly computing
20610 whether we have to check reduction validity for outer loop
20613 2017-06-09 Richard Biener <rguenther@suse.de>
20615 PR tree-optimization/79483
20616 * graphite-scop-detection.c (order): New global.
20617 (get_order): Compute bb to order mapping that satisfies code
20618 generation constraints.
20619 (cmp_pbbs): New helper.
20620 (build_scops): Start domwalk at entry block, sort generated
20623 2017-06-09 Richard Biener <rguenther@suse.de>
20625 PR middle-end/81007
20626 * ipa-polymorphic-call.c
20627 (ipa_polymorphic_call_context::restrict_to_inner_class):
20628 Skip FIELD_DECLs with error_mark_node type.
20629 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20632 2017-06-09 Martin Liska <mliska@suse.cz>
20634 * predict.c (struct branch_predictor): New struct.
20635 (test_prediction_value_range): New test.
20636 (predict_c_tests): New function.
20637 * selftest-run-tests.c (selftest::run_tests): Run the function.
20638 * selftest.h: Declare new tests.
20640 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20643 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20644 gen_add3_insn did not fail.
20645 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20646 r0, construct that number in a temporary reg and add that reg to r0.
20647 If asked to put the result in r0 as well, fail.
20649 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20651 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20652 for early expansion of vec_eqv.
20654 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20656 PR middle-end/81005
20657 * ubsan.c (instrument_null): Avoid pointless code temporary.
20658 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20660 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20664 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20667 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20670 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20672 * predict.c (unlikely_executed_edge_p): Move ahead.
20673 (probably_never_executed_edge_p): Use it.
20675 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20677 PR middle-end/79988
20678 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20679 gimple_call_builtin_p call.
20681 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20683 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20684 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20685 rtl_check_failed_type2, rtl_check_failed_code1,
20686 rtl_check_failed_code2, rtl_check_failed_code_mode,
20687 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20688 rtvec_check_failed_bounds, rtl_check_failed_flag,
20689 _fatal_insn_not_found, _fatal_insn): Likewise.
20690 * tree.h (tree_contains_struct_check_failed,
20691 tree_check_failed, tree_not_check_failed,
20692 tree_class_check_failed, tree_range_check_failed,
20693 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20694 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20695 tree_operand_check_failed, omp_clause_check_failed,
20696 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20699 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20701 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20702 flag_branch_probabilities.
20703 * ipa-inline.c (edge_badness): Likewise.
20704 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20705 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20706 * predict.c (maybe_hot_frequency_p): Likewise.
20707 (probably_never_executed): Likewise.
20708 * sched-ebb.c (schedule_ebbs): Likewise.
20709 * sched-rgn.c (find_single_block_region): Likewise.
20710 * tracer.c (tail_duplicate): Likewise.
20712 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20714 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20715 longer requires x_flag_profile_use.
20717 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20719 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20720 instead of flag_reorder_blocks_and_partition.
20721 * dbxout.c (dbxout_function_end): Likewise.
20722 * dwarf2out.c (gen_subprogram_die): Likewise.
20723 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20724 * hw-doloop.c (reorg_loops): Likewise.
20725 * varasm.c (assemble_start_function,
20726 assemble_end_function): Likewise.
20727 (decide_function_section): Do not check for
20728 flag_reorder_blocks_and_partition.
20730 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20732 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20734 (chkp_get_hard_register_fake_addr_expr): Ditto.
20735 (chkp_build_addr_expr): Add check for hard reg case.
20736 (chkp_parse_array_and_component_ref): Ditto.
20737 (chkp_find_bounds_1): Ditto.
20738 (chkp_process_stmt): Don't generate bounds store for
20741 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20743 * predict.c (maybe_hot_bb_p): Do not check profile status.
20744 (maybe_hot_edge_p): Likewise.
20745 (probably_never_executed): Check for zero counts even if profile
20747 (unlikely_executed_edge_p): New function.
20748 (unlikely_executed_stmt_p): New function.
20749 (unlikely_executed_bb_p): New function.
20750 (set_even_probabilities): Use unlikely predicates.
20751 (combine_predictions_for_bb): Likewise.
20752 (predict_paths_for_bb): Likewise.
20753 (predict_paths_leading_to_edge): Likewise.
20754 (determine_unlikely_bbs): New function.
20755 (estimate_bb_frequencies): Use it.
20756 (compute_function_frequency): Use zero counts even if profile is
20758 * profile-count.h: Fix typo.
20760 2017-08-08 Julia Koval <julia.koval@intel.com>
20762 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20763 _mm512_mask_cvtsepi16_storeu_epi8,
20764 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20765 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20766 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20767 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20768 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20769 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20770 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20771 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20772 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20773 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20774 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20775 __builtin_ia32_pmovuswb256mem_mask,
20776 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20777 __builtin_ia32_pmovwb512mem_mask): New builtins.
20779 2017-08-08 Julia Koval <julia.koval@intel.com>
20781 PR target/73350,80862
20782 * config/i386/subst.md (round): Fix round pattern.
20783 * config/i386/i386.c (ix86_erase_embedded_rounding):
20784 Fix erasing rounding for the fixed pattern.
20786 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20788 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20790 2017-06-08 Martin Liska <mliska@suse.cz>
20792 PR gcov-profile/80911
20793 * gcov.c (block_info::block_info): New constructor.
20795 2017-06-07 Carl Love <cel@us.ibm.com>
20797 * config/rs6000/rs6000-c: The return type of the following
20798 built-in functions was implemented as int not long long. Fix sign
20799 of return value for the unsigned version of vec_mulo and vec_mule.
20800 vector unsigned long long vec_bperm (vector unsigned long long,
20801 vector unsigned char)
20802 vector signed long long vec_mule (vector signed int,
20804 vector unsigned long long vec_mule (vector unsigned int,
20805 vector unsigned int)
20806 vector signed long long vec_mulo (vector signed int,
20808 vector unsigned long long vec_mulo (vector unsigned int,
20809 vector unsigned int)
20810 * doc/extend.texi: Fix the documentation for the built-in
20813 2017-06-07 Carl Love <cel@us.ibm.com>
20816 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20819 2017-06-07 Carl Love <cel@us.ibm.com>
20821 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20822 support, Generate doublehv for signed int/float for BE case only.
20824 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20826 * doc/invoke.texi (mcx16): Rewrite.
20828 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20830 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20831 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20832 *mov<mode>_softfloat, and an anonymous splitter): Use
20833 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20835 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20837 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20839 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20840 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20841 (rs6000_debug_reg_global): Adjust.
20842 (rs6000_init_hard_regno_mode_ok): Adjust.
20843 (rs6000_dbx_register_number): Adjust.
20844 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20845 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20846 Remove SPE_ACC and SPEFSCR.
20847 (REG_ALLOC_ORDER): Ditto.
20848 (FRAME_POINTER_REGNUM): Change to 111.
20849 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20850 (REG_CLASS_NAMES): Ditto.
20851 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20852 (REGISTER_NAMES): Ditto.
20853 (ADDITIONAL_REG_NAMES): Ditto.
20854 (rs6000_reg_names): Ditto.
20855 * config/rs6000/rs6000.md: Renumber some register number
20858 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20860 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20862 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20863 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20865 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20866 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20867 Delete the SPE high registers.
20868 (REG_ALLOC_ORDER): Ditto.
20869 (enum reg_class): Remove SPE_HIGH_REGS.
20870 (REG_CLASS_NAMES): Ditto.
20871 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20872 (REGISTER_NAMES): Ditto.
20873 (rs6000_reg_names): Ditto.
20874 * doc/tm.texi.in: Remove SPE as example.
20875 * doc/tm.texi: Regenerate.
20877 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20879 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20880 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20881 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20882 * config/rs6000/rs6000.md (type): Remove "brinc".
20884 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20886 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20887 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20888 * config/rs6000/linuxspe.h: Delete file.
20889 * config/rs6000/rs6000.md: Don't include spe.md.
20890 * config/rs6000/spe.h: Delete file.
20891 * config/rs6000/spe.md: Delete file.
20892 * config/rs6000/t-rs6000: Remove spe.md.
20894 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20896 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20897 (reg_or_none500mem_operand): Delete.
20898 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20899 instead of reg_or_none500mem_operand.
20901 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20903 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20904 handling of SPE flags.
20905 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20907 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20909 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20911 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20912 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20913 paired_divv2sf3): Similar.
20914 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20915 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20916 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20918 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20919 Rename the paired_* instruction patterns.
20920 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20922 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20923 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20924 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20925 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20926 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20927 PAIRED_VECTOR_MODE.
20928 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20929 (spe_func_has_64bit_regs_p): Delete.
20930 (spe_expand_predicate_builtin): Delete.
20931 (spe_expand_evsel_builtin): Delete.
20932 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20933 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20934 (invalid_e500_subreg): Delete.
20935 (rs6000_legitimize_address): Always force_reg op2 as well, for
20936 paired single memory accesses.
20937 (rs6000_member_type_forces_blk): Delete.
20938 (rs6000_spe_function_arg): Delete.
20939 (rs6000_expand_unop_builtin): Delete SPE handling.
20940 (rs6000_expand_binop_builtin): Ditto.
20941 (spe_expand_stv_builtin): Delete.
20942 (bdesc_2arg_spe): Delete.
20943 (spe_expand_builtin): Delete.
20944 (spe_expand_predicate_builtin): Delete.
20945 (spe_expand_evsel_builtin): Delete.
20946 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20947 (spe_init_builtins): Delete.
20948 (spe_func_has_64bit_regs_p): Delete.
20949 (savres_routine_name): Delete "info" parameter. Adjust callers.
20950 (rs6000_emit_stack_reset): Ditto.
20951 (rs6000_dwarf_register_span): Delete.
20952 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20953 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20954 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20956 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20958 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20959 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20960 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20961 mulv2sf3, divv2sf3): Delete expanders.
20963 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20965 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20967 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20969 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20970 * config/rs6000/rs6000.c: Ditto.
20972 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20974 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20975 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20976 comparison_operator.
20978 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20980 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20981 * config/rs6000/rs6000.opt: Ditto.
20982 * config/rs6000/t-rtems: Ditto.
20984 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20986 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20987 TARGET_E500_SINGLE by 0, simplify.
20988 * config/rs6000/rs6000.c: Ditto.
20989 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20990 (spe_build_register_parallel): Delete.
20991 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20992 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20993 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20994 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20995 (E500_CONVERT): Delete.
20996 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20998 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21000 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21001 * config/rs6000/dfp.md: Ditto.
21002 (negdd2, *negdd2_fpr): Merge.
21003 (absdd2, *absdd2_fpr): Merge.
21004 (negtd2, *negtd2_fpr): Merge.
21005 (abstd2, *abstd2_fpr): Merge.
21006 * config/rs6000/e500.h: Delete file.
21007 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21008 TARGET_FPRS by 1 and simplify.
21009 * config/rs6000/rs6000-c.c: Ditto.
21010 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21011 TARGET_DF_SPE by 0.
21012 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21014 * config/rs6000/rs6000.md: Ditto.
21015 (floatdidf2, *floatdidf2_fpr): Merge.
21016 (move_from_CR_gt_bit): Delete.
21017 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21018 (E500_CR_IOR_COMPARE): Delete.
21019 (All patterns that require !TARGET_FPRS): Delete.
21020 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21022 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21024 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21026 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21028 * graphds.c (add_edge): Intitialize edge's attached data.
21029 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21030 pointer parameter. Call pointed function on each edge during
21031 graph traversing. Skip traversing the edge when the function
21033 (graphds_dfs, graphds_scc): Ditto.
21034 (for_each_edge): New parameter. Pass the new parameter to callback
21036 * graphds.h (skip_edge_callback): New function pointer type.
21037 (graphds_dfs, graphds_scc): New function pointer parameter.
21038 (graphds_edge_callback, for_each_edge): New parameter.
21040 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21042 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21043 out code checking if runtime alias check is possible to below ...
21044 Call the new function.
21045 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21046 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21048 2017-06-07 Marek Polacek <polacek@redhat.com>
21051 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21052 TYPE_OVERFLOW_WRAPS check.
21054 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21056 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21057 if versioning is required.
21058 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21059 peeling with the check for versioning.
21061 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21063 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21064 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21065 Set true to new parameter if new ssa variable is defined.
21066 (vect_gen_vector_loop_niters): Refactor. Set range information
21067 for the new vector loop bound variable.
21068 (vect_do_peeling): Ditto.
21070 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21072 * tree-affine.c (ssa.h): Include header file.
21073 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21074 has wrapping overflow behavior.
21076 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21078 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21080 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21082 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21083 (tree_to_aff_combination): ... here.
21085 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21087 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21088 reg_pressure model function.
21089 (ivopts_global_cost_for_size): Delete.
21090 (determine_set_costs, iv_ca_recount_cost): Call new model function
21091 ivopts_estimate_reg_pressure.
21093 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21095 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21096 expensive than udiv. Remove floating point cases from mod.
21098 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21100 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21101 Increase idiv cost.
21103 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21105 * config/aarch64/aarch64.md
21106 (copysignsf3): Fix mask generation.
21108 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21110 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21112 (class dump_manager): Add register_dumps method.
21113 * dumpfile.c: Include langhooks.h.
21114 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21115 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21116 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21117 (dump_manager::dump_register): Start with 512 entries instead of 32.
21118 (dump_manager::register_dumps): New method.
21119 * toplev.c (general_init): Instead of invoking register_dumps
21120 langhook, invoke register_dumps method on the dump manager.
21121 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21124 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21126 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21127 Say that # requires an associated define_split to exist, and that
21128 the define_split must be suitable for use after register allocation.
21130 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21132 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21133 (compute_outgoing_frequencies): Also initialize zero counts.
21134 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21135 around loops; preserve more of profile when nothing changes.
21137 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21139 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21141 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21142 * config/arm/arm-cpu-cdata.h: Regenerate.
21143 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21144 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21145 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21146 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21147 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21149 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21151 * config/arm/t-rmprofile: Likewise.
21152 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21154 2017-06-06 David S. Miller <davem@davemloft.net>
21157 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21158 blockage if function uses alloca.
21160 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21162 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21163 New "uid" fields to hold pretty-print IDs of group and ref.
21164 Memory references are now identified as <group_id>:<ref_id>
21165 instead of using [random] addresses.
21166 (dump_mem_details): Simplify, no functional change.
21167 (dump_mem_ref): Simplify and make output more concise.
21168 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21169 (find_or_create_group): Initialize group uid.
21170 (record_ref): Initialize ref uid. Improve debug output.
21171 (prune_group_by_reuse, should_issue_prefetch_p,)
21172 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21173 (mark_nontemporal_store, determine_loop_nest_reuse):
21174 Improve debug output.
21176 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21178 * dbgcnt.def (prefetch): New debug counter.
21179 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21180 (schedule_prefetches): Stop issueing prefetches if debug counter
21183 2017-06-06 Tom de Vries <tom@codesourcery.com>
21185 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21186 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21188 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21190 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21191 Use aarch64_reg_or_zero predicate for operand 4.
21192 (aarch64_compare_and_swap<mode> define_insn_and_split):
21193 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21194 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21196 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21198 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21199 (arm_compute_save_core_reg_mask): This.
21200 (thumb1_compute_save_reg_mask): Rename into ...
21201 (thumb1_compute_save_core_reg_mask): This.
21202 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21203 (arm_compute_frame_layout): Likewise.
21205 2017-06-06 Richard Biener <rguenther@suse.de>
21207 PR tree-optimization/80974
21208 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21209 keep or clear leaders SSA info.
21211 2017-06-06 Tom de Vries <tom@codesourcery.com>
21213 * config/nvptx/nvptx.c (split_mode_p): New function.
21214 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21216 2017-06-06 Tom de Vries <tom@codesourcery.com>
21218 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21220 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21223 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21226 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21228 * shrink-wrap.c (handle_simple_exit): Update profile.
21229 (try_shrink_wrapping): Upate profile.
21231 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21233 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21234 (tree_guess_outgoing_edge_probabilities): New.
21235 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21236 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21238 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21240 * ipa-split.c (split_function): Initialize return bb profile.
21242 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21244 * profile.c (compute_branch_probabilities): Also initialize
21245 EXIT_BLOCK profile.
21247 2017-06-06 Richard Biener <rguenther@suse.de>
21249 PR tree-optimization/80928
21250 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21251 (vect_analyze_loop_operations): Properly guard analysis for
21253 (vect_transform_loop): Likewise.
21254 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21255 (vect_model_induction_cost): Do not cost for pure SLP.
21256 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21257 of induction in inner loop vectorization.
21258 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21259 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21260 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21262 (vect_analyze_slp_cost_1): Cost induction.
21263 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21264 (vect_get_slp_vect_defs): Likewise.
21265 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21266 (vect_transform_stmt): Handle SLP reductions.
21267 * tree-vectorizer.h (vectorizable_induction): Adjust.
21269 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21271 * config/rs6000/rs6000.c (make_resolver_func): Update
21272 init_lowered_empty_function call.
21274 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21276 * doc/invoke.texi: Document the -fprofile-abs-path option.
21277 * common.opt (fprofile-abs-path): New option.
21278 * gcov-io.h (gcov_write_filename): Declare.
21279 * gcov-io.c (gcov_write_filename): New function.
21280 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21281 * profile.c (output_location): Likewise.
21283 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21285 * shring-wrap.c: Revert accidental commit.
21287 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21289 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21291 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21293 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21295 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21296 profile in callgraph edge.
21297 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21298 (apply_scale): Likewise.
21299 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21300 Also scale profile when inlining function with zero profile.
21301 (initialize_cfun): Update exit block profile even when it is zero.
21302 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21303 when profile is read.
21305 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21307 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21308 (CLONE_*): New constants to define the processors we can generate
21309 code for with the target_clone attribute.
21310 (rs6000_clone_map): New array to identify which clone processors
21311 the current program is running on.
21312 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21313 target_clone attribute.
21314 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21315 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21316 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21317 (cpu_expand_builtin): Add support for target_clone attribute.
21318 (rs6000_valid_attribute_p): Allow "default" attribute.
21319 (get_decl_name): New debug function to simplify printing the
21320 current function name in debugging statements.
21321 (rs6000_clone_priority): New functions to support the target_clone
21322 attribute, and be able to generate code to switch between ISA 2.05
21323 through ISA 3.0 (power6 through power9).
21324 (rs6000_compare_version_priority): Likewise.
21325 (rs6000_get_function_versions_dispatcher): Likewise.
21326 (make_resolver_func): Likewise.
21327 (add_condition_to_bb): Likewise.
21328 (dispatch_function_versions): Likewise.
21329 (rs6000_generate_version_dispatcher_body): Likewise.
21330 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21331 (fusion_gpr_load_p): Fix a spacing issue.
21332 * doc/extend.texi (Common Function Attributes): Document that the
21333 PowerPC supports the target_clone attribute.
21335 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21337 * config/arm/arm.h: explain F symbol found in description of ARM
21338 register allocation in its legend.
21340 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21342 * config/mips/frame-header-opt.c: Include profile-count.h.
21343 * config/riscv/riscv.c: Include profile-count.h
21345 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21347 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21349 (sm_set_flag_if_changed): Add bbs field.
21350 (execute_sm_if_changed_flag_set): Pass BBS.
21351 (execute_sm): Update.
21353 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21355 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21358 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21360 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21361 (peephole2): New peephole2 to emit the above.
21362 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21364 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21366 * config/aarch64/aarch64.c (define_peephole2 above
21367 *sub_<shift>_<mode>): New peephole.
21369 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21371 * config/i386/i386.c (make_resolver_func): Update.
21372 * Makefile.in: Add profile-count.h and profile-count.o
21373 * auto-profile.c (afdo_indirect_call): Update to new API.
21374 (afdo_set_bb_count): Update.
21375 (afdo_propagate_edge): Update.
21376 (afdo_propagate_circuit): Update.
21377 (afdo_calculate_branch_prob): Update.
21378 (afdo_annotate_cfg): Update.
21379 * basic-block.h: Include profile-count.h
21380 (struct edge_def): Turn count to profile_count.
21381 (struct basic_block_def): Likewie.
21382 (REG_BR_PROB_BASE): Move to profile-count.h
21383 (RDIV): Move to profile-count.h
21384 * bb-reorder.c (max_entry_count): Turn to profile_count.
21385 (find_traces): Update.
21386 (rotate_loop):Update.
21387 (connect_traces):Update.
21388 (sanitize_hot_paths):Update.
21389 * bt-load.c (migrate_btr_defs): Update.
21390 * cfg.c (RDIV): Remove.
21391 (init_flow): Use alloc_block.
21392 (alloc_block): Uninitialize count.
21393 (unchecked_make_edge): Uninitialize count.
21394 (check_bb_profile): Update.
21395 (dump_edge_info): Update.
21396 (dump_bb_info): Update.
21397 (update_bb_profile_for_threading): Update.
21398 (scale_bbs_frequencies_int): Update.
21399 (scale_bbs_frequencies_gcov_type): Update.
21400 (scale_bbs_frequencies_profile_count): New.
21401 * cfg.h (update_bb_profile_for_threading): Update.
21402 (scale_bbs_frequencies_profile_count): Declare.
21403 * cfgbuild.c (compute_outgoing_frequencies): Update.
21404 (find_many_sub_basic_blocks): Update.
21405 * cfgcleanup.c (try_forward_edges): Update.
21406 (try_crossjump_to_edge): Update.
21407 * cfgexpand.c (expand_gimple_tailcall): Update.
21408 (construct_exit_block): Update.
21409 * cfghooks.c (verify_flow_info): Update.
21410 (dump_bb_for_graph): Update.
21411 (split_edge): Update.
21412 (make_forwarder_block): Update.
21413 (duplicate_block): Update.
21414 (account_profile_record): Update.
21415 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21416 (get_estimated_loop_iterations): Update.
21417 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21418 (single_likely_exit): Update.
21419 * cfgloopmanip.c (scale_loop_profile): Update.
21421 (set_zero_probability): Update.
21422 (lv_adjust_loop_entry_edge): Update.
21423 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21424 (purge_dead_edges): Update.
21425 (rtl_account_profile_record): Update.
21426 * cgraph.c (cgraph_node::create): Uninitialize count.
21427 (symbol_table::create_edge): Uninitialize count.
21428 (cgraph_update_edges_for_call_stmt_node): Update.
21429 (cgraph_edge::dump_edge_flags): Update.
21430 (cgraph_node::dump): Update.
21431 (cgraph_edge::maybe_hot_p): Update.
21432 * cgraph.h: Include profile-count.h
21433 (create_clone), create_edge, create_indirect_edge): Update.
21434 (cgraph_node): Turn count to profile_count.
21435 (cgraph_edge0: Likewise.
21436 (make_speculative, clone): Update.
21437 (create_edge): Update.
21438 (init_lowered_empty_function): Update.
21439 * cgraphclones.c (cgraph_edge::clone): Update.
21440 (duplicate_thunk_for_node): Update.
21441 (cgraph_node::create_clone): Update.
21442 * cgraphunit.c (cgraph_node::analyze): Update.
21443 (cgraph_node::expand_thunk): Update.
21444 * final.c (dump_basic_block_info): Update.
21445 * gimple-streamer-in.c (input_bb): Update.
21446 * gimple-streamer-out.c (output_bb): Update.
21447 * graphite.c (print_global_statistics): Update.
21448 (print_graphite_scop_statistics): Update.
21449 * hsa-brig.c: Include basic-block.h.
21450 * hsa-dump.c: Include basic-block.h.
21451 * hsa-gen.c (T sum_slice): Update.
21452 (convert_switch_statements):Update.
21453 * hsa-regalloc.c: Include basic-block.h.
21454 * ipa-chkp.c (chkp_produce_thunks): Update.
21455 * ipa-cp.c (struct caller_statistics): Update.
21456 (init_caller_stats): Update.
21457 (gather_caller_stats): Update.
21458 (ipcp_cloning_candidate_p): Update.
21459 (good_cloning_opportunity_p): Update.
21460 (get_info_about_necessary_edges): Update.
21461 (dump_profile_updates): Update.
21462 (update_profiling_info): Update.
21463 (update_specialized_profile): Update.
21464 (perhaps_add_new_callers): Update.
21465 (decide_about_value): Update.
21466 (ipa_cp_c_finalize): Update.
21467 * ipa-devirt.c (struct odr_type_warn_count): Update.
21468 (struct decl_warn_count): Update.
21469 (struct final_warning_record): Update.
21470 (possible_polymorphic_call_targets): Update.
21471 (ipa_devirt): Update.
21472 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21473 * ipa-icf.c (sem_function::merge): Update.
21474 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21475 * ipa-inline.c (compute_uninlined_call_time): Update.
21476 (compute_inlined_call_time): Update.
21477 (want_inline_small_function_p): Update.
21478 (want_inline_self_recursive_call_p): Update.
21479 (edge_badness): Update.
21480 (lookup_recursive_calls): Update.
21481 (recursive_inlining): Update.
21482 (inline_small_functions): Update.
21483 (dump_overall_stats): Update.
21484 (dump_inline_stats): Update.
21485 * ipa-profile.c (ipa_profile_generate_summary): Update.
21486 (ipa_propagate_frequency): Update.
21487 (ipa_profile): Update.
21488 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21489 * ipa-utils.c (ipa_merge_profiles): Update.
21490 * loop-doloop.c (doloop_modify): Update.
21491 * loop-unroll.c (report_unroll): Update.
21492 (unroll_loop_runtime_iterations): Update.
21493 * lto-cgraph.c (lto_output_edge): Update.
21494 (lto_output_node): Update.
21495 (input_node): Update.
21496 (input_edge): Update.
21497 (merge_profile_summaries): Update.
21498 * lto-streamer-in.c (input_cfg): Update.
21499 * lto-streamer-out.c (output_cfg): Update.
21500 * mcf.c (create_fixup_graph): Update.
21501 (adjust_cfg_counts): Update.
21502 (sum_edge_counts): Update.
21503 * modulo-sched.c (sms_schedule): Update.
21504 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21505 * predict.c (maybe_hot_count_p): Update.
21506 (probably_never_executed): Update.
21507 (dump_prediction): Update.
21508 (combine_predictions_for_bb): Update.
21509 (propagate_freq): Update.
21510 (handle_missing_profiles): Update.
21511 (counts_to_freqs): Update.
21512 (rebuild_frequencies): Update.
21513 (force_edge_cold): Update.
21514 * predict.h: Include profile-count.h
21515 (maybe_hot_count_p, counts_to_freqs): UPdate.
21516 * print-rtl-function.c: Do not include cfg.h
21517 * print-rtl.c: Include basic-block.h
21518 * profile-count.c: New file.
21519 * profile-count.h: New file.
21520 * profile.c (is_edge_inconsistent): Update.
21521 (correct_negative_edge_counts): Update.
21522 (is_inconsistent): Update.
21523 (set_bb_counts): Update.
21524 (read_profile_edge_counts): Update.
21525 (compute_frequency_overlap): Update.
21526 (compute_branch_probabilities): Update; Initialize and deinitialize
21528 (branch_prob): Update.
21529 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21530 (edge_gcov_count): New.
21531 (bb_gcov_count): New.
21532 * shrink-wrap.c (try_shrink_wrapping): Update.
21533 * tracer.c (better_p): Update.
21534 * trans-mem.c (expand_transaction): Update.
21535 (ipa_tm_insert_irr_call): Update.
21536 (ipa_tm_insert_gettmclone_call): Update.
21537 * tree-call-cdce.c: Update.
21538 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21539 (gimple_duplicate_sese_tail): Update.
21540 (gimple_account_profile_record): Update.
21541 (execute_fixup_cfg): Update.
21542 * tree-inline.c (copy_bb): Update.
21543 (copy_edges_for_bb): Update.
21544 (initialize_cfun): Update.
21545 (freqs_to_counts): Update.
21546 (copy_cfg_body): Update.
21547 (expand_call_inline): Update.
21548 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21549 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21550 (try_unroll_loop_completely): Update.
21551 (try_peel_loop): Update.
21552 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21553 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21554 * tree-ssa-loop-split.c (connect_loops): Update.
21555 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21556 * tree-ssa-reassoc.c (branch_fixup): Update.
21557 * tree-ssa-tail-merge.c (replace_block_by): Update.
21558 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21559 (compute_path_counts): Update.
21560 (update_profile): Update.
21561 (recompute_probabilities): Update.
21562 (update_joiner_offpath_counts): Update.
21563 (estimated_freqs_path): Update.
21564 (freqs_to_counts_path): Update.
21565 (clear_counts_path): Update.
21566 (ssa_fix_duplicate_block_edges): Update.
21567 (duplicate_thread_path): Update.
21568 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21569 (struct switch_conv_info): Update.
21570 * tree-tailcall.c (decrease_profile): Update.
21571 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21572 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21573 * value-prof.c (check_counter): Update.
21574 (gimple_divmod_fixed_value): Update.
21575 (gimple_mod_pow2): Update.
21576 (gimple_mod_subtract): Update.
21577 (gimple_ic_transform): Update.
21578 (gimple_stringop_fixed_value): Update.
21579 * value-prof.h (gimple_ic): Update.
21581 2017-06-02 Carl Love <cel@us.ibm.com>
21583 * config/rs6000/rs6000-c: Add support for built-in functions
21584 vector double vec_doublee (vector signed int);
21585 vector double vec_doublee (vector unsigned int);
21586 vector double vec_doublee (vector float);
21587 vector double vec_doubleh (vector signed int);
21588 vector double vec_doubleh (vector unsigned int);
21589 vector double vec_doubleh (vector float);
21590 vector double vec_doublel (vector signed int);
21591 vector double vec_doublel (vector unsigned int);
21592 vector double vec_doublel (vector float);
21593 vector double vec_doubleo (vector signed int);
21594 vector double vec_doubleo (vector unsigned int);
21595 vector double vec_doubleo (vector float);.
21596 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21597 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21599 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21600 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21601 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21603 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21604 vec_doublel, vec_doubleh.
21605 * doc/extend.texi: Update the built-in documentation file for the
21606 new built-in functions.
21608 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21611 * ipa-inline-analysis.c (free_growth_caches): Set
21612 edge_removal_hook_holder to NULL after removing it.
21614 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21616 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21617 comparision with zero.
21619 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21620 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21621 for early expansion of vec_min and vec_max builtins.
21622 (builtin_function_type): Add min/max unsigned variants to those
21623 identified as having unsigned arguments.
21625 2017-06-02 Olivier Hainque <hainque@adacore.com>
21627 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21629 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21631 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21632 Use VALL_F16 iterator rather than VALL.
21634 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21636 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21637 Emit CBNZ inside loop when doing a strong exchange and comparing
21638 against zero. Generate the CC flags after the loop.
21640 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21642 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21643 (dl_section_ref): New.
21644 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21645 On AIX, append an expression to subtract the size of the
21646 section length to dl_section_ref.
21648 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21650 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21651 for early expansion of vector absolute builtins.
21653 2017-06-02 Richard Biener <rguenther@suse.de>
21655 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21656 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21658 2017-06-02 Richard Biener <rguenther@suse.de>
21660 PR tree-optimization/80948
21661 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21662 stmt order as well.
21664 2017-06-02 Richard Biener <rguenther@suse.de>
21666 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21668 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21669 for inductions as relevant.
21671 2017-06-02 Richard Biener <rguenther@suse.de>
21673 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21674 (vectorizable_induction): ... this. Remove dead code.
21676 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21678 * builtins. (expand_builtin_alloca): Remove second parameter and
21679 infer its value from the first parameter instead.
21680 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21682 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21684 PR rtl-optimization/80903
21685 * loop-doloop.c (add_test): Unshare sequence.
21687 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21689 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21691 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21693 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21695 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21696 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21697 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21698 (xlogue_layout::get_stub_rtx): Make static.
21699 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21700 (xlogue_layout::compute_stub_managed_regs): Rename to...
21701 (xlogue_layout::count_stub_managed_regs): ...this.
21702 (xlogue_layout::is_stub_managed_reg): New function.
21703 (xlogue_layout::m_stub_names): Rename to...
21704 (xlogue_layout::s_stub_names): ...this, make static.
21705 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21706 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21707 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21708 xlogue_layout::s_stub_names): Instantiate statics.
21709 (stub_managed_regs): Remove.
21710 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21711 (disable_call_ms2sysv_xlogues): Rename to...
21712 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21713 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21715 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21716 change after reload_completed.
21717 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21719 (ix86_expand_prologue): Likewise.
21720 (ix86_expand_epilogue): Likewise.
21721 (ix86_expand_split_stack_prologue): Likewise.
21722 (ix86_compute_frame_layout): Remove frame parameter ...
21723 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21724 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21726 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21727 (ix86_frame): Move from here ...
21728 * config/i386/i386.h (ix86_frame): ... to here.
21729 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21730 complete ix86_frame data structure instead. Remove some_ld_name.
21732 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21734 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21735 symbols that hold a DECL_VALUE_EXPR.
21737 2017-06-01 Martin Jambor <mjambor@suse.cz>
21739 PR tree-optimization/80898
21740 * tree-sra.c (process_subtree_disqualification): Removed.
21741 (disqualify_candidate): Do not acll
21742 process_subtree_disqualification.
21743 (subtree_mark_written_and_enqueue): New function.
21744 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21745 RHS has been disqualified and re-queue LHS if necessary. Apart
21746 from that, ignore disqualified RHS.
21748 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21750 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21751 address fetch for z10 or later.
21753 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21755 * config/arc/arc.md (tst_movb): Add guard when splitting.
21757 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21759 * config/arc/arc.c (arc_can_eliminate): Test against
21760 arc_frame_pointer_needed.
21762 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21764 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21765 to prevent store reordering.
21766 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21767 (type): Add block type.
21768 (stack_tie): Define special instruction to be used in
21771 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21773 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21774 constraint. It is not valid for the pattern.
21775 (noncommutative_binary_comparison): Likewise.
21777 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21779 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21782 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21784 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21785 be used by the reg-alloc.
21787 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21789 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21790 reg-alloc when having mul64 or mul32x16 instructions.
21791 (mulsidi3): Likewise.
21792 (umulsidi3): Likewise.
21793 (mulsi32x16): New pattern.
21794 (mulsi64): Likewise.
21795 (mulsidi64): Likewise.
21796 (umulsidi64): Likewise.
21797 (MUL32x16_REG): Define.
21798 (mul64_600): Use MUL32x16_REG.
21799 (mac64_600): Likewise.
21800 (umul64_600): Likewise.
21801 (umac64_600): Likewise.
21803 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21805 * config/arc/arc.md (mulsi3_700): Make it commutative.
21807 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21809 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21811 (*sign_extendsidi2_insn): Likewise for movstosw.
21813 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21815 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21816 the type of the input discriminant value. Convert the
21817 discriminant value of signedness vary.
21819 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21821 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21822 Add to -Wall section.
21824 2017-06-01 Richard Biener <rguenther@suse.de>
21826 PR middle-end/66313
21827 * fold-const.c (fold_plusminus_mult_expr): If the factored
21828 factor may be zero use a wrapping type for the inner operation.
21829 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21830 and handle moved defs.
21831 (process_assignment): Properly guard the unary op case. Return a
21832 tri-state indicating that moving the stmt before the call may allow
21833 to continue. Pass through to_move.
21834 (find_tail_calls): Handle moving unrelated defs before
21837 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21840 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21841 splitter result in the canonical way.
21843 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21845 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21846 also for 32bit target. Update insn attributes.
21847 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21849 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21851 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21852 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21853 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21854 (_mm_maskz_min_ss): New intrinsics.
21856 2017-05-31 Martin Liska <mliska@suse.cz>
21858 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21859 Change comment style to one we normally use.
21860 (vectorizable_reduction): Likewise.
21861 (vectorizable_induction): Likewise.
21862 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21863 (vectorizable_call): Likewise.
21864 (vectorizable_simd_clone_call): Likewise.
21865 (vectorizable_conversion): Likewise.
21866 (vectorizable_assignment): Likewise.
21867 (vectorizable_shift): Likewise.
21868 (vectorizable_operation): Likewise.
21869 (vectorizable_store): Likewise.
21870 (vectorizable_load): Likewise.
21871 * tree-vectorizer.h: Likewise.
21873 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21875 * passes.c (emergency_dump_function): New.
21876 * tree-pass.h (emergency_dump_function): Declare.
21877 * plugin.c (plugins_internal_error_function): Remove.
21878 * plugin.h (plugins_internal_error_function): Remove declaration.
21879 * toplev.c (internal_error_function): New static function. Use it...
21880 (general_init): ...here.
21882 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21884 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21885 (arc_rtx_costs): Add costs for new patterns.
21886 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21887 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21889 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21891 * tree-ssa-strlen.c (get_next_strinfo): New function.
21892 (get_stridx_plus_constant): Use it.
21893 (zero_length_string): Likewise.
21894 (adjust_related_strinfos): Likewise.
21895 (adjust_last_stmt): Likewise.
21897 2017-05-31 Richard Biener <rguenther@suse.de>
21900 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21901 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21903 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21905 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21906 loop_vinfo argument and use of dependence distance vectors.
21907 Check instead whether the two references differ only in their
21908 initial value and assume that they have the same alignment if the
21909 difference is a multiple of the vector alignment.
21910 (vect_analyze_data_refs_alignment): Update call accordingly.
21912 2017-05-31 Martin Liska <mliska@suse.cz>
21915 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21917 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21919 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21920 (create_intersect_range_checks): Move from ...
21921 * tree-data-ref.c (create_intersect_range_checks_index)
21922 (create_intersect_range_checks): ... to here.
21923 (create_runtime_alias_checks): New function factored from ...
21924 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21925 here. Call above function.
21926 * tree-data-ref.h (create_runtime_alias_checks): New function.
21928 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21930 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21931 segment length for dr_b and compute it in wide_int.
21933 2017-05-31 Richard Biener <rguenther@suse.de>
21935 PR tree-optimization/80906
21936 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21937 and pass through iv_map.
21938 (copy_bb_and_scalar_dependences): Adjust.
21939 (translate_pending_phi_nodes): Likewise.
21940 (copy_loop_close_phi_args): Handle code-generating IVs instead
21943 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21945 * diagnostic-color.c (color_dict): Add "type-diff".
21946 (parse_gcc_colors): Update comment.
21947 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21948 -fdiagnostics-show-template-tree and -fno-elide-type.
21949 (GCC_COLORS): Add type-diff to example.
21951 (-fdiagnostics-show-template-tree): New.
21952 (-fno-elide-type): New.
21953 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21954 the pp_format_decoder callback. Call any m_format_postprocessor's
21956 (pretty_printer::pretty_printer): Initialize
21957 m_format_postprocessor.
21958 (pretty_printer::~pretty_printer): Delete any
21959 m_format_postprocessor.
21960 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21961 (class format_postprocessor): New class.
21962 (struct pretty_printer::format_decoder): Document the new parameters.
21963 (struct pretty_printer::m_format_postprocessor): New field.
21964 * tree-diagnostic.c (default_tree_printer): Update for new
21965 bool and const char ** params.
21966 * tree-diagnostic.h (default_tree_printer): Likewise.
21968 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21970 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21971 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21972 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21973 rs6000_gen_cell_microcode code.
21974 (rs6000_final_prescan_insn): Delete.
21975 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21976 "warn-cell-microcode" entries.
21977 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21978 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21979 throughout. Change cc_reg_not_micro_cr0_operand to
21980 cc_reg_not_cr0_operand throughout.
21981 (*extendhi<mode>2_noload): Delete.
21982 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21983 (mwarn-cell-microcode): Delete.
21984 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21985 -mgen-cell-microcode and -mwarn-cell-microcode.
21987 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21990 * config/i386/constraints.md (Yd): New constraint.
21992 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21993 and (?Yd, r) alternatives. Update insn attributes.
21994 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21995 and (?*Yd, r) alternatives. Update insn attributes.
21996 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21998 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22000 * gimplify.c (gimplify_modify_expr): Don't create a
22001 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22004 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22006 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22008 2017-05-30 Richard Biener <rguenther@suse.de>
22010 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22011 and reduc_def fields.
22012 (STMT_VINFO_REDUC_TYPE): New define.
22013 (STMT_VINFO_REDUC_DEF): Likewise.
22014 (vect_force_simple_reduction): Adjust prototype.
22015 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22016 (vect_is_simple_reduction): Remove check_reduction argument.
22017 (vect_force_simple_reduction): Adjust and set
22018 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22019 (vectorizable_reduction): Do not re-do reduction analysis
22020 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22021 * tree-parloops.c (gather_scalar_reductions): Adjust.
22023 2017-05-30 Richard Biener <rguenther@suse.de>
22025 PR middle-end/80901
22026 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22029 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22031 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22032 Introduce unknown_misalignment parameter and remove vf.
22033 (vect_peeling_hash_get_lowest_cost):
22034 Pass unknown_misalignment parameter.
22035 (vect_enhance_data_refs_alignment):
22036 Fix unsupportable data ref treatment.
22038 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22040 * tree-vect-data-refs.c (vect_get_data_access_cost):
22041 Workaround for SLP handling.
22042 (vect_enhance_data_refs_alignment):
22043 Compute costs for doing no peeling at all, compare to the best
22044 peeling costs so far and avoid peeling if cheaper.
22046 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22048 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22049 Return peeling info and set costs to zero for unlimited cost
22051 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22052 with unknown misalignment. Compute and costs for unknown
22053 misalignment, compare them to the costs for known misalignment
22054 and choose the cheapest for peeling.
22056 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22058 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22059 (vect_get_peeling_costs_all_drs): Create function.
22060 (vect_peeling_hash_get_lowest_cost):
22061 Use vect_get_peeling_costs_all_drs.
22062 (vect_peeling_supportable): Create function.
22063 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22065 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22067 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22068 DR_HAS_NEGATIVE_STEP.
22069 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22070 (vect_enhance_data_refs_alignment): Use.
22071 (vect_duplicate_ssa_name_ptr_info): Use.
22072 * tree-vectorizer.h (dr_misalignment): Use.
22073 (known_alignment_for_access_p): Use.
22075 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22078 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22080 (has_section_name): New function.
22082 2017-05-30 Martin Liska <mliska@suse.cz>
22085 * auto-profile.c (get_function_decl_from_block): Fix
22088 2017-05-30 Richard Biener <rguenther@suse.de>
22090 PR middle-end/80876
22091 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22093 2017-05-30 Martin Liska <mliska@suse.cz>
22095 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22096 * dumpfile.h (struct dump_file_info): Remove ctors.
22098 2017-05-30 Martin Liska <mliska@suse.cz>
22100 * predict.def: Fix GNU coding style.
22102 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22104 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22105 Mark 'to' argument with ATTRIBUTE_UNUSED.
22107 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22109 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22110 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22111 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22114 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22116 * doc/install.texi (Options specification): Restore entry of
22117 --enable-sjlj-exceptions.
22119 2017-05-27 Michael Eager <eager@eagercon.com>
22122 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22124 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22126 * config/microblaze/microblaze.h
22127 (FIXED_REGISTERS): Update in macro.
22128 (CALL_USED_REGISTERS): Update in macro.
22130 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22132 * doc/install.texi: Add links to macOS binary distributions.
22134 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22138 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22140 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22142 2017-05-26 Martin Liska <mliska@suse.cz>
22144 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22146 2017-05-26 Martin Liska <mliska@suse.cz>
22148 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22149 always leading ';; '.
22150 (dump_bb_info): Likewise.
22151 (brief_dump_cfg): Likewise.
22152 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22153 * dumpfile.c: Remove usage of TDF_VERBOSE.
22154 * dumpfile.h (enum dump_kind): Likewise.
22155 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22156 * print-tree.c (debug_verbose): Remove.
22157 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22158 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22159 * tree-diagnostic.c (default_tree_printer): Replace
22160 TDF_DIAGNOSTIC with TDF_SLIM.
22162 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22164 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22165 in parameter loop, rather than loop_vinfo.
22166 (create_intersect_range_checks): Ditto.
22167 (vect_create_cond_for_alias_checks): Update call to above functions.
22169 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22171 PR tree-optimization/80815
22172 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22173 for merging runtime alias checks. Handle negative DR_STEPs.
22175 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22177 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22179 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22180 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22181 out code pruning runtime alias checks.
22182 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22183 factored out from above.
22184 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22186 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22188 (prune_runtime_alias_test_list): New decalaration.
22190 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22192 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22193 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22194 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22195 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22196 (operator==, comp_dr_with_seg_len_pair): Ditto.
22197 (vect_prune_runtime_alias_test_list): Ditto.
22199 2017-05-26 Martin Liska <mliska@suse.cz>
22202 * params.def: Bound partial-inlining-entry-probability param.
22204 2017-05-26 Marek Polacek <polacek@redhat.com>
22207 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22210 2017-05-26 Richard Biener <rguenther@suse.de>
22212 PR tree-optimization/80842
22213 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22216 2017-05-26 Richard Biener <rguenther@suse.de>
22218 PR tree-optimization/80844
22219 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22221 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22223 * doc/md.texi (Machine Constraints): Update x86 family
22224 machine constraints section to match 'config/i386/constraints.md'.
22226 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22228 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22230 2017-05-25 Nathan Sidwell <nathan@acm.org>
22232 * doc/invoke.texi (--enable-languages): Update documentation.
22234 2017-05-25 Martin Liska <mliska@suse.cz>
22236 * dumpfile.c: Add TDF_FOLDING.
22237 * dumpfile.h (enum dump_kind): Likewise.
22238 * genmatch.c (dt_simplify::gen_1): Use it.
22240 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22242 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22244 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22246 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22247 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22249 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22251 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22252 * match.pd (X == C): Rewrite it here.
22253 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22254 with_certain_nonzero_bits2): New predicates.
22255 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22257 2017-05-24 Nathan Sidwell <nathan@acm.org>
22259 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22262 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22265 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22267 * config/powerpcspe: New port. Files are copied from the rs6000
22268 port, with "rs6000" in filenames replaced by "powerpcspe".
22270 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22272 PR rtl-optimization/80754
22273 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22275 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22277 * config/sparc/sparc.md (length): Return the correct value for -mflat
22278 sibcalls to match output_sibcall.
22280 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22284 * config/rs6000/rs6000.c (struct machine_function): Add new field
22286 (rs6000_get_separate_components): Init that field, use it.
22287 (rs6000_components_for_bb): Use the field.
22289 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22291 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22293 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22295 PR middle-end/80823
22296 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22298 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22301 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22302 address against address_operand predicate.
22303 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22305 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22307 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22308 parameters passed indirectly.
22310 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22312 * config/i386/i386.md (*movdi_internal): Remove SSE4
22313 alternative 18 (?r, *v). Update insn attributes.
22314 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22315 Update insn attributes.
22316 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22317 Update insn attributes.
22318 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22319 alternative 1 (r, v). Remove isa attribute.
22320 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22321 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22322 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22324 2017-05-23 Tom de Vries <tom@codesourcery.com>
22326 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22329 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22331 * cgraphunit.c (symbol_table::process_new_functions): Update.
22332 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22333 (inline_generate_summary): Rename to ...
22334 (ipa_fn_summary_generate): ... this one.
22335 (inline_read_summary): Rename to ...
22336 (ipa_fn_summary_read): ... this one.
22337 (inline_write_summary): Rename to ...
22338 (ipa_fn_summary_write): ... this one.
22339 (inline_free_summary): Rename to ...
22340 (ipa_free_fn_summary): ... this one.
22341 (pass_data_local_fn_summary, pass_local_fn_summary,
22342 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22343 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22344 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22345 make_pass_ipa_fn_summary): New.
22346 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22347 inline_write_summary, inline_free_summary): Remove.
22348 (ipa_free_fn_summary) : New.
22349 * ipa-inline.c (ipa_inline): Update.
22350 (pass_ipa_inline): Do not generate summaries.
22351 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22353 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22354 and add pass_ipa_fn_summary.
22355 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22357 (make_pass_inline_parameters): Remove.
22359 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22361 * omp-low.c (struct omp_context): Remove "default_kind" member.
22364 * omp-offload.c (execute_oacc_device_lower): Remove the
22365 parallelism dimensions function attributes for unparallelized
22366 OpenACC kernels constructs.
22368 2017-05-23 Martin Liska <mliska@suse.cz>
22370 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22372 (cgraph_edge::make_speculative): Likewise.
22373 (cgraph_edge::resolve_speculation): Likewise.
22374 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22375 (cgraph_node::dump): Likewise.
22376 * cgraph.h: Likewise.
22377 * cgraphunit.c (analyze_functions): Likewise.
22378 (symbol_table::compile): Likewise.
22379 * ipa-cp.c (print_all_lattices): Likewise.
22380 (determine_versionability): Likewise.
22381 (initialize_node_lattices): Likewise.
22382 (ipcp_verify_propagated_values): Likewise.
22383 (estimate_local_effects): Likewise.
22384 (update_profiling_info): Likewise.
22385 (create_specialized_node): Likewise.
22386 (perhaps_add_new_callers): Likewise.
22387 (decide_about_value): Likewise.
22388 (decide_whether_version_node): Likewise.
22389 (identify_dead_nodes): Likewise.
22390 (ipcp_store_bits_results): Likewise.
22391 * ipa-devirt.c (dump_targets): Likewise.
22392 (ipa_devirt): Likewise.
22393 * ipa-icf.c (sem_item::dump): Likewise.
22394 (sem_function::equals): Likewise.
22395 (sem_variable::equals): Likewise.
22396 (sem_item_optimizer::read_section): Likewise.
22397 (sem_item_optimizer::execute): Likewise.
22398 (congruence_class::dump): Likewise.
22399 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22400 (dump_inline_summary): Likewise.
22401 (estimate_node_size_and_time): Likewise.
22402 (inline_analyze_function): Likewise.
22403 * ipa-inline-transform.c (inline_call): Likewise.
22404 * ipa-inline.c (report_inline_failed_reason): Likewise.
22405 (want_early_inline_function_p): Likewise.
22406 (edge_badness): Likewise.
22407 (update_edge_key): Likewise.
22408 (inline_small_functions): Likewise.
22409 * ipa-profile.c (ipa_profile): Likewise.
22410 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22411 (ipa_make_edge_direct_to_target): Likewise.
22412 (remove_described_reference): Likewise.
22413 (ipa_impossible_devirt_target): Likewise.
22414 (propagate_controlled_uses): Likewise.
22415 (ipa_print_node_params): Likewise.
22416 (ipcp_transform_function): Likewise.
22417 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22418 (propagate_pure_const): Likewise.
22419 * ipa-reference.c (generate_summary): Likewise.
22420 (read_write_all_from_decl): Likewise.
22421 (propagate): Likewise.
22422 (ipa_reference_read_optimization_summary): Likewise.
22423 * ipa-utils.c (ipa_merge_profiles): Likewise.
22424 * ipa.c (walk_polymorphic_call_targets): Likewise.
22425 (symbol_table::remove_unreachable_nodes): Likewise.
22426 (ipa_single_use): Likewise.
22427 * passes.c (execute_todo): Likewise.
22428 * predict.c (drop_profile): Likewise.
22429 * symtab.c (symtab_node::get_dump_name): New function.
22430 (symtab_node::dump_name): Likewise.
22431 (symtab_node::dump_asm_name): Likewise.
22432 (symtab_node::dump_references): Likewise.
22433 (symtab_node::dump_referring): Likewise.
22434 (symtab_node::dump_base): Likewise.
22435 (symtab_node::debug_symtab): Likewise.
22436 * tree-sra.c (convert_callers_for_node): Likewise.
22437 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22438 * value-prof.c (init_node_map): Likewise.
22440 2017-05-23 Martin Liska <mliska@suse.cz>
22442 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22443 and symtab_node::debug_symtab to symbol_table::debug.
22444 * cgraphunit.c (analyze_functions): Use the renamed function.
22445 (symbol_table::compile): Likewise.
22446 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22447 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22448 * passes.c (execute_todo): Likewise.
22449 * symtab.c (symbol_table::dump): New function.
22450 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22452 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22454 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22455 that nonconst implies exec.
22457 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22459 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22460 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22461 (inline_edge_summary_vec): Turn into ...
22462 (ipa_call_summaries): ... this one.
22463 (redirect_to_unreachable, edge_set_predicate,
22464 evaluate_properties_for_edge, inline_summary_alloc,
22465 reset_ipa_call_summary, reset_inline_summary,
22466 inline_summary_t::duplicate): Update.
22467 (inline_edge_duplication_hook): Turn to ...
22468 (ipa_call_summary_t::duplicate): ... this one.
22469 (inline_edge_removal_hook): Turn to ...
22470 (ipa_call_summary_t::remove): ... this one.
22471 (dump_inline_edge_summary): Turn to ...
22472 (dump_ipa_call_summary): ... this one.
22473 (estimate_function_body_sizes): Update.
22474 (inline_update_callee_summaries): Update.
22475 (remap_edge_change_prob): Update.
22476 (remap_edge_summaries): Update.
22477 (inline_merge_summary): Update.
22478 (do_estimate_edge_time): Update.
22479 (inline_generate_summary): Update.
22480 (inline_read_section): Update.
22481 (inline_read_summary): Update.
22482 (inline_free_summary): Update.
22483 * ipa-inline.c (can_inline_edge_p): Update.
22484 (compute_inlined_call_time): Update.
22485 (want_inline_small_function_p): Update.
22486 (edge_badness): Update.
22487 (early_inliner): Update.
22488 * ipa-inline.h (inline_edge_summary): Turn to ...
22489 (ipa_call_summary): ... this one.
22490 (ipa_call_summary_t): New class.
22491 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22492 (ipa_call_summaries): New.
22493 (inline_edge_summary): Remove.
22494 (estimate_edge_growth): Update.
22495 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22496 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22497 * ipa-split.c (execute_split_functions): Update.
22498 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22500 2017-05-23 Tom de Vries <tom@codesourcery.com>
22502 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22503 attributes): Document rdrand effective target.
22505 2017-05-23 Tom de Vries <tom@codesourcery.com>
22507 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22508 attributes): Sort alphabetically.
22510 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22512 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22514 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22517 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22518 V2DF/V2DI splat into two separate patterns, one that handles
22519 registers, and the other that only handles memory. Drop support
22520 for splatting from a GPR on ISA 2.07 and then splitting the
22521 splat into direct move and splat.
22522 (vsx_splat_<mode>_reg): Likewise.
22523 (vsx_splat_<mode>_mem): Likewise.
22525 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22527 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22529 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22531 PR middle-end/80809
22532 * omp-low.c (finish_taskreg_remap): New function.
22533 (finish_taskreg_scan): If unit size of ctx->record_type
22534 is non-constant, unshare the size expression and replace
22535 decls in it with possible outer var refs.
22537 PR middle-end/80809
22538 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22539 GOVD_SHARED rather than GOVD_PRIVATE with it.
22540 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22541 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22543 PR middle-end/80853
22544 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22545 as last argument to build_outer_var_ref for pointer bases of array
22546 section reductions.
22548 2017-05-19 Martin Sebor <msebor@redhat.com>
22550 * print-tree.c (print_node): Print DECL_READ_P flag.
22552 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22554 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22555 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22556 * cgraph.c: Likewise.
22557 * cgraphunit.c: Likewise.
22558 * gengtype.c: Likewise.
22559 * ipa-cp.c: Likewise.
22560 * ipa-devirt.c: Likewise.
22561 * ipa-icf.c: Likewise.
22562 * ipa-predicate.c: Likewise.
22563 * ipa-profile.c: Likewise.
22564 * ipa-prop.c: Likewise.
22565 * ipa-split.c: Likewise.
22567 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22568 edge_predicate_pool, dump_inline_hints,
22569 inline_summary::account_size_time, redirect_to_unreachable,
22570 edge_set_predicate, set_hint_predicate,
22571 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22572 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22573 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22574 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22575 ipa_call_summary_t::remove, initialize_growth_caches,
22576 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22577 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22578 mark_modified, unmodified_parm_1, unmodified_parm,
22579 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22580 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22581 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22582 will_be_nonconstant_predicate, record_modified_bb_info,
22583 get_minimal_bb, record_modified, param_change_prob,
22584 phi_result_unknown_predicate, predicate_for_phi_result,
22585 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22586 estimate_function_body_sizes, compute_inline_parameters,
22587 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22588 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22589 inline_update_callee_summaries, remap_edge_change_prob,
22590 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22591 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22592 inline_analyze_function, inline_summary_t::insert,
22593 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22594 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22595 inline_free_summary): Move to ipa-fnsummary.h
22596 (predicate_t): Remove.
22597 * ipa-fnsummary.c: New file.
22598 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22599 (enum inline_hints_vals, inline_hints, agg_position_info,
22600 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22601 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22602 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22603 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22604 inline_read_summary, inline_write_summary, inline_free_summary,
22605 inline_analyze_function, initialize_inline_failed,
22606 inline_merge_summary, inline_update_overall_summary,
22607 compute_inline_parameters): Move to ipa-fnsummary.h
22608 * ipa-fnsummary.h: New file.
22609 * ipa-inline-transform.h: Include ipa-inline.h.
22610 * ipa-inline.c: LIkewise.
22612 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22614 * ipa-inline.c (edge_badness): Use inlined_time instead of
22615 inline_summaries->get.
22617 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22619 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22621 2017-05-22 Nathan Sidwell <nathan@acm.org>
22623 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22624 (fdump-lang): Document 'raw' option.
22625 * dumpfile.h (TDI_tu): Delete.
22626 * dumpfile.c (dump_files): Remove translation-unit.
22627 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22629 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22631 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22632 command option from $(AWK) call.
22633 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22634 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22635 [FORMAT]: Remove handling of variable.
22636 * config/avr/t-multilib: Regenerate.
22638 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22640 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22642 (dump_inline_summary): Do not print self_time.
22643 (estimate_function_body_sizes): Do not set self_time.
22644 (compute_inline_parameters): Likewise.
22645 (inline_read_section, inline_write_summary): Do not stream self_time.
22646 * ipa-inline.h (inline_summary): Drop self_time.
22648 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22650 * ipa-inline-analysis.c (account_size_time): Rename to ...
22651 (inline_summary::account_size_time): ... this one.
22652 (reset_ipa_call_summary): Turn to ...
22653 (ipa_call_summary::reset): ... this one.
22654 (reset_inline_summary): Turn to ...
22655 (inline_summary::reset): ... this one.
22656 (inline_summary_t::remove): Update.
22657 (inline_summary_t::duplicate): Update.
22658 (ipa_call_summary_t::remove): Update.
22659 (dump_inline_summary): Update.
22660 (estimate_function_body_sizes): Update.
22661 (compute_inline_parameters): Update.
22662 (estimate_node_size_and_time): Update.
22663 (inline_merge_summary): Update.
22664 (inline_update_overall_summary): Update.
22665 (inline_read_section): Update.
22666 (inline_write_summary): Update.
22667 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22668 add account_size_time and reset member functions.
22669 (ipa_call_summary): Add reset function.
22670 * ipa-predicate.h (predicate::operator &): Constify.
22672 2017-05-22 Richard Biener <rguenther@suse.de>
22674 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22676 2017-05-19 Jason Merrill <jason@redhat.com>
22678 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22680 2017-05-19 Marek Polacek <polacek@redhat.com>
22683 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22684 TYPE_OVERFLOW_WRAPS checks.
22686 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22688 * tree-core.h (enum omp_clause_default_kind): Add
22689 "OMP_CLAUSE_DEFAULT_PRESENT".
22690 * tree-pretty-print.c (dump_omp_clause): Handle it.
22691 * gimplify.c (enum gimplify_omp_var_data): Add
22692 "GOVD_MAP_FORCE_PRESENT".
22693 (gimplify_adjust_omp_clauses_1): Map it to
22694 "GOMP_MAP_FORCE_PRESENT".
22695 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22697 * gimplify.c (oacc_default_clause): Clarify.
22699 2017-05-19 Nathan Sidwell <nathan@acm.org>
22701 LANG_HOOK_REGISTER_DUMPS
22702 * toplev.c (general_init): Call register dump lang hook.
22703 * doc/invoke.texi: Document -fdump-lang option family.
22704 * dumpfile.c (dump_files): Remove class dump here.
22705 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22706 * dumpfile.h (tree_dump_index): Remove TDI_class.
22707 * langhooks-def.h (lhd_register_dumps): Declare.
22708 (LANG_HOOKS_REGISTER_DUMPS): Define.
22709 (LANG_HOOKS_INITIALIZER): Add it.
22710 * langhooks.c (lhd_register_dumps): Define.
22711 * langhooks.h (struct lang_hooks): Add register_dumps.
22713 2017-05-19 Nathan Sidwell <nathan@acm.org>
22715 * context.h (context::set_passes): New.
22716 * context.c (context::context): Do not create pass manager.
22717 * toplev.c (general_init): Create pass manager here.
22719 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22721 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22722 use this splitter if two add or or instructions would also work for
22723 the constant we want to generate.
22725 2017-05-19 Richard Biener <rguenther@suse.de>
22728 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22729 predicate evaluation.
22731 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22733 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22735 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22738 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22740 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22741 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22742 (inline_edge_summary_vec): Turn into ...
22743 (ipa_call_summaries): ... this one.
22744 (redirect_to_unreachable, edge_set_predicate,
22745 evaluate_properties_for_edge, inline_summary_alloc,
22746 reset_ipa_call_summary, reset_inline_summary,
22747 inline_summary_t::duplicate): Update.
22748 (inline_edge_duplication_hook): Turn to ...
22749 (ipa_call_summary_t::duplicate): ... this one.
22750 (inline_edge_removal_hook): Turn to ...
22751 (ipa_call_summary_t::remove): ... this one.
22752 (dump_inline_edge_summary): Turn to ...
22753 (dump_ipa_call_summary): ... this one.
22754 (estimate_function_body_sizes): Update.
22755 (inline_update_callee_summaries): Update.
22756 (remap_edge_change_prob): Update.
22757 (remap_edge_summaries): Update.
22758 (inline_merge_summary): Update.
22759 (do_estimate_edge_time): Update.
22760 (inline_generate_summary): Update.
22761 (inline_read_section): Update.
22762 (inline_read_summary): Update.
22763 (inline_free_summary): Update.
22764 * ipa-inline.c (can_inline_edge_p): Update.
22765 (compute_inlined_call_time): Update.
22766 (want_inline_small_function_p): Update.
22767 (edge_badness): Update.
22768 (early_inliner): Update.
22769 * ipa-inline.h (inline_edge_summary): Turn to ...
22770 (ipa_call_summary): ... this one.
22771 (ipa_call_summary_t): New class.
22772 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22773 (ipa_call_summaries): New.
22774 (inline_edge_summary): Remove.
22775 (estimate_edge_growth): Update.
22776 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22777 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22778 * ipa-split.c (execute_split_functions): Update.
22779 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22781 2017-05-19 Richard Biener <rguenther@suse.de>
22783 PR middle-end/80764
22784 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22786 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22788 * config/rs6000/rs6000.c (struct machine_function): Add field
22789 fpr_is_wrapped_separately.
22790 (rs6000_get_separate_components): Use 64 components. Handle the
22791 new FPR components.
22792 (rs6000_components_for_bb): Handle the FPR components.
22793 (rs6000_emit_prologue_components): Handle the FPR components.
22794 (rs6000_emit_epilogue_components): Handle the FPR components.
22795 (rs6000_set_handled_components): Handle the FPR components.
22796 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22797 that are already separately shrink-wrapped.
22798 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22799 that are already separately shrink-wrapped.
22801 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22804 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22807 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22808 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22809 cases where the register allocator uses a move and an offsettable
22810 memory operation to/from a FPR register on ISA 2.06/2.07.
22811 (define_peephole2 for Altivec d-form store): Likewise.
22813 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22816 * config/i386/mmx.md (*mov<mode>_internal): Enable
22817 alternatives 11, 12, 13 and 14 also for 32bit targets.
22818 Remove alternatives 15, 16, 17 and 18.
22819 * config/i386/sse.md (vec_concatv2di): Change
22820 alternative (!x, *y) to (x, ?!*Yn).
22822 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22824 * dumpfile.h (enum dump_kind): Remove stray comma.
22826 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22828 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22829 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22830 predicate::num_conditions
22831 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22832 (CHANGED): turn into predicate::changed.
22833 (agg_position_info): Move to ipa-predicate.h
22834 (add_condition, predicate::add_clause, predicate::operator &=,
22835 predicate::or_with, predicate::evaluate, predicate::probability,
22836 dump_condition, dump_clause, predicate::dump,
22837 predicate::remap_after_duplication, predicate::remap_after_inlining,
22838 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22839 (evaluate_conditions_for_known_args): Update.
22840 (set_cond_stmt_execution_predicate): Update.
22841 * ipa-inline.h: Include ipa-predicate.h
22842 (condition, inline_param_summary, conditions, agg_position_info,
22843 predicate): Move to ipa-predicate.h
22844 * ipa-predicate.c: New file.
22845 * ipa-predicate.h: New file.
22847 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22849 * final.c (leaf_function_p): Check we are not in a sequence.
22851 2017-05-18 Martin Liska <mliska@suse.cz>
22853 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22854 * dumpfile.c (dump_register): Use new enum dump_kind.
22855 (get_dump_file_name): Likewise.
22856 (dump_enable_all): Likewise.
22857 (dump_switch_p_1): Likewise.
22858 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22859 * dumpfile.h (enum dump_kind): New enum type.
22860 (struct dump_file_info): Create constructor and
22861 format fields and comments.
22862 * passes.c (pass_manager::register_one_dump_file):
22864 * statistics.c (statistics_early_init): Likewise.
22865 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22866 TDF_TREE with TDF_SLIM.
22867 (gather_memory_references_ref): Likewise.
22869 2017-05-18 Martin Liska <mliska@suse.cz>
22871 * vec.h (struct vnull): Use it.
22873 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22875 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22876 (true_predicate, false_predicate, true_predicate_p,
22877 false_predicate_p): Remove.
22878 (single_cond_predicate, not_inlined_predicate): Turn to member function
22880 (add_condition): Update.
22881 (add_clause): Turn to...
22882 (predicate::add_clause): ... this one; update; allow passing NULL
22884 (and_predicates): Turn to ...
22885 (predicate::operator &=): ... this one.
22886 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22887 (or_predicates): Turn to ...
22888 (predicate::or_with): ... this one.
22889 (evaluate_predicate): Turn to ...
22890 (predicate::evaluate): ... this one.
22891 (predicate_probability): Turn to ...
22892 (predicate::probability): ... this one.
22893 (dump_condition): Update.
22894 (dump_predicate): Turn to ...
22895 (predicate::dump): ... this one.
22896 (account_size_time): Update.
22897 (edge_set_predicate): Update.
22898 (set_hint_predicate): UPdate.
22899 (evaluate_conditions_for_known_args): Update.
22900 (evaluate_properties_for_edge): Update.
22901 (remap_predicate_after_duplication): Turn to...
22902 (predicate::remap_after_duplication): ... this one.
22903 (remap_hint_predicate_after_duplication): Update.
22904 (inline_summary_t::duplicate): UPdate.
22905 (dump_inline_edge_summary): Update.
22906 (dump_inline_summary): Update.
22907 (set_cond_stmt_execution_predicate): Update.
22908 (set_switch_stmt_execution_predicate): Update.
22909 (compute_bb_predicates): Update.
22910 (will_be_nonconstant_expr_predicate): Update.
22911 (will_be_nonconstant_predicate): Update.
22912 (phi_result_unknown_predicate): Update.
22913 (predicate_for_phi_result): Update.
22914 (array_index_predicate): Update.
22915 (estimate_function_body_sizes): Update.
22916 (estimate_node_size_and_time): Update.
22917 (estimate_ipcp_clone_size_and_time): Update.
22918 (remap_predicate): Rename to ...
22919 (predicate::remap_after_inlining): ... this one.
22920 (remap_hint_predicate): Update.
22921 (inline_merge_summary): Update.
22922 (inline_update_overall_summary): Update.
22923 (estimate_size_after_inlining): Update.
22924 (read_predicate): Rename to ...
22925 (predicate::stream_in): ... this one.
22926 (read_inline_edge_summary): Update.
22927 (write_predicate): Rename to ...
22928 (predicate::stream_out): ... this one.
22929 (write_inline_edge_summary): Update.
22930 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22931 (clause_t): Turn to uint32_t
22932 (predicate): Turn to class; implement constructor and operators
22934 (size_time_entry): Update.
22935 (inline_summary): Update.
22936 (inline_edge_summary): Update.
22938 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22940 * fold-const.c (fold_binary_loc): Move transformation...
22941 * match.pd (C - X CMP X): ... here.
22943 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22945 * config/sparc/sparc.c (sparc_option_override): Set function
22946 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22947 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22949 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22951 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22953 2017-05-18 Marek Polacek <polacek@redhat.com>
22956 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22957 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22958 gimple_assign_load_p.
22960 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22962 PR middle-end/80692
22963 * real.c (do_compare): Give decimal_do_compare preference over
22964 comparing just the signs.
22966 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22968 * doc/md.texi (Canonicalization of Instructions): Describe the
22969 canonical form of instructions that inherently set a condition
22972 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22974 PR middle-end/80775
22975 * tree-cfg.c: Move deletion of unreachable case statements to after
22976 the merging of consecutive case labels.
22978 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22980 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22981 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22982 restoring of callee-saved registers.
22984 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22986 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22987 * config/visium/visium.c (single_set_and_flags): Likewise.
22988 * config/visium/visium.md (Substitutions): Likewise.
22990 2017-05-17 Martin Liska <mliska@suse.cz>
22992 * cfg.c: Introduce dump_flags_t type and
22993 use it instead of int type.
22995 * cfghooks.c: Likewise.
22996 * cfghooks.h (struct cfg_hooks): Likewise.
22997 * cfgrtl.c: Likewise.
22998 * cfgrtl.h: Likewise.
22999 * cgraph.c (cgraph_node::get_body): Likewise.
23000 * coretypes.h: Likewise.
23001 * domwalk.c: Likewise.
23002 * domwalk.h: Likewise.
23003 * dumpfile.c (struct dump_option_value_info): Likewise.
23004 (dump_enable_all): Likewise.
23005 (dump_switch_p_1): Likewise.
23006 (opt_info_switch_p): Likewise.
23007 * dumpfile.h (enum tree_dump_index): Likewise.
23008 (struct dump_file_info): Likewise.
23009 * genemit.c: Likewise.
23010 * generic-match-head.c: Likewise.
23011 * gengtype.c (open_base_files): Likewise.
23012 * gimple-pretty-print.c: Likewise.
23013 * gimple-pretty-print.h: Likewise.
23014 * graph.c (print_graph_cfg): Likewise.
23015 * graphite-scop-detection.c (dot_all_sese): Likewise.
23016 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23017 * loop-unroll.c (report_unroll): Likewise.
23018 * passes.c (pass_manager::register_one_dump_file): Likewise.
23019 * print-tree.c: Likewise.
23020 * statistics.c: Likewise.
23021 * tree-cfg.c: Likewise.
23022 * tree-cfg.h: Likewise.
23023 * tree-dfa.c: Likewise.
23024 * tree-dfa.h: Likewise.
23025 * tree-dump.c (dump_function): Likewise.
23026 * tree-dump.h (struct dump_info): Likewise.
23027 * tree-pretty-print.c: Likewise.
23028 * tree-pretty-print.h: Likewise.
23029 * tree-ssa-live.c: Likewise.
23030 * tree-ssa-live.h: Likewise.
23031 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23032 * tree-vect-loop.c: Likewise.
23033 * tree-vect-slp.c: Likewise.
23035 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23036 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23038 PR tree-optimization/80457
23039 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23040 of all arguments to a statement as scalar_to_vec operations.
23041 (vectorizable_call): Adjust call to vect_model_simple_cost for
23043 (vectorizable_conversion): Likewise.
23044 (vectorizable_assignment): Likewise.
23045 (vectorizable_shift): Likewise.
23046 (vectorizable_operation): Likewise.
23047 (vectorizable_comparison): Likewise.
23048 (vect_is_simple_cond): Record the def types for operands.
23049 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23050 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23051 for statement argument count.
23053 2017-05-16 Carl Love <cel@us.ibm.com>
23055 * config/rs6000/rs6000-c: Add support for built-in functions
23056 vector unsigned long long vec_bperm (vector unsigned long long,
23057 vector unsigned char)
23058 vector signed long long vec_mule (vector signed int,
23060 vector unsigned long long vec_mule (vector unsigned int,
23061 vector unsigned int)
23062 vector signed long long vec_mulo (vector signed int,
23064 vector unsigned long long vec_mulo (vector unsigned int,
23065 vector unsigned int)
23066 vector signed char vec_sldw (vector signed char,
23067 vector signed char,
23069 vector unsigned char vec_sldw (vector unsigned char,
23070 vector unsigned char,
23072 vector signed short vec_sldw (vector signed short,
23073 vector signed short,
23075 vector unsigned short vec_sldw (vector unsigned short,
23076 vector unsigned short,
23078 vector signed int vec_sldw (vector signed int,
23081 vector unsigned int vec_sldw (vector unsigned int,
23082 vector unsigned int,
23084 vector signed long long vec_sldw (vector signed long long,
23085 vector signed long long,
23087 vector unsigned long long vec_sldw (vector unsigned long long,
23088 vector unsigned long long,
23090 * config/rs6000/rs6000-c: Add support for built-in functions
23091 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23092 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23093 * doc/extend.texi: Update the built-in documentation for the
23094 new built-in functions.
23096 2017-05-16 Marek Polacek <polacek@redhat.com>
23100 * tree.c (save_expr): Don't fold the expression.
23102 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23104 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23105 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23106 and (?*y,m). Update insn attributes.
23108 2017-05-16 Martin Liska <mliska@suse.cz>
23110 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23111 flags argument of print_gimple_stmt, print_gimple_expr,
23112 print_generic_stmt and print_generic_expr.
23113 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23114 * coretypes.h: Likewise.
23115 * except.c (dump_eh_tree): Likewise.
23116 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23117 * gimple-pretty-print.h: Likewise.
23118 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23119 (backprop::push_to_worklist): Likewise.
23120 (backprop::pop_from_worklist): Likewise.
23121 (backprop::process_use): Likewise.
23122 (backprop::intersect_uses): Likewise.
23123 (note_replacement): Likewise.
23124 * gimple-ssa-store-merging.c
23125 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23126 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23127 (pass_store_merging::execute): Likewise.
23128 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23129 (ssa_base_cand_dump_callback): Likewise.
23130 (dump_incr_vec): Likewise.
23131 (replace_refs): Likewise.
23132 (replace_mult_candidate): Likewise.
23133 (create_add_on_incoming_edge): Likewise.
23134 (create_phi_basis): Likewise.
23135 (insert_initializers): Likewise.
23136 (all_phi_incrs_profitable): Likewise.
23137 (introduce_cast_before_cand): Likewise.
23138 (replace_one_candidate): Likewise.
23139 * gimplify.c (gimplify_expr): Likewise.
23140 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23141 (set_rename): Likewise.
23142 (rename_uses): Likewise.
23143 (copy_loop_phi_nodes): Likewise.
23144 (add_close_phis_to_merge_points): Likewise.
23145 (copy_loop_close_phi_args): Likewise.
23146 (copy_cond_phi_args): Likewise.
23147 (graphite_copy_stmts_from_block): Likewise.
23148 (translate_pending_phi_nodes): Likewise.
23149 * graphite-poly.c (print_pdr): Likewise.
23150 (dump_gbb_cases): Likewise.
23151 (dump_gbb_conditions): Likewise.
23152 (print_scop_params): Likewise.
23153 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23154 (build_cross_bb_scalars_use): Likewise.
23155 (gather_bbs::before_dom_children): Likewise.
23156 * hsa-dump.c (dump_hsa_immed): Likewise.
23157 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23158 (get_replacement_map): Likewise.
23159 * ipa-inline-analysis.c (dump_condition): Likewise.
23160 (estimate_function_body_sizes): Likewise.
23161 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23162 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23163 * ipa-prop.c (ipa_dump_param): Likewise.
23164 (ipa_print_node_jump_functions_for_edge): Likewise.
23165 (ipa_modify_call_arguments): Likewise.
23166 (ipa_modify_expr): Likewise.
23167 (ipa_dump_param_adjustments): Likewise.
23168 (ipa_dump_agg_replacement_values): Likewise.
23169 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23170 * ipa-pure-const.c (check_stmt): Likewise.
23171 (pass_nothrow::execute): Likewise.
23172 * ipa-split.c (execute_split_functions): Likewise.
23173 * omp-offload.c (dump_oacc_loop_part): Likewise.
23174 (dump_oacc_loop): Likewise.
23175 * trans-mem.c (tm_log_emit): Likewise.
23176 (tm_memopt_accumulate_memops): Likewise.
23177 (dump_tm_memopt_set): Likewise.
23178 (dump_tm_memopt_transform): Likewise.
23179 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23180 (print_loop): Likewise.
23181 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23182 (chkp_gather_checks_info): Likewise.
23183 (chkp_get_check_result): Likewise.
23184 (chkp_remove_check_if_pass): Likewise.
23185 (chkp_use_outer_bounds_if_possible): Likewise.
23186 (chkp_reduce_bounds_lifetime): Likewise.
23187 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23188 (chkp_mark_completed_bounds): Likewise.
23189 (chkp_register_incomplete_bounds): Likewise.
23190 (chkp_mark_invalid_bounds): Likewise.
23191 (chkp_maybe_copy_and_register_bounds): Likewise.
23192 (chkp_build_returned_bound): Likewise.
23193 (chkp_get_bound_for_parm): Likewise.
23194 (chkp_build_bndldx): Likewise.
23195 (chkp_get_bounds_by_definition): Likewise.
23196 (chkp_generate_extern_var_bounds): Likewise.
23197 (chkp_get_bounds_for_decl_addr): Likewise.
23198 * tree-chrec.c (chrec_apply): Likewise.
23199 * tree-data-ref.c (dump_data_reference): Likewise.
23200 (dump_subscript): Likewise.
23201 (dump_data_dependence_relation): Likewise.
23202 (analyze_overlapping_iterations): Likewise.
23203 * tree-inline.c (expand_call_inline): Likewise.
23204 (tree_function_versioning): Likewise.
23205 * tree-into-ssa.c (dump_defs_stack): Likewise.
23206 (dump_currdefs): Likewise.
23207 (dump_names_replaced_by): Likewise.
23208 (dump_update_ssa): Likewise.
23209 (update_ssa): Likewise.
23210 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23211 * tree-parloops.c (build_new_reduction): Likewise.
23212 (try_create_reduction_list): Likewise.
23213 (ref_conflicts_with_region): Likewise.
23214 (oacc_entry_exit_ok_1): Likewise.
23215 (oacc_entry_exit_single_gang): Likewise.
23216 * tree-pretty-print.h: Likewise.
23217 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23218 (get_scalar_evolution): Likewise.
23219 (add_to_evolution): Likewise.
23220 (get_loop_exit_condition): Likewise.
23221 (analyze_evolution_in_loop): Likewise.
23222 (analyze_initial_condition): Likewise.
23223 (analyze_scalar_evolution): Likewise.
23224 (instantiate_scev): Likewise.
23225 (number_of_latch_executions): Likewise.
23226 (gather_chrec_stats): Likewise.
23227 (final_value_replacement_loop): Likewise.
23228 (scev_const_prop): Likewise.
23229 * tree-sra.c (dump_access): Likewise.
23230 (disqualify_candidate): Likewise.
23231 (create_access): Likewise.
23232 (reject): Likewise.
23233 (maybe_add_sra_candidate): Likewise.
23234 (create_access_replacement): Likewise.
23235 (analyze_access_subtree): Likewise.
23236 (analyze_all_variable_accesses): Likewise.
23237 (sra_modify_assign): Likewise.
23238 (initialize_constant_pool_replacements): Likewise.
23239 (find_param_candidates): Likewise.
23240 (decide_one_param_reduction): Likewise.
23241 (replace_removed_params_ssa_names): Likewise.
23242 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23243 * tree-ssa-copy.c (dump_copy_of): Likewise.
23244 (copy_prop_visit_cond_stmt): Likewise.
23245 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23246 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23247 (record_equivalences_from_stmt): Likewise.
23248 * tree-ssa-dse.c (compute_trims): Likewise.
23249 (delete_dead_call): Likewise.
23250 (delete_dead_assignment): Likewise.
23251 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23252 (forward_propagate_into_cond): Likewise.
23253 (pass_forwprop::execute): Likewise.
23254 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23255 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23257 (move_computations_worker): Likewise.
23258 (execute_sm): Likewise.
23259 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23260 (remove_exits_and_undefined_stmts): Likewise.
23261 (remove_redundant_iv_tests): Likewise.
23262 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23263 (adjust_iv_update_pos): Likewise.
23264 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23265 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23266 (value_replacement): Likewise.
23267 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23268 * tree-ssa-pre.c (print_pre_expr): Likewise.
23269 (get_representative_for): Likewise.
23270 (create_expression_by_pieces): Likewise.
23271 (insert_into_preds_of_block): Likewise.
23272 (eliminate_insert): Likewise.
23273 (eliminate_dom_walker::before_dom_children): Likewise.
23274 (eliminate): Likewise.
23275 (remove_dead_inserted_code): Likewise.
23276 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23277 * tree-ssa-reassoc.c (get_rank): Likewise.
23278 (eliminate_duplicate_pair): Likewise.
23279 (eliminate_plus_minus_pair): Likewise.
23280 (eliminate_not_pairs): Likewise.
23281 (undistribute_ops_list): Likewise.
23282 (eliminate_redundant_comparison): Likewise.
23283 (update_range_test): Likewise.
23284 (optimize_range_tests_var_bound): Likewise.
23285 (optimize_vec_cond_expr): Likewise.
23286 (rewrite_expr_tree): Likewise.
23287 (rewrite_expr_tree_parallel): Likewise.
23288 (linearize_expr): Likewise.
23289 (break_up_subtract): Likewise.
23290 (linearize_expr_tree): Likewise.
23291 (attempt_builtin_powi): Likewise.
23292 (attempt_builtin_copysign): Likewise.
23293 (transform_stmt_to_copy): Likewise.
23294 (transform_stmt_to_multiply): Likewise.
23295 (dump_ops_vector): Likewise.
23296 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23297 (print_scc): Likewise.
23298 (set_ssa_val_to): Likewise.
23299 (visit_reference_op_store): Likewise.
23300 (visit_use): Likewise.
23301 (sccvn_dom_walker::before_dom_children): Likewise.
23302 (run_scc_vn): Likewise.
23303 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23305 (expr_hash_elt::print): Likewise.
23306 (const_and_copies::pop_to_marker): Likewise.
23307 (const_and_copies::record_const_or_copy_raw): Likewise.
23308 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23309 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23310 (dump_predicates): Likewise.
23311 (find_uninit_use): Likewise.
23312 (warn_uninitialized_phi): Likewise.
23313 (pass_late_warn_uninitialized::execute): Likewise.
23314 * tree-ssa.c (verify_vssa): Likewise.
23315 (verify_ssa): Likewise.
23316 (maybe_optimize_var): Likewise.
23317 * tree-vrp.c (dump_value_range): Likewise.
23318 (dump_all_value_ranges): Likewise.
23319 (dump_asserts_for): Likewise.
23320 (register_edge_assert_for_2): Likewise.
23321 (vrp_visit_cond_stmt): Likewise.
23322 (vrp_visit_switch_stmt): Likewise.
23323 (vrp_visit_stmt): Likewise.
23324 (vrp_visit_phi_node): Likewise.
23325 (simplify_cond_using_ranges_1): Likewise.
23326 (fold_predicate_in): Likewise.
23327 (evrp_dom_walker::before_dom_children): Likewise.
23328 (evrp_dom_walker::push_value_range): Likewise.
23329 (evrp_dom_walker::pop_value_range): Likewise.
23330 (execute_early_vrp): Likewise.
23332 2017-05-16 Richard Biener <rguenther@suse.de>
23334 * dwarf2out.c (loc_list_from_tree_1): Do not create
23335 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23337 2017-05-16 Richard Biener <rguenther@suse.de>
23339 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23341 (note_variable_value_in_expr): If we resolved the decl ref
23342 do not push to the stack.
23344 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23346 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23347 operations in fast-math mode.
23348 (vaddq_f16): Likewise.
23349 (vmul_f16): Likewise.
23350 (vmulq_f16): Likewise.
23351 (vsub_f16): Likewise.
23352 (vsubq_f16): Likewise.
23353 * config/arm/neon.md (add<mode>3): New.
23355 (fma:<VH:mode>3): New. Also remove outdated comment.
23358 2017-05-16 Martin Liska <mliska@suse.cz>
23362 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23363 (odr_types_equivalent_p): Likewise.
23365 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23367 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23369 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23372 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23373 non-interunit SSE move alternatives with '?'.
23374 (zero-extendsidi peephole2): New peephole to skip intermediate
23375 general register in SSE zero-extend sequence.
23377 2017-05-15 Jeff Law <law@redhat.com>
23379 * reorg.c (relax_delay_slots): Create a new variable to hold
23380 the temporary target rather than clobbering TARGET_LABEL.
23382 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23383 missing argument to extract_bit_field call.
23384 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23386 2017-05-15 Martin Liska <mliska@suse.cz>
23389 * gcc.c (process_command): Do not allow empty argument of -o option.
23391 2017-05-15 Renlin Li <renlin.li@arm.com>
23393 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23394 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23395 * config/aarch64/constraints.md (Usf): Add long call check.
23396 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23397 (call_value): Likewise.
23398 (sibcall): Likewise.
23399 (sibcall_value): Likewise.
23401 (call_value_insn): New.
23402 (sibcall_insn): Update rtx pattern.
23403 (sibcall_value_insn): Likewise.
23404 (call_internal): Remove.
23405 (call_value_internal): Likewise.
23406 (sibcall_internal): Likewise.
23407 (sibcall_value_internal): Likewise.
23408 (call_reg): Likewise.
23409 (call_symbol): Likewise.
23410 (call_value_reg): Likewise.
23411 (call_value_symbol): Likewise.
23413 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23416 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23418 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23420 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23421 compatible with CCGOCmode and with CCZmode.
23423 2017-05-14 Martin Sebor <msebor@redhat.com>
23425 PR middle-end/77671
23426 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23427 (gimple_fold_builtin_snprintf): Same.
23428 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23429 (gimple_fold_builtin_snprintf): Same.
23430 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23431 of character types.
23432 (is_call_safe): New function.
23433 (try_substitute_return_value): Call it.
23434 (try_simplify_call): New function.
23435 (pass_sprintf_length::handle_gimple_call): Call it.
23437 2017-05-14 Martin Sebor <msebor@redhat.com>
23439 PR middle-end/80669
23440 * builtins.c (expand_builtin_stpncpy): Simplify.
23442 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23444 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23445 * config/i386/i386.h
23446 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23447 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23448 (struct machine_function): Add new members call_ms2sysv,
23449 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23450 (struct machine_frame_state): New fields sp_realigned and
23451 sp_realigned_offset.
23452 * config/i386/i386.c
23453 (enum xlogue_stub): New enum.
23454 (enum xlogue_stub_sets): New enum.
23455 (class xlogue_layout): New class.
23456 (struct ix86_frame): New fields stack_realign_allocate_offset,
23457 stack_realign_offset and outlined_save_offset. Modify comments to
23458 detail stack layout when using out-of-line stubs.
23459 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23460 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23461 -mcall-ms2sysv-xlogues.
23462 (stub_managed_regs): New static variable.
23463 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23464 registers managed by out-of-line stub.
23465 (disable_call_ms2sysv_xlogues): New function.
23466 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23467 m->call_ms2sysv when appropriate and compute frame layout for
23469 (sp_valid_at, fp_valid_at): New inline functions.
23470 (choose_basereg): New function.
23471 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23473 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23474 Use align parameter of choose_baseaddr to generated aligned SSE movs
23476 (pro_epilogue_adjust_stack): Modify to track
23477 machine_frame_state::sp_realigned.
23478 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23479 (ix86_nsaved_sseregs): Likewise.
23480 (ix86_emit_save_regs): Likewise.
23481 (ix86_emit_save_regs_using_mov): Likewise.
23482 (ix86_emit_save_sse_regs_using_mov): Likewise.
23483 (get_scratch_register_on_entry): Likewise.
23484 (gen_frame_set): New function.
23485 (gen_frame_load): Likewise.
23486 (gen_frame_store): Likewise.
23487 (emit_outlined_ms2sysv_save): Likewise.
23488 (emit_outlined_ms2sysv_restore): Likewise.
23489 (ix86_expand_prologue): Modify stack re-alignment code and call
23490 emit_outlined_ms2sysv_save when appropriate.
23491 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23492 parameter rtx_insn *insn, which allows the function to be used to only
23493 generate the notes.
23494 (ix86_expand_epilogue): Modify validity checks of frame and stack
23495 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23496 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23497 * config/i386/predicates.md
23498 (save_multiple): New predicate.
23499 (restore_multiple): Likewise.
23500 * config/i386/sse.md
23501 (save_multiple<mode>): New pattern.
23502 (save_multiple_realign<mode>): Likewise.
23503 (restore_multiple<mode>): Likewise.
23504 (restore_multiple_and_return<mode>): Likewise.
23505 (restore_multiple_leave_return<mode>): Likewise.
23506 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23508 2017-05-14 Julia Koval <julia.koval@intel.com>
23510 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23511 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23512 (__builtin_ia32_xsetbv): New builtins.
23513 * config/i386/i386.c (ix86_expand_special_args_builtin):
23515 (ix86_expand_builtin): Special expand for new intrinsics.
23516 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23517 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23518 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23520 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23522 * cfganal.c (inverted_post_order_compute): Change argument type
23524 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23525 * df-core.c (rest_of_handle_df_initialize): Adjust.
23526 (rest_of_handle_df_finish): Likewise.
23527 (df_analyze_1): Likewise.
23528 (df_analyze): Likewise.
23529 (loop_inverted_post_order_compute): Change argument to be a vec *.
23530 (df_analyze_loop): Adjust.
23531 (df_get_n_blocks): Likewise.
23532 (df_get_postorder): Likewise.
23533 * df.h (struct df_d): Change field to be a vec.
23534 * lcm.c (compute_laterin): Adjust.
23535 (compute_available): Likewise.
23536 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23537 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23538 * tree-ssa-pre.c (compute_antic): Likewise.
23540 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23542 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23543 (depth_first_search::depth_first_search): Change structure init
23544 function to this constructor.
23545 (depth_first_search::add_bb): Rename function to this member.
23546 (depth_first_search::execute): Likewise.
23547 (flow_dfs_compute_reverse_finish): Adjust.
23549 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23551 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23552 (longest_simple_path): Likewise.
23553 * shrink-wrap.c (spread_components): Likewise.
23554 (disqualify_problematic_components): Likewise.
23555 (emit_common_heads_for_components): Likewise.
23556 (emit_common_tails_for_components): Likewise.
23557 (insert_prologue_epilogue_for_components): Likewise.
23559 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23561 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23564 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23566 * df-core.c (df_set_blocks): Start using auto_bitmap.
23567 (df_compact_blocks): Likewise.
23568 * df-problems.c (df_rd_confluence_n): Likewise.
23569 * df-scan.c (df_insn_rescan_all): Likewise.
23570 (df_process_deferred_rescans): Likewise.
23571 (df_update_entry_block_defs): Likewise.
23572 (df_update_exit_block_uses): Likewise.
23573 (df_entry_block_bitmap_verify): Likewise.
23574 (df_exit_block_bitmap_verify): Likewise.
23575 (df_scan_verify): Likewise.
23576 * lra-constraints.c (lra_constraints): Likewise.
23577 (undo_optional_reloads): Likewise.
23578 (lra_undo_inheritance): Likewise.
23579 * lra-remat.c (calculate_gen_cands): Likewise.
23580 (do_remat): Likewise.
23581 * lra-spills.c (assign_spill_hard_regs): Likewise.
23582 (spill_pseudos): Likewise.
23583 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23584 (bitmap_set_subtract_values): Likewise.
23586 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23588 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23589 management with auto_bitmap.
23590 (fix_inter_tick): Likewise.
23591 (fix_recovery_deps): Likewise.
23592 * ira.c (add_store_equivs): Likewise.
23593 (find_moveable_pseudos): Likewise.
23594 (split_live_ranges_for_shrink_wrap): Likewise.
23595 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23596 (rtx_reuse_manager::seen_def_p): Likewise.
23597 (rtx_reuse_manager::set_seen_def): Likewise.
23598 * print-rtl.h (class rtx_reuse_manager): Likewise.
23600 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23602 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23604 (migrate_btr_def): Likewise.
23605 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23606 * df-core.c (loop_post_order_compute): Likewise.
23607 (loop_inverted_post_order_compute): Likewise.
23608 * hsa-common.h: Likewise.
23609 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23610 * init-regs.c (initialize_uninitialized_regs): Likewise.
23611 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23612 (inline_small_functions): Likewise.
23613 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23614 * ira.c (combine_and_move_insns): Likewise.
23615 (build_insn_chain): Likewise.
23616 * loop-invariant.c (find_invariants): Likewise.
23617 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23618 * predict.c (tree_predict_by_opcode): Likewise.
23619 (predict_paths_leading_to): Likewise.
23620 (predict_paths_leading_to_edge): Likewise.
23621 (estimate_loops_at_level): Likewise.
23622 (estimate_loops): Likewise.
23623 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23624 (spread_components): Likewise.
23625 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23626 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23627 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23628 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23629 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23630 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23631 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23632 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23633 (mark_threaded_blocks): Likewise.
23634 (thread_through_all_blocks): Likewise.
23635 * tree-ssa.c (verify_ssa): Likewise.
23636 (execute_update_addresses_taken): Likewise.
23637 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23639 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23641 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23643 (post_order_compute): Likewise.
23644 (inverted_post_order_compute): Likewise.
23645 (pre_and_rev_post_order_compute_fn): Likewise.
23647 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23649 * genrecog.c (int_set::int_set): Explicitly construct our
23650 auto_vec base class.
23651 * vec.h (auto_vec::auto_vec): New constructor.
23653 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23655 * bitmap.h (class auto_bitmap): New constructor taking
23656 bitmap_obstack * argument.
23658 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23660 * bitmap.h (class auto_bitmap): Change type of m_bits to
23661 bitmap_head, and adjust ctor / dtor and member operators.
23663 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23665 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23666 when returned register mode doesn't match original mode.
23668 2017-05-12 Jeff Law <law@redhat.com>
23669 Jakub Jelinek <jakub@redhat.com>
23671 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23672 we look for cc setter after the compare-elim changes.
23673 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23674 within the vector to match what compare-elim now expects.
23675 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23676 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23678 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23679 after the compare-elim changes.
23680 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23681 the vector to match what compare-elim now expects.
23682 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23683 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23684 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23685 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23686 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23688 * config/visium/visium.c (single_set_and_flags): Fix where
23689 we look for cc setter after the compare-elim changes.
23690 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23691 with the vector to match what compare-elim now expects.
23692 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23693 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23694 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23695 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23696 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23697 (neg<mode>2_insn_set_overflow): Likewise.
23699 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23701 PR middle-end/79794
23702 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23703 maybe_expand_insn call, set ops[0].target. If still set after call,
23704 set alt_rtl. Add extra arg to recursive calls.
23705 (extract_bit_field): Add alt_rtl argument. Pass to
23707 * expmed.h (extract_bit_field): Fix prototype.
23708 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23709 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23710 to extract_bit_field_calls.
23711 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23712 Pass alt_rtl to extract_bit_field calls.
23713 * calls.c (store_unaligned_arguments_into_psuedos)
23714 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23715 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23717 * optabs.h (struct expand_operand): Add target bitfield.
23719 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23721 * compare-elim.c (try_eliminate_compare): Canonicalize
23722 operation with embedded compare to
23723 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23724 (set (reg) (operation)].
23726 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23728 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23731 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23732 cost of adding a carry flag for ADC instruction.
23733 [case MINUS]: Ignore the cost of subtracting a carry flag
23734 for SBB instruction.
23736 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23738 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23740 * config/rs6000/bmiintrin.h: New file.
23741 * config/rs6000/bmi2intrin.h: New file.
23742 * config/rs6000/x86intrin.h: New file.
23744 2017-05-12 Jeff Law <law@redhat.com>
23746 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23749 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23751 PR middle-end/80707
23752 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23754 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23756 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23757 early expansion of vector divide builtins.
23758 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23759 builtins identified as having unsigned arguments.
23761 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23763 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23764 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23765 expansion of vector logical operations (and, andc, or, xor,
23768 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23770 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23771 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23773 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23775 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23776 early GIMPLE expansion of vector multiplies.
23778 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23780 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23781 TARGET_HAVE_MOVT conditional.
23782 (movt splitter): Likewise.
23784 2017-05-12 Richard Biener <rguenther@suse.de>
23786 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23787 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23788 Fold all stmts not inplace.
23790 2017-05-12 Richard Biener <rguenther@suse.de>
23792 PR tree-optimization/80713
23793 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23794 inserted_exprs bit for not removed stmts.
23796 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23798 PR middle-end/69921
23799 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23800 parallelized" attribute for parallelized OpenACC kernels.
23801 * omp-offload.c (execute_oacc_device_lower): Use it.
23803 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23804 Set "oacc kernels" attribute.
23805 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23806 parameter. Adjust all users.
23807 (oacc_fn_attrib_kernels_p): Remove function.
23808 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23809 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23810 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23811 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23812 assert "oacc kernels" attribute is set.
23814 2017-05-11 Carl Love <cel@us.ibm.com>
23816 * config/rs6000/rs6000-c: Add support for built-in functions
23817 vector unsigned char vec_popcnt (vector signed char)
23818 vector unsigned char vec_popcnt (vector unsigned char)
23819 vector unsigned short vec_popcnt (vector signed short)
23820 vector unsigned short vec_popcnt (vector unsigned short)
23821 vector unsigned int vec_popcnt (vector signed int)
23822 vector unsigned int vec_popcnt (vector unsigned int)
23823 vector unsigned long long vec_popcnt (vector signed long long)
23824 vector unsigned long long vec_popcnt (vector unsigned long long)
23825 vector signed long long vec_slo (vector signed long long,
23826 vector signed char)
23827 vector signed long long vec_slo (vector signed long long,
23828 vector unsigned char)
23829 vector unsigned long long vec_slo (vector unsigned long long,
23830 vector signed char)
23831 vector unsigned long long vec_slo (vector unsigned long long,
23832 vector unsigned char)
23833 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23834 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23835 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23836 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23837 * doc/extend.texi: Update the built-in documentation file for the
23838 new built-in functions.
23840 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23842 * attribs.h (sorted_attr_string): Move machine independent
23843 functions for target clone support from the i386 port to common
23844 code. Rename ix86_function_versions to common_function_versions.
23845 Rename make_name to make_unique_name.
23846 (common_function_versions): Likewise.
23847 (make_unique_name): Likewise.
23848 (make_dispatcher_decl): Likewise.
23849 (is_function_default_version): Likewise.
23850 * attribs.c (attr_strcmp): Likewise.
23851 (sorted_attr_string): Likewise.
23852 (common_function_versions): Likewise.
23853 (make_unique_name): Likewise.
23854 (make_dispatcher_decl): Likewise.
23855 (is_function_default_version): Likewise.
23856 * config/i386/i386.c (attr_strcmp): Likewise.
23857 (sorted_attr_string): Likewise.
23858 (ix86_function_versions): Likewise.
23859 (make_name): Likewise.
23860 (make_dispatcher_decl): Likewise.
23861 (is_function_default_version): Likewise.
23862 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23864 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23867 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23868 Account for direct move costs for vec_construct of integer
23871 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23874 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23875 (UNSPEC_STX_ATOMIC): Ditto.
23876 (loaddi_via_sse): New insn.
23877 (storedi_via_sse): Ditto.
23878 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23879 Update corresponding peephole2 patterns.
23880 (atomic_storedi_fpu): Ditto.
23882 2017-05-11 Julia Koval <julia.koval@intel.com>
23884 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23885 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23887 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23888 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23889 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23891 2017-05-11 Nathan Sidwell <nathan@acm.org>
23893 * graphite-poly.c: Include dumpfile.h.
23895 * dumpfle.h (dump_function): Declare here ...
23896 * tree-dump.h (dump_function): ... not here.
23897 * dumpfile.c: #include tree-cfg.h.
23898 (dump_function): Move here from ...
23899 * tree-dump.c (dump_function): ... here.
23900 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23901 * graphite-poly.c: Don't include tree-dump.h.
23902 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23903 * print-tree.c: Likewise.
23904 * stor-layout.c: Likewise.
23905 * tree-nested.c: Likewise.
23907 * dumpfile.c (dump_start): Use TDF_FLAGS.
23908 (dump_enable_all): Fix TDF_KIND check thinko.
23910 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23912 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23913 array entries to represent two legal parameterizations of the
23914 overloaded __builtin_cmpb function, as represented by the
23915 P6_OV_BUILTIN_CMPB constant.
23916 (altivec_resolve_overloaded_builtin): Add special case handling
23917 for the __builtin_cmpb function, as represented by the
23918 P6_OV_BUILTIN_CMPB constant.
23919 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23920 (BU_P6_64BIT_2): New macro.
23921 (BU_P6_OVERLOAD_2): New macro
23922 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23923 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23924 (CMPB): Add overload support to represent both 32-bit and 64-bit
23925 compare-bytes function.
23926 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23927 support for TARGET_CMPB.
23928 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23929 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23930 documentation of the __builtin_cmpb overloaded built-in function.
23932 2017-05-11 Richard Biener <rguenther@suse.de>
23934 PR tree-optimization/80705
23935 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23936 bases are not vectorizable.
23938 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23940 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23941 when counting register pressure.
23943 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23945 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23946 (iv_ca_more_deps): Renamed to ...
23947 (iv_ca_compare_deps): ... this.
23948 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23950 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23952 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23954 (determine_group_iv_costs): ... here.
23955 (find_inv_vars_cb): Record inv var if it's not recorded before.
23957 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23959 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23960 (get_shiftadd_cost): Ditto.
23962 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23964 * tree-ssa-address.c: Include header file.
23965 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23967 (add_to_parts): Refactor.
23968 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23969 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23972 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23974 PR tree-optimization/53090
23975 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23977 (extract_cond_operands): Detect condition with IV on both sides
23978 and return COMP_IV_EXPR_2.
23979 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23980 (rewrite_use_compare): Simplify by removing call to function
23981 extract_cond_operands.
23983 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23985 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23986 (extract_cond_operands): Detect condition comparing against non-
23987 invariant bound and return appropriate enum value.
23988 (find_interesting_uses_cond): Update use of extract_cond_operands.
23989 Handle its return value accordingly.
23990 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23992 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23994 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23995 nonlinear iv_use computation in loop invariant sensitive way.
23997 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23999 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24000 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24002 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24004 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24005 (dump_cand): Support iv_cand.inv_exprs.
24006 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24008 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24011 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24013 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24015 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24016 as local function. Include necessary header files.
24017 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24019 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24021 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24023 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24025 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24026 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24027 RSHIFT_EXPR and BIT_NOT_EXPR.
24029 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24031 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24032 (adjust_setup_cost): New parameter supporting round up adjustment.
24033 (struct address_cost_data): Delete.
24034 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24035 (split_address_cost, ptr_difference_cost): Delete.
24036 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24037 (struct ainc_cost_data): New struct.
24038 (get_address_cost_ainc): New function.
24039 (get_address_cost, get_computation_cost): Reimplement.
24040 (determine_group_iv_cost_address): Record inv_expr for all uses of
24042 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24043 (iv_ca_has_deps): Reimplemented to ...
24044 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24046 (iv_ca_extend): Call iv_ca_more_deps.
24048 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24050 * tree-ssa-address.c (struct mem_address): Move to header file.
24051 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24052 * tree-ssa-address.h (struct mem_address): Move from C file.
24053 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24055 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24057 * tree-affine.h (aff_combination_type): New interface.
24058 (aff_combination_zero_p): Remove static.
24059 (aff_combination_const_p): New interface.
24060 (aff_combination_singleton_var_p): New interfaces.
24062 2017-05-11 Richard Biener <rguenther@suse.de>
24064 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24065 Skip unreachable blocks and destinations.
24066 (eliminate): Move stmt removal and fixup ...
24067 (fini_eliminate): ... here. Skip inserted exprs.
24068 (pass_pre::execute): Move fini_pre after fini_eliminate.
24069 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24070 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24071 PRE to get rid of dead code that has invalid SSA form and
24072 split critical edges again.
24074 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24076 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24078 2017-05-11 Richard Biener <rguenther@suse.de>
24080 * passes.c (execute_function_todo): Verify loops if they are
24081 said to be up-to-date.
24082 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24083 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24085 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24088 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24089 handle calling assemble_external ourself.
24092 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24093 modes with zero size. Enhance comment.
24095 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24097 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24098 built-ins for vec_xl and vec_xst with short and char pointer
24101 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24103 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24104 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24105 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24106 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24107 (_mm_maskz_min_round_ss): New intrinsics.
24108 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24109 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24110 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24111 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24112 (__builtin_ia32_minss_mask_round): New builtins.
24113 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24114 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24115 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24117 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24118 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24120 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24123 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24125 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24126 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24127 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24128 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24129 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24130 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24131 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24132 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24133 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24134 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24135 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24136 (__builtin_ia32_mulss_mask_round): New builtins.
24137 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24138 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24139 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24141 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24142 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24144 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24147 2017-05-10 Julia Koval <julia.koval@intel.com>
24149 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24150 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24151 (_mm256_setr_m128i): New intrinsics.
24153 2017-05-10 Julia Koval <julia.koval@intel.com>
24155 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24156 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24157 (_mm_maskz_rcp14_ss): New intrinsics.
24158 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24159 (__builtin_ia32_rcp14ss_mask): New builtins.
24160 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24162 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24164 PR tree-optimization/51513
24165 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24166 (assert_unreachable_fallthru_edge_p): Use it.
24167 (group_case_labels_stmt): Likewise.
24168 * tree-cfg.h: Prototype it.
24169 * stmt.c: Include cfghooks.h and tree-cfg.h.
24170 (emit_case_dispatch_table) <gap_label>: New local variable.
24171 Use it to fill dispatch table gaps.
24172 Test for default_label before updating probabilities.
24173 (expand_case) <default_label>: Remove unneeded initialization.
24174 Test for unreachable default case statement and remove its edge.
24175 Set default_label accordingly.
24176 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24178 2017-05-10 Carl Love <cel@us.ibm.com>
24180 * config/rs6000/rs6000-c: Add support for built-in functions
24181 vector signed char vec_neg (vector signed char)
24182 vector signed short int vec_neg (vector short int)
24183 vector signed int vec_neg (vector signed int)
24184 vector signed long long vec_neg (vector signed long long)
24185 vector float vec_neg (vector float)
24186 vector double vec_neg (vector double)
24187 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24189 * config/rs6000/altivec.h: Add define for vec_neg
24190 * doc/extend.texi: Update the built-in documentation for the
24191 new built-in functions.
24193 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24195 PR tree-optimization/77644
24196 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24198 2017-05-10 Nathan Sidwell <nathan@acm.org>
24200 * dumpfile.h (TDI_lang_all): New.
24201 (TDF_KIND): New. Renumber others
24202 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24204 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24206 (get_dump_file_name): Adjust suffix generation.
24207 (dump_enable_all): Use TDF_KIND.
24208 * doc/invoke.texi (-fdump-lang-all): Document.
24210 * dumpfile.h: Tabify.
24212 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24215 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24216 Move member access before delete.
24218 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24220 * tree-inline.c (expand_call_inline): Split block at stmt
24223 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24226 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24227 are now unused after splitting mov{sf,sd}_hardfloat.
24228 (f32_lr2): Likewise.
24229 (f32_lm): Likewise.
24230 (f32_lm2): Likewise.
24231 (f32_li): Likewise.
24232 (f32_li2): Likewise.
24233 (f32_lv): Likewise.
24234 (f32_sr): Likewise.
24235 (f32_sr2): Likewise.
24236 (f32_sm): Likewise.
24237 (f32_sm2): Likewise.
24238 (f32_si): Likewise.
24239 (f32_si2): Likewise.
24240 (f32_sv): Likewise.
24241 (f32_dm): Likewise.
24242 (f32_vsx): Likewise.
24243 (f32_av): Likewise.
24244 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24245 For movsf, order stores so the VSX stores occur before the GPR
24246 store which encourages the register allocator to use a traditional
24247 FPR instead of a GPR. For movsd, order the stores so that the GPR
24248 store comes before the VSX stores to allow the power6 to work.
24249 This is due to the power6 not having a 32-bit integer store
24250 instruction from a FPR.
24251 (movsf_hardfloat): Likewise.
24252 (movsd_hardfloat): Likewise.
24254 2017-05-09 Martin Sebor <msebor@redhat.com>
24256 PR translation/80280
24257 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24260 PR translation/80280
24261 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24262 data member added in r247778.
24263 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24265 2017-05-09 Nathan Sidwell <nathan@acm.org>
24267 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24269 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24272 2017-05-09 Marek Polacek <polacek@redhat.com>
24274 * doc/invoke.texi: Fix typo.
24276 2017-05-09 Richard Biener <rguenther@suse.de>
24278 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24279 (vrp_val_is_min): Likewise.
24280 (set_value_range_to_value): Likewise.
24281 (set_value_range_to_nonnegative): Likewise.
24282 (gimple_assign_nonzero_p): Likewise.
24283 (gimple_stmt_nonzero_p): Likewise.
24284 (vrp_int_const_binop): Likewise. Remove unreachable case.
24285 (adjust_range_with_scev): Adjust comments.
24286 (compare_range_with_value): Likewise.
24287 (extract_range_from_phi_node): Likewise.
24288 (test_for_singularity): Likewise.
24290 2017-05-09 Richard Biener <rguenther@suse.de>
24292 * tree-vrp.c (get_single_symbol): Add assert that we don't
24293 get overflowed constants as invariant part.
24294 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24295 checks. Use wi::cmp instead of recursing for integer constants.
24296 (compare_values): Just ignore whether we assumed undefined
24297 overflow instead of failing the compare.
24298 (extract_range_for_var_from_comparison_expr): Add comment before the
24299 TREE_NO_WARNING sets.
24300 (test_for_singularity): Likewise.
24301 (extract_range_from_comparison): Do not disable optimization
24302 when we assumed undefined overflow.
24303 (extract_range_basic): Remove init of unused var.
24305 2017-05-09 Richard Biener <rguenther@suse.de>
24307 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24308 (extract_range_from_multiplicative_op_1): Adjust.
24309 (extract_range_from_binary_expr_1): Use int_const_binop.
24311 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24314 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24315 rs6000_store_data_bypass_p in seven define_bypass directives and
24316 in several comments.
24317 * config/rs6000/rs6000-protos.h: Add prototype for
24318 rs6000_store_data_bypass_p function.
24319 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24320 function implements slightly different (rs6000-specific) semantics
24321 than store_data_bypass_p, returning false rather than aborting
24322 with assertion error when arguments do not satisfy the
24323 requirements of store data bypass.
24324 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24325 rs6000_store_data_bypass_p.
24327 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24329 * config/xtensa/xtensa-protos.h
24330 (xtensa_initial_elimination_offset): New declaration.
24331 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24332 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24333 macro definition, add case for FRAME_POINTER_REGNUM when
24334 FRAME_GROWS_DOWNWARD.
24335 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24336 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24337 xtensa_initial_elimination_offset.
24339 2017-05-08 Nathan Sidwell <nathan@acm.org>
24341 * doc/invoke.texi: Alphabetize -fdump options.
24343 2017-05-08 Martin Sebor <msebor@redhat.com>
24345 PR translation/80280
24346 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24348 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24350 * target.def (compute_frame_layout): New optional target hook.
24351 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24352 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24353 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24355 * reload1.c (verify_initial_elim_offsets): Likewise.
24356 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24357 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24358 (arm_get_frame_offsets): Split up into this ...
24359 (arm_compute_frame_layout): ... and this function.
24361 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24363 * config/aarch64/constraints.md (Usa): New constraint.
24364 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24366 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24368 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24369 with_multilib_list after it has been checked.
24371 2017-05-08 Richard Biener <rguenther@suse.de>
24373 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24374 (bitmap_set_subtract_values): Likewise.
24376 2017-05-08 Richard Biener <rguenther@suse.de>
24378 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24379 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24381 (gimple_stmt_nonzero_warnv_p): Rename to ...
24382 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24384 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24385 (extract_range_basic): Adjust, do not disable propagation on
24386 strict overflow sensitive simplification.
24387 (vrp_visit_cond_stmt): Likewise.
24389 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24391 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24392 body size unconditionally.
24394 2017-05-07 Jeff Law <law@redhat.com>
24397 2017-05-06 Jeff Law <law@redhat.com>
24398 PR tree-optimization/78496
24399 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24402 PR tree-optimization/78496
24403 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24404 (simplify_stmt_using_ranges): Call it.
24405 (vrp_dom_walker::before_dom_children): Extract equivalences
24406 from an ASSERT_EXPR with an equality comparison against a
24409 2017-05-06 Jeff Law <law@redhat.com>
24411 PR tree-optimization/78496
24412 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24415 PR tree-optimization/78496
24416 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24417 (simplify_stmt_using_ranges): Call it.
24418 (vrp_dom_walker::before_dom_children): Extract equivalences
24419 from an ASSERT_EXPR with an equality comparison against a
24422 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24424 * lra-constraints.c (lra_copy_reg_equiv): New function.
24425 (split_reg): Use it to copy equivalence information from the
24426 original register to the spill register.
24428 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24430 PR rtl-optimization/75964
24431 * simplify-rtx.c (simplify_const_relational_operation): Remove
24432 invalid handling of comparisons of integer ABS.
24434 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24436 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24437 initialize to zero.
24438 (init_regs): Remove declaration.
24439 (function_arg_advance_32): Initialize error_p as boolean variable.
24441 2017-05-05 Nathan Sidwell <nathan@acm.org>
24443 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24444 lines. Use for (;;).
24446 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24448 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24449 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24450 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24451 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24452 VF=2 that require versioning.
24454 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24456 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24459 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24461 * diagnostic.h (diagnostic_override_option_index): Convert from
24462 macro to inline function.
24464 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24466 * diagnostic.c (last_module_changed_p): New function.
24467 (set_last_module): New function.
24468 (diagnostic_report_current_module): Convert macro usage to
24469 the above functions.
24470 * diagnostic.h (diagnostic_context::last_module): Strengthen
24471 from const line_map * to const line_map_ordinary *.
24472 (diagnostic_last_module_changed): Delete macro.
24473 (diagnostic_set_last_module): Delete macro.
24475 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24477 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24478 with diagnostic_report_diagnostic.
24479 (diagnostic_n_impl_richloc): Likewise.
24480 * diagnostic.h (report_diagnostic): Delete macro.
24481 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24482 with diagnostic_report_diagnostic.
24483 * substring-locations.c (format_warning_va): Likewise.
24485 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24487 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24488 save/restor of format_spec. Move option-printing code to...
24489 (print_option_information): ...this new function, and
24490 reimplement by simply printing to the pretty_printer,
24491 rather than appending to the format string.
24493 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24495 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24496 handling logic into...
24497 (update_effective_level_from_pragmas): ...this new function.
24499 2017-05-04 Andrew Waterman <andrew@sifive.com>
24501 * config/riscv/riscv.opt (mstrict-align): New option.
24502 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24503 (SLOW_UNALIGNED_ACCESS): Define.
24504 (riscv_slow_unaligned_access): Declare.
24505 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24507 (riscv_slow_unaligned_access): New variable.
24508 (rocket_tune_info): Set slow_unaligned_access to true.
24509 (optimize_size_tune_info): Set slow_unaligned_access to false.
24510 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24511 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24512 (riscv_option_override): Set riscv_slow_unaligned_access.
24513 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24515 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24517 * config/riscv/riscv.md: Unify indentation.
24519 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24524 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24526 (extendsi<mode>2): Add support for doing sign extension via
24527 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24528 don't have ISA 3.0 instructions.
24529 (extendsi<mode>2 splitter): Likewise.
24530 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24531 generate the normal insns since SImode can now go in vector
24532 registers. Disallow the special UNSPECs needed for previous
24533 machines to hide SImode being used. Add new insns
24534 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24535 (fix_trunc<mode>si2_stfiwx): Likewise.
24536 (fix_trunc<mode>si2_internal): Likewise.
24537 (fixuns_trunc<mode>si2): Likewise.
24538 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24539 (fctiw<u>z_<mode>_smallint): Likewise.
24540 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24541 of floating point to 32-bit integer from doing a direct move to
24542 the GPR registers to do a store.
24543 (fctiwz_<mode>): Break long line.
24545 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24547 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24548 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24549 (addr_list, addr_offset_valid_p): New.
24550 (split_address_groups): Check offset validity with above function.
24551 (gt-tree-ssa-loop-ivopts.h): Include header file.
24553 2017-05-05 Nathan Sidwell <nathan@acm.org>
24555 * config.gcc (arm*-*-*): Add missing 'fi'.
24557 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24559 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24560 included in -fopt-info does not matter.
24561 * doc/optinfo.texi (-fopt-info): Fix description of default
24562 behavour. Explicitly say order of options included in -fopt-info
24565 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24567 * config.gcc: Allow combinations of aprofile and rmprofile values for
24568 --with-multilib-list.
24569 * config/arm/t-multilib: New file.
24570 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24571 variables. Remove setting of ISA and floating-point ABI in
24572 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24573 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24574 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24575 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24577 * config/arm/t-rmprofile: Likewise except for the matches changes.
24578 * doc/install.texi (--with-multilib-list): Document the combination of
24579 aprofile and rmprofile values and warn about pitfalls in doing that.
24581 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24583 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24584 (movdi_aarch64): Likewise.
24586 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24588 PR tree-optimization/80632
24589 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24591 (build_arrays): Initialize it for virtual phis.
24592 (fix_phi_nodes): Use it for virtual phis.
24594 PR tree-optimization/80558
24595 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24596 [x, y] op z into [x op, y op z] for op & or | if conditions
24599 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24600 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24603 * config/arm/arm.md (use_literal_pool): Remove.
24604 (64-bit immediate split): No longer takes cost into consideration
24605 if arm_disable_literal_pool is enabled.
24606 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24607 used when arm_disable_literal_pool is enabled.
24608 (arm_max_const_double_inline_cost): Remove use of
24609 arm_disable_literal_pool.
24610 (push_minipool_fix): Add assert.
24611 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24612 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24613 (no_literal_pool_sf_immediate): New.
24615 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24617 PR tree-optimization/80613
24618 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24619 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24621 2017-05-05 Richard Biener <rguenther@suse.de>
24623 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24625 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24627 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24628 of this flag from insn conditions due to removal from r247495.
24630 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24632 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24634 (arm_early_store_addr_dep_ptr): Likewise.
24635 * config/arm/aarch-common-protos.h
24636 (arm_early_load_addr_dep_ptr): Add prototype.
24637 (arm_early_store_addr_dep_ptr): Likewise.
24638 * config/arm/cortex-a53.md: Add new bypasses.
24640 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24642 * tree.c (next_type_uid): Change type to unsigned.
24643 (type_hash_canon): Decrement back next_type_uid if
24644 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24645 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24648 2017-05-04 Martin Sebor <msebor@redhat.com>
24650 * builtins.c: Fix a trivial typo in a comment.
24652 PR middle-end/79234
24653 * builtins.c (check_sizes): Adjust to handle reading past the end.
24654 Avoid printing excessive upper bound of ranges. Use %E to print
24655 tree nodes instead of converting them to %wu.
24656 (expand_builtin_memchr): New function.
24657 (compute_dest_size): Rename...
24658 (compute_objsize): ...to this.
24659 (expand_builtin_memcpy): Adjust.
24660 (expand_builtin_mempcpy): Adjust.
24661 (expand_builtin_strcat): Adjust.
24662 (expand_builtin_strcpy): Adjust.
24663 (check_strncat_sizes): Adjust.
24664 (expand_builtin_strncat): Adjust.
24665 (expand_builtin_strncpy): Adjust and simplify.
24666 (expand_builtin_memset): Adjust.
24667 (expand_builtin_bzero): Adjust.
24668 (expand_builtin_memcmp): Adjust.
24669 (expand_builtin): Handle memcmp.
24670 (maybe_emit_chk_warning): Check strncat just once.
24672 2017-05-04 Martin Sebor <msebor@redhat.com>
24674 PR preprocessor/79214
24675 PR middle-end/79222
24676 PR middle-end/79223
24677 * builtins.c (check_sizes): Add inlining context and issue
24678 warnings even when -Wno-system-headers is set.
24679 (check_strncat_sizes): Same.
24680 (expand_builtin_strncat): Same.
24681 (expand_builtin_memmove): New function.
24682 (expand_builtin_stpncpy): Same.
24683 (expand_builtin): Handle memmove and stpncpy.
24685 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24687 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24688 which is not used any more.
24690 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24692 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24694 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24696 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24697 (cortexa53_tunings): Likewise.
24698 (cortexa57_tunings): Likewise.
24699 (cortexa72_tunings): Likewise.
24700 (cortexa73_tunings): Likewise.
24702 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24704 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24705 Set loop alignment to 8.
24707 2017-05-04 Martin Sebor <msebor@redhat.com>
24709 PR translation/80280
24710 * builtins.c (expand_builtin_object_size): Add missing quoting to
24711 %D and like directives.
24712 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24713 (hsa_type_for_tree_type): Same.
24714 (verify_function_arguments): Same.
24715 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24716 * varasm.c (get_section): Same.
24719 2017-05-04 Martin Sebor <msebor@redhat.com>
24721 PR translation/80280
24722 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24724 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24726 * config/aarch64/aarch64.c (generic_addrcost_table):
24727 Change HI/TI mode setting.
24729 2017-05-04 Martin Jambor <mjambor@suse.cz>
24731 PR tree-optimization/80622
24732 * tree-sra.c (comes_initialized_p): New function.
24733 (build_accesses_from_assign): Only set write lazily when
24734 comes_initialized_p is false.
24735 (analyze_access_subtree): Use comes_initialized_p.
24736 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24737 instead of testing for PARM_DECL.
24739 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24741 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24742 constraint on operand 0 to allow more general addressing modes.
24743 Adjust output template.
24744 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24746 * config/aarch64/aarch64-protos.h
24747 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24748 * config/aarch64/constraints.md (Dp): New address constraint.
24749 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24752 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24754 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24755 update use of estimate_ipcp_clone_size_and_time.
24756 (estimate_local_effects): Update use of
24757 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24758 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24759 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24760 Return nonspecialized time.
24762 2017-05-04 Richard Biener <rguenther@suse.de>
24764 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24765 for the last VUSE which def dominates the PHI. Directly call
24767 (get_continuation_for_phi_1): Remove.
24769 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24771 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24772 to explain the use of truncating division. Cap the number of
24773 iterations to the maximum given by nb_iterations_upper_bound,
24776 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24778 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24779 * configure: Regenerate.
24780 * config.in: Regenerate.
24781 * config/i386/driver-mingw32.c: new file.
24782 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24783 * config.host: Link driver-mingw32.o on MinGW host.
24784 * doc/install.texi: Document new --enable-mingw-wildcard configure
24787 2017-05-04 Marek Polacek <polacek@redhat.com>
24789 PR tree-optimization/80612
24790 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24792 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24793 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24795 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24796 (movt splitter): Likewise.
24797 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24798 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24799 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24800 block for Thumb-1 with MOVT.
24801 (thumb2_legitimate_address_p): Move code block ...
24802 (can_avoid_literal_pool_for_label_p): ... into this new function.
24803 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24805 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24806 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24807 "M-profile targets with the MOVT instruction".
24809 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24811 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24812 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24813 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24815 2017-05-04 Martin Liska <mliska@suse.cz>
24817 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24818 variable cond_code.
24820 2017-05-04 Richard Biener <rguenther@suse.de>
24822 * tree.c (array_at_struct_end_p): Handle arrays at struct
24823 end with flexarrays more conservatively. Refactor and treat
24824 arrays of arrays or aggregates more strict. Fix
24825 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24826 * tree.c (array_at_struct_end_p): Adjust prototype.
24827 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24828 * gimple-fold.c (get_range_strlen): Likewise.
24829 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24831 2017-05-04 Richard Biener <rguenther@suse.de>
24833 PR tree-optimization/31130
24834 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24836 (supports_overflow_infinity): Likewise.
24837 (is_negative_overflow_infinity): Likewise.
24838 (is_positive_overflow_infinity): Likewise.
24839 (is_overflow_infinity): Likewise.
24840 (stmt_overflow_infinity): Likewise.
24841 (overflow_infinity_range_p): Likewise.
24842 (usable_range_p): Remove as always returning true.
24843 (make_overflow_infinity): Remove.
24844 (negative_overflow_infinity): Likewise.
24845 (positive_overflow_infinity): Likewise.
24846 (avoid_overflow_infinity): Likewise.
24847 (set_value_range): Adjust accordingly.
24848 (set_value_range_to_nonnegative): Likewise, remove now unused
24849 overflow_infinity arg.
24850 (vrp_operand_equal_p): Adjust.
24851 (update_value_range): Likewise.
24852 (range_int_cst_singleton_p): Likewise.
24853 (operand_less_p): Likewise.
24854 (compare_values_warnv): Likewise.
24855 (extract_range_for_var_from_comparison_expr): Likewise.
24856 (vrp_int_const_binop): Likewise.
24857 (zero_nonzero_bits_from_vr): Likewise.
24858 (extract_range_from_multiplicative_op_1): Likewise.
24859 (extract_range_from_binary_expr_1): Likewise.
24860 (extract_range_from_unary_expr): Likewise.
24861 (extract_range_from_comparison): Likewise.
24862 (extract_range_basic): Likewise.
24863 (adjust_range_with_scev): Likewise.
24864 (compare_ranges): Likewise.
24865 (compare_range_with_value): Likewise.
24866 (dump_value_range): Likewise.
24867 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24869 (simplify_cond_using_ranges): Adjust.
24871 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24873 * brig-builtins.def: Added a builtin for class_f64.
24874 * builtin-types.def: Added a builtin type needed by class_f64.
24876 2017-05-03 Jason Merrill <jason@redhat.com>
24878 * timevar.def: Add TV_CONSTEXPR.
24880 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24882 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24884 2017-05-03 Martin Jambor <mjambor@suse.cz>
24886 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24887 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24888 * ipa-cp.c (ipcp_propagate_stage): Do not call
24889 ipa_update_after_lto_read.
24890 * ipa-inline.c (ipa_inline): Likewise.
24892 2017-05-03 Martin Jambor <mjambor@suse.cz>
24894 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24895 tag. Added a default constructor and a destructor.
24896 (ipa_edge_args_sum_t): New class;
24897 (ipa_edge_args_sum): Declare.
24898 (ipa_edge_args_vector): Remove declaration.
24899 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24900 (ipa_free_edge_args_substructures): Remove declaration.
24901 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24902 (ipa_edge_args_info_available_for_edge_p): Likewise.
24903 * ipa-prop.c (ipa_edge_args_vector): Removed.
24904 (edge_removal_hook_holder): Likewise.
24905 (edge_duplication_hook_holder): Likewise.
24906 (ipa_edge_args_sum): New variable.
24907 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24908 ipa_edge_args_vector.
24909 (ipa_free_edge_args_substructures): Likewise.
24910 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24911 ipa_edge_args_vector.
24912 (ipa_edge_removal_hook): Turned into method
24913 ipa_edge_args_sum_t::remove.
24914 (ipa_edge_duplication_hook): Turned into method
24915 ipa_edge_args_sum_t::duplicate.
24916 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24917 registering edge hooks.
24918 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24919 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24920 ipa_edge_args_sum instead of ipa_edge_args_vector.
24921 * ipa-profile.c (ipa_profile): Likewise.
24923 2017-05-03 Martin Jambor <mjambor@suse.cz>
24925 * symbol-summary.h (function_summary): New method exists.
24926 (function_summary::symtab_removal): Deallocate through release.
24927 (call_summary): New class.
24928 (gt_ggc_mx): New overload.
24929 (gt_pch_nx): Likewise.
24930 (gt_pch_nx): Likewise.
24932 2017-05-03 Jeff Law <law@redhat.com>
24934 PR tree-optimization/78496
24935 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24936 from simplify_cond_using_ranges. Split off code to walk
24937 backwards through casts into ...
24938 (simplify_cond_using_ranges_2): New function.
24939 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24940 (execute_vrp): After identifying jump threads, call
24941 simplify_cond_using_ranges_2.
24943 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24946 * ipa-inline.h (inline_summary): Add ctor.
24947 (create_ggc): Do not use ggc_cleared_alloc.
24949 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24950 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24952 * gcc.c (handle_braces): Support escaping in switch matching
24954 * doc/invoke.texi (Spec Files): Document it.
24955 Remove superfluous @code markup in items.
24957 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24959 * diagnostic-show-locus.c (struct column_range): New struct.
24960 (get_affected_columns): New function.
24961 (get_printed_columns): New function.
24962 (struct correction): New struct.
24963 (correction::ensure_capacity): New function.
24964 (correction::ensure_terminated): New function.
24965 (struct line_corrections): New struct.
24966 (line_corrections::~line_corrections): New dtor.
24967 (line_corrections::add_hint): New function.
24968 (layout::print_trailing_fixits): Reimplement in terms of the new
24970 (selftest::test_overlapped_fixit_printing): New function.
24971 (selftest::diagnostic_show_locus_c_tests): Call it.
24973 2017-05-03 Nathan Sidwell <nathan@acm.org>
24975 Canonicalize canonical type hashing
24976 * tree.h (type_hash_canon_hash): Declare.
24977 * tree.c (type_hash_list, attribute_hash_list): Move into
24978 type_hash_canon_hash.
24979 (build_type_attribute_qual_variant): Break out hash code calc into
24980 type_hash_canon_hash.
24981 (type_hash_canon_hash): New. Generic type hash computation.
24982 (build_range_type_1, build_array_type_1, build_function_type,
24983 build_method_type_directly, build_offset_type, build_complex_type,
24984 make_vector_type): Call it.
24986 2017-05-03 Richard Biener <rguenther@suse.de>
24988 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24989 When all DRs have unknown misaligned do not always peel
24990 when there is a store but apply the same costing model as if
24991 there were only loads.
24993 2017-05-03 Richard Biener <rguenther@suse.de>
24996 PR tree-optimization/80492
24997 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24998 compare_base_decls returning dont-know properly.
25000 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25002 * config/arm/iterators.md (CCSI): New mode iterator.
25003 (arch): New mode attribute.
25004 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25005 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25006 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25007 code iterator for success result mode.
25008 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25009 the corresponding new insn generators.
25011 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25014 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25015 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25017 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25019 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25020 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25021 (DDR_A): Wrap DDR argument in brackets.
25022 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25023 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25024 (DDR_REVERSED_P): Likewise.
25026 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25028 PR tree-optimization/79472
25029 * tree-switch-conversion.c (struct switch_conv_info): Add
25030 contiguous_range and default_case_nonstandard fields.
25031 (collect_switch_conv_info): Compute contiguous_range and
25032 default_case_nonstandard fields, don't clear final_bb if
25033 contiguous_range and only the default case doesn't have the required
25035 (check_all_empty_except_final): Set default_case_nonstandard instead
25036 of failing if contiguous_range and the default case doesn't have empty
25038 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25039 and only the default case doesn't have the required constants. Skip
25041 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25042 if default_case_nonstandard.
25043 (build_constructors): Build constant 1 just once. Assert that default
25044 values aren't inserted in between cases if contiguous_range. Skip
25046 (build_arrays): Skip virtual phis.
25047 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25048 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25049 Handle virtual phis.
25050 (gen_inbound_check): Handle default_case_nonstandard case.
25051 (process_switch): Adjust check_final_bb caller. Call
25052 gather_default_values with the first non-default case instead of
25053 default case if default_case_nonstandard.
25055 2017-05-02 Nathan Sidwell <nathan@acm.org>
25057 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25058 check. Fix formatting.
25060 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25062 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25063 errors when comparing specialized and unspecialized times.
25065 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25067 * diagnostic-show-locus.c
25068 (layout::should_print_annotation_line_p): Make private.
25069 (layout::print_annotation_line): Make private.
25070 (layout::annotation_line_showed_range_p): Make private.
25071 (layout::show_ruler): Make private.
25072 (layout::print_source_line): Make private. Pass in line and
25073 line_width, rather than calling location_get_source_line. Drop
25075 (layout::print_leading_fixits): New method.
25076 (layout::print_any_fixits): Rename to...
25077 (layout::print_trailing_fixits): ...this, and make private.
25078 Don't print newline fixits.
25079 (diagnostic_show_locus): Move logic for printing one row into...
25080 (layout::print_line): ...this new function. Move the
25081 location_get_source_line call and error-handling from
25082 print_source_line to here. Call print_leading_fixits, and rename
25083 print_any_fixits to print_trailing_fixits.
25084 (selftest::test_fixit_insert_containing_newline): Update now that
25085 newlines are partially supported.
25086 (selftest::test_fixit_insert_containing_newline_2): New test.
25087 (selftest::test_fixit_replace_containing_newline): Update comments.
25088 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25089 * edit-context.c (class added_line): New class.
25090 (class edited_line): Describe newline handling in comment.
25091 (edited_line::actually_edited_p): New method.
25092 (edited_line::print_content): Delete redundant decl.
25093 (edited_line::m_predecessors): New field.
25094 (edited_file::print_content): Call edited_line::print_content.
25095 (edited_file::print_diff): Update to support newlines.
25096 (edited_file::print_diff_hunk): Likewise.
25097 (edited_file::print_run_of_changed_lines): New function.
25098 (edited_file::print_diff_line): Convert to...
25099 (print_diff_line): ...this.
25100 (edited_file::get_effective_line_count): New function.
25101 (edited_line::edited_line): Initialize new field m_predecessors.
25102 (edited_line::~edited_line): Clean up m_predecessors.
25103 (edited_line::apply_fixit): Handle newlines.
25104 (edited_line::get_effective_line_count): New function.
25105 (edited_line::print_content): New function.
25106 (edited_line::print_diff_lines): New function.
25107 (selftest::test_applying_fixits_insert_containing_newline): New
25109 (selftest::test_applying_fixits_replace_containing_newline): New
25111 (selftest::insert_line): New function.
25112 (selftest::test_applying_fixits_multiple_lines): Add example of
25114 (selftest::edit_context_c_tests): Call the new tests.
25116 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25118 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25119 parameter cand. Update dump information.
25120 (get_computation_cost): Update uses.
25122 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25124 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25125 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25126 (get_computation_at, rewrite_use_address): Update use of
25127 get_computation_aff.
25129 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25131 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25132 (get_computation): Delete.
25133 (get_computation_cost): Implement like get_computation_cost_at.
25134 Use get_computation_at.
25135 (get_computation_cost_at): Delete.
25136 (rewrite_use_nonlinear_expr): Use get_computation_at.
25137 (rewrite_use_compare, remove_unused_ivs): Ditto.
25139 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25141 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25143 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25145 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25146 (ivopts_global_cost_for_size): Rename parameter and update uses.
25147 (iv_ca_recount_cost): Update uses.
25148 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25149 candidates seperately in n_invs and n_cands.
25150 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25152 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25154 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25155 (find_inv_vars_cb): New.
25156 (find_depends): Renamed to ...
25157 (find_inv_vars): ... this.
25158 (add_candidate_1, force_var_cost): Call find_inv_vars.
25159 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25161 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25163 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25164 inv_vars. Add inv_exprs.
25165 (struct iv_cand): Rename depends_on to inv_vars.
25166 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25167 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25168 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25169 (dump_cand): Dump inv_vars.
25170 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25171 (record_invariant, find_depends, add_candidate_1): Ditto.
25172 (set_group_iv_cost, force_var_cost): Ditto.
25173 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25174 (get_computation_cost_at, get_computation_cost): Ditto.
25175 (determine_group_iv_cost_generic): Ditto.
25176 (determine_group_iv_cost_address): Ditto.
25177 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25178 (determine_group_iv_costs): Ditto.
25179 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25180 (iv_ca_set_remove_invariants): Renamed to ...
25181 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25182 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25183 (iv_ca_set_add_invariants): Renamed to ...
25184 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25185 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25186 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25187 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25188 (create_new_ivs): Remove useless dump.
25190 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25192 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25194 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25195 (iv_ca_set_no_cp, create_new_iv): Ditto.
25197 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25199 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25201 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25203 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25204 function tree_check2.
25206 2017-05-02 Martin Liska <mliska@suse.cz>
25208 * doc/gcov.texi: Add missing preposition.
25209 * gcov.c (function_info::function_info): Properly fill up
25210 all member variables.
25212 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25214 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25216 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25218 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25220 2017-05-02 Martin Liska <mliska@suse.cz>
25223 * lto-streamer-in.c (lto_read_tree_1): Remove
25224 LTO_STREAMER_DEBUG.
25225 * lto-streamer.c (struct tree_hash_entry): Likewise.
25226 (struct tree_entry_hasher): Likewise.
25227 (tree_entry_hasher::hash): Likewise.
25228 (tree_entry_hasher::equal): Likewise.
25229 (lto_streamer_init): Likewise.
25230 (lto_orig_address_map): Likewise.
25231 (lto_orig_address_get): Likewise.
25232 (lto_orig_address_remove): Likewise.
25233 * lto-streamer.h: Likewise.
25234 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25235 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25237 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25239 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25240 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25241 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25242 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25243 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25244 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25245 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25246 (mm_maskz_sub_ss): New intrinsics.
25247 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25248 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25249 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25250 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25251 (__builtin_ia32_subss_mask_round): New builtins.
25252 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25253 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25254 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25256 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25257 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25259 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25262 2017-05-02 Martin Jambor <mjambor@suse.cz>
25264 PR tree-optimization/78687
25265 * tree-sra.c (access): New field parent.
25266 (process_subtree_disqualification): New function.
25267 (disqualify_candidate): Call it.
25268 (build_accesses_from_assign): Reset write flag if creating an
25270 (build_access_subtree): Fill in parent field and also prpagate
25271 down grp_write flag.
25272 (create_artificial_child_access): New parameter set_grp_write, set
25273 grp_write to its value.
25274 (propagate_subaccesses_across_link): Also propagate grp_write flag
25276 (propagate_all_subaccesses): Push the closest parent back to work
25277 queue if add_access_to_work_queue returned true.
25279 2017-05-02 Richard Biener <rguenther@suse.de>
25281 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25282 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25283 -fstrict-overflow documentation.
25284 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25285 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25286 flag_strict_overflow.
25287 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25288 * lto-opts.c (lto_write_options): Do not stream it.
25289 * lto-wrapper.c (merge_and_complain): Do not handle it.
25290 * opts.c (default_options_table): Do not set -fstrict-overflow.
25291 (finish_options): Likewise do not clear it when sanitizing.
25292 * simplify-rtx.c (simplify_const_relational_operation): Do not
25293 test flag_strict_overflow.
25295 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25297 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25298 using enabled attribute.
25299 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25300 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25301 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25302 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25303 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25304 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25305 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25306 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25307 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25308 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25310 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25312 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25314 2017-05-02 Richard Biener <rguenther@suse.de>
25316 PR tree-optimization/80591
25318 2017-04-10 Richard Biener <rguenther@suse.de>
25320 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25323 2017-05-02 Richard Biener <rguenther@suse.de>
25325 PR tree-optimization/80549
25326 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25327 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25328 headers if they do not have a preheader.
25330 2017-05-02 Martin Liska <mliska@suse.cz>
25333 * common.opt: Fix typo.
25334 * doc/invoke.texi: Likewise.
25336 2017-05-01 Jan Beulich <jbeulich@suse.com>
25338 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25339 swapping, add (x,x,m,x,n) alternative.
25341 2017-05-01 Nathan Sidwell <nathan@acm.org>
25343 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25344 unnecessary unadjusted_alignment check.
25346 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25349 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25351 * gimplify.c (gimplify_cilk_detach): New function.
25352 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25353 * tree-core.h: Document EXPR_CILK_SPAWN.
25354 * tree.h (EXPR_CILK_SPAWN): Define.
25356 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25358 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25359 to use new fixit_hint representation, using the "replace" logic.
25360 (get_line_span_for_fixit_hint): Likewise.
25361 (layout::print_any_fixits): Likewise.
25362 (selftest::test_one_liner_many_fixits): Rename to...
25363 (selftest::test_one_liner_many_fixits_1): ...this, and update
25364 comment and expected output to reflect that the multiple fix-it
25365 hints are now consolidated into one insertion.
25366 (selftest::test_one_liner_many_fixits_2): New test.
25367 (selftest::test_diagnostic_show_locus_one_liner): Update for
25369 (selftest::test_fixit_consolidation): Update for fix-it API
25371 * diagnostic.c (print_parseable_fixits): Likewise.
25372 * edit-context.c (edited_line::m_line_events): Convert from
25373 auto_vec <line_event *> to auto_vec <line_event>.
25374 (class line_event): Convert from abstract base class to a concrete
25375 class, taking over the role of replace_event.
25376 (class insert_event): Delete.
25377 (class replace_event): Rename to class line_event. Convert to
25379 (edit_context::add_fixits): Reimplement.
25380 (edit_context::apply_insert): Delete.
25381 (edit_context::apply_replace): Rename to...
25382 (edit_context::apply_fixit): ...this. Convert to half-open range.
25383 (edited_file::apply_insert): Delete.
25384 (edited_file::apply_replace): Rename to...
25385 (edited_file::apply_fixit): ...this.
25386 (edited_line::~edited_line): Drop deletion of events.
25387 (edited_line::apply_insert): Delete.
25388 (edited_line::apply_replace): Rename to...
25389 (edited_line::apply_fixit): ...this. Convert to half-open range.
25390 Update for change to type of m_line_events.
25391 * edit-context.h (edit_context::apply_insert): Delete.
25392 (edit_context::apply_replace): Rename to...
25393 (edit_context::apply_fixit): ...this.
25395 2017-05-01 Martin Sebor <msebor@redhat.com>
25397 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25400 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25403 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25404 __get_cpuid_max returns 0.
25405 (__get_cpuid_count): Ditto.
25407 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25409 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25410 replacement expression is another instance of one of its arguments.
25412 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25415 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25416 check for stack push/pop autoinc.
25417 * config/i386/i386.c (ix86_agi_dependent): Return false
25418 if the only reason why modified_in_p returned true is that
25419 addr is SP based and set_insn is a push or pop.
25421 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25423 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25426 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25429 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25430 (account_size_time): Use two predicates - exec_pred and
25432 (evaluate_conditions_for_known_args): Compute both clause and
25434 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25435 (inline_summary_t::duplicate): Update.
25436 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25438 (compute_inline_parameters): Likewise.
25439 (estimate_edge_size_and_time): Update caluclation of time.
25440 (estimate_node_size_and_time): Compute both time and nonspecialized
25442 (estimate_ipcp_clone_size_and_time): Update.
25443 (inline_merge_summary): Update.
25444 (do_estimate_edge_time): Update.
25445 (do_estimate_edge_size): Update.
25446 (do_estimate_edge_hints): Update.
25447 (inline_read_section, inline_write_summary): Stream both new predicates.
25448 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25450 (compute_inlined_call_time): Cleanup.
25451 (big_speedup_p): Update.
25452 (edge_badness): Update.
25453 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25454 (size_time_entry): Replace predicate by exec_predicate and
25455 nonconst_predicate.
25456 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25457 (estimate_edge_time): Return also nonspec_time.
25458 (reset_edge_growth_cache): Update.
25460 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25462 PR rtl-optimization/80491
25463 * ifcvt.c (noce_process_if_block): When looking for x setter
25464 with missing else_bb, don't check only the insn right before
25465 cond_earliest, but look for the last insn that x is modified in
25466 within the same bb.
25468 PR rtl-optimization/80491
25469 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25471 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25473 PR tree-optimization/80487
25474 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25476 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25478 PR tree-optimization/79697
25479 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25480 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25481 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25483 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25484 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25486 2017-04-28 Martin Sebor <msebor@redhat.com>
25488 PR tree-optimization/80523
25489 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25490 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25492 (maybe_warn, format_directive, parse_directive): Use new functions.
25493 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25495 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25497 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25499 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25501 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25502 target_header_dir): Set correctly.
25503 * configure: Regenerated.
25504 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25505 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25506 instead of SYSTEM_HEADER_DIR.
25508 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25510 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25511 (estimate_local_effects): Likewise.
25512 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25513 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25514 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25515 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25516 do_estimate_edge_time, estimate_edge_time): Likewise.
25517 * ipa-inline-analysis.c (estimate_node_size_and_time,
25518 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25519 (estimate_time_after_inlining): Remove.
25521 2017-04-28 Martin Liska <mliska@suse.cz>
25523 * doc/gcov.texi: Enhance documentation of gcov.
25525 2017-04-28 Martin Liska <mliska@suse.cz>
25527 * doc/gcov.texi: Sort options in alphabetic order.
25528 * doc/gcov-dump.texi: Likewise.
25529 * doc/gcov-tool.texi: Likewise.
25530 * gcov.c (print_usage): Likewise.
25531 * gcov-dump.c (print_usage): Likewise.
25532 * gcov-tool.c (print_merge_usage_message): Likewise.
25533 (print_rewrite_usage_message): Likewise.
25534 (print_overlap_usage_message): Likewise.
25536 2017-04-28 Martin Liska <mliska@suse.cz>
25538 PR gcov-profile/53915
25539 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25541 2017-04-28 Martin Liska <mliska@suse.cz>
25543 PR gcov-profile/79891
25544 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25545 is marked by compiler as living on a line.
25546 (get_cycles_count): Remove usage of the union.
25547 (output_intermediate_file): Likewise.
25548 (find_source): Fix GNU coding style.
25549 (accumulate_line_counts): Remove old non-all block mode.
25550 (output_lines): Remove usage of the union.
25551 * profile.c (output_location): Include all BBs, even if
25552 belonging to a same line (and file) as a previous BB.
25554 2017-04-28 Martin Liska <mliska@suse.cz>
25556 * gcov.c (process_args): Handle new argument 'w'.
25557 (read_graph_file): Assign ID to BBs.
25558 (output_branch_count): Display BB # if verbose flag is set.
25559 (output_lines): Likewise for arcs.
25560 (print_usage): Add '--verbose' option help.
25561 * doc/gcov.texi: Document --verbose (-w) option.
25563 2017-04-28 Martin Liska <mliska@suse.cz>
25565 * gcov.c (struct block_location_info): New struct.
25566 (process_file): Fill up the new structure.
25567 (read_graph_file): Replace usage of encoding by the newly added
25569 (add_line_counts): Likewise.
25570 (accumulate_line_counts): Remove usage of the union.
25571 (function_info::function_info): New function.
25572 (function_info::~function_info): Likewise.
25573 (process_file): Call delete instead of release_function.
25574 (release_function): Release the function.
25575 (release_structures): Call delete instead of release_function.
25576 (solve_flow_graph): Replace usage of num_blocks.
25577 (find_exception_blocks): Likewise.
25578 (output_lines): Fix GNU coding style.
25580 2017-04-28 Martin Liska <mliska@suse.cz>
25583 * coverage.c (coverage_remove_note_file): New function.
25584 * coverage.h: Declare the function.
25585 * toplev.c (finalize): Clean if an error has been seen.
25587 2017-04-28 Martin Liska <mliska@suse.cz>
25589 PR gcov-profile/80031
25590 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25591 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25592 * gcov.c (read_graph_file): Read just number of blocks.
25593 * profile.c (branch_prob): Do not stream 0 flags per a basic
25596 2017-04-28 Martin Liska <mliska@suse.cz>
25598 * gcov-dump.c (tag_*): Add new argument to declarations.
25599 (dump_gcov_file): Likewise.
25600 (tag_blocks): Add and use new argument depth.
25601 (tag_arcs): Likewise.
25602 (tag_lines): Likewise.
25603 (tag_counters): Likewise.
25604 (tag_summary): Likewise.
25605 (dump_working_sets): Use depth to do a proper indentation.
25607 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25610 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25611 * symtab.c (symtab_node::debug_symtab): Move definition here.
25613 2017-04-28 Richard Biener <rguenther@suse.de>
25615 * lto-streamer.h (LTO_major_version): Bump to 7.
25617 2017-04-28 Richard Biener <rguenther@suse.de>
25619 * tree-vrp.c (assert_info): New struct.
25620 (add_assert_info): New helper.
25621 (register_edge_assert_for_2): Refactor to add asserts to a vector
25623 (register_edge_assert_for_1): Likewise.
25624 (register_edge_assert_for): Likewise.
25625 (finish_register_edge_assert_for): New helper actually registering
25626 asserts where live on edge.
25627 (find_conditional_asserts): Adjust.
25628 (find_switch_asserts): Likewise.
25629 (evrp_dom_walker::try_find_new_range): Generalize.
25630 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25632 2017-04-27 Marek Polacek <polacek@redhat.com>
25635 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25636 arg10 and arg11 to itype.
25638 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25640 * doc/extend.texi (Object Size Checking): Improve grammar.
25642 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25645 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25646 that the logic for permitting reciprocal estimates matches that
25649 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25652 * tree.c (type_cache_hasher::equal): Only compare
25653 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25654 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25655 non-aggregate element types.
25656 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25657 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25659 2017-04-27 Richard Biener <rguenther@suse.de>
25661 PR middle-end/80533
25662 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25663 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25664 keeping a reference to a trailing array.
25666 2017-04-27 Richard Biener <rguenther@suse.de>
25668 PR middle-end/80539
25669 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25670 being in loop-closed SSA form conservatively.
25671 (chrec_fold_multiply_poly_poly): Likewise.
25673 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25675 PR middle-end/79665
25676 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25677 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25679 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25682 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25683 (aarch64_function_arg_alignment): Return unsigned int again, but still
25684 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25685 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25686 Don't emit -Wpsabi note.
25687 (aarch64_function_arg_boundary): Likewise.
25688 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25691 2017-04-26 Nathan Sidwell <nathan@acm.org>
25693 * tree.h (crc32_unsigned_n): Declare.
25694 (crc32_unsigned, crc32_unsigned): Make inline.
25695 * tree.c (crc32_unsigned_bits): Replace with ...
25696 (crc32_unsigned_n): ... this.
25697 (crc32_unsigned, crc32_byte): Remove.
25698 (crc32_string): Remove unnecessary braces.
25700 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25702 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25703 * ipa-inline-analysis.c (MAX_TIME): Remove.
25704 (account_size_time): Use sreal for time.
25705 (dump_inline_summary): Update.
25706 (estimate_function_body_sizes): Update.
25707 (estimate_edge_size_and_time): Update.
25708 (estimate_calls_size_and_time): Update.
25709 (estimate_node_size_and_time): Update.
25710 (inline_merge_summary): Update.
25711 (inline_update_overall_summary): Update.
25712 (estimate_time_after_inlining): Update.
25713 (inline_read_section): Update.
25714 (inline_write_summary): Update.
25715 * ipa-inline.c (compute_uninlined_call_time): Update.
25716 (compute_inlined_call_time): Update.
25717 (recursive_inlining): Update.
25718 (inline_small_functions): Update.
25719 (dump_overall_stats): Update.
25720 * ipa-inline.h: Include sreal.h.
25721 (size_time_entry): Turn time to sreal.
25722 (inline_summary): Turn self_time nad time to sreal.
25724 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25726 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25728 (sreal::stream_out, sreal::stream_in): New.
25729 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25731 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25733 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25736 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25739 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25741 (dimode_scalar_chain::compute_convert_gain): Ditto.
25742 (dimode_scalar_chain::make_vector_copies): Ditto.
25743 (dimode_scalar_chain::convert_reg): Ditto.
25744 (dimode_scalar_chain::convert_insn): Ditto.
25745 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25746 (VI248_AVX512BW_1): New mode iterator.
25747 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25748 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25751 2017-04-25 Martin Sebor <msebor@redhat.com>
25753 PR tree-optimization/80497
25754 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25755 constants are representable in HOST_WIDE_INT.
25756 (parse_directive): Ditto.
25758 2017-04-25 Martin Sebor <msebor@redhat.com>
25761 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25762 (new_zero_array): Adjust signature.
25763 (dom_info::dom_init): Used unsigned rather that size_t.
25764 (dom_info::dom_info): Same.
25766 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25767 Jakub Jelinek <jakub@redhat.com>
25770 * config/arm/arm.c: Include gimple.h.
25771 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25772 returns negative, increment ncrn only if it returned positive.
25773 (arm_needs_doubleword_align): Return int instead of bool,
25774 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25775 members, but if there is any such non-FIELD_DECL
25776 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25777 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25778 returns negative, increment nregs only if it returned positive.
25779 (arm_setup_incoming_varargs): Likewise.
25780 (arm_function_arg_boundary): Emit -Wpsabi note if
25781 arm_needs_doubleword_align returns negative, return
25782 DOUBLEWORD_ALIGNMENT only if it returned positive.
25784 2017-04-25 Marek Polacek <polacek@redhat.com>
25787 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25788 first argument to type.
25790 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25793 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25794 type checks to test for compatibility instead of equality.
25796 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25797 Jakub Jelinek <jakub@redhat.com>
25800 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25802 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25803 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25804 the alignment computation, but return their maximum in warn_alignment.
25805 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25806 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25808 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25809 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25812 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25814 * config/arc/simdext.md (dmpyh): Fix typo.
25816 2017-04-25 Richard Biener <rguenther@suse.de>
25818 PR tree-optimization/80492
25819 * alias.c (compare_base_decls): Handle registers with asm
25820 specification conservatively.
25821 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25822 compare_base_decls returning dont-know properly.
25824 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25826 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25827 (legitimate_offset_address_p): New function.
25828 (arc_legitimate_address_p): Use above function.
25830 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25832 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25834 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25836 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25837 ACCH registers whenever they are available.
25839 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25841 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25842 double regs fix when not used.
25844 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25846 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25848 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25849 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25851 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25853 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25854 when emitting short ADD instructions.
25856 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25858 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25860 (cmpsi_cc_c_insn): Likewise.
25861 (cbranchsi4_scratch): Compute proper instruction length using
25862 compact_hreg_operand.
25863 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25865 2017-04-25 Richard Biener <rguenther@suse.de>
25867 PR middle-end/80509
25868 * passes.c (pass_manager::pass_manager): Initialize
25869 m_name_to_pass_map.
25871 2017-04-25 Richard Biener <rguenther@suse.de>
25873 PR tree-optimization/79201
25874 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25876 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25879 * config/s390/vector.md: Split MEM->GPR vector moves for
25880 non-s_operand addresses.
25882 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25885 * config/s390/predicates.md (reload_const_wide_int_operand): New
25887 * config/s390/s390.md ("movti"): Remove d/P alternative.
25888 ("movti_bigconst"): New pattern definition.
25890 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25893 * s390-protos.h (s390_expand_cs_hqi): Removed.
25894 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25895 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25896 modes as well as CCZ1mode and CCZmode.
25897 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25898 signature of s390_emit_compare_and_swap.
25899 (s390_expand_cs_hqi): Likewise, make static.
25900 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25901 compare-and-swap, in some cases.
25902 (s390_expand_cs): Wrapper function.
25903 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25905 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25906 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25907 patterns for small and large integers. Forbid symref memory operands.
25908 Move expander to s390.c. Require cc register.
25909 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25910 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25911 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25912 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25913 symref memory operands. Remove CC mode and call s390_match_ccmode
25915 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25917 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25919 * config/s390/s390.md (define_peephole2): New peephole to help
25920 combining the load-and-test pattern with volatile memory.
25922 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25924 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25925 with CCZmode for TARGET_Z196.
25927 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25929 PR rtl-optimization/80501
25930 * combine.c (make_compound_operation_int): Set subreg_code to SET
25931 even for AND with mask of the sign bit of mode.
25933 PR rtl-optimization/80500
25934 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25935 sum's initial value.
25937 2017-04-25 Julian Brown <julian@codesourcery.com>
25938 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25940 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25942 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25944 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25946 2017-04-25 Julian Brown <julian@codesourcery.com>
25947 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25949 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25950 (thunderx2t99_sha): New Reservation.
25952 2017-04-25 Julian Brown <julian@codesourcery.com>
25953 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25955 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25956 type for 1-element load.
25958 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25960 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25962 2017-04-24 Martin Jambor <mjambor@suse.cz>
25964 PR tree-optimization/80293
25965 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25966 char arrays not totally scalarizable if it is false.
25967 (analyze_all_variable_accesses): Pass correct value in the new
25968 parameter. Add a statistics counter.
25970 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25972 PR middle-end/79931
25973 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25975 2017-04-24 Richard Biener <rguenther@suse.de>
25977 PR tree-optimization/80494
25978 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25979 out for complex types.
25981 2017-04-24 Richard Biener <rguenther@suse.de>
25983 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25984 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25985 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25986 (DFS): Adjust and never fail.
25987 (sccvn_dom_walker::fail): Remove.
25988 (sccvn_dom_walker::before_dom_children): Adjust.
25989 (run_scc_vn): Likewise and never fail.
25990 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25991 (pass_fre::execute): Likewise.
25993 2017-04-24 Richard Biener <rguenther@suse.de>
25995 PR tree-optimization/79725
25996 * tree-ssa-sink.c (statement_sink_location): Return whether
25997 failure reason was zero uses. Move that check later.
25998 (sink_code_in_bb): Deal with zero uses by removing the stmt
26001 2017-04-24 Richard Biener <rguenther@suse.de>
26004 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26005 pointer-based references.
26007 2017-04-24 Richard Biener <rguenther@suse.de>
26010 * pass_manager.h (pass_manager::operator new): Remove.
26011 (pass_manager::operator delete): Likewise.
26012 * passes.c (pass_manager::operator new): Remove.
26013 (pass_manager::operator delete): Likewise.
26014 (pass_manager::pass_manager): Zero individual pass members.
26016 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26019 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26020 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26021 Check "XEXP (src, 1)" operand here.
26022 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26023 Check "XEXP (src, 1)" operand here.
26024 (dimode_scalar_chain::make_vector_copies): Detect count register
26025 of a shift instruction. Zero extend count register from QImode
26026 to DImode to satisfy vector shift pattern count operand predicate.
26027 Substitute vector shift count operand with a DImode copy.
26028 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26031 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26033 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26034 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26035 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26036 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26037 (UNSPEC_NOREX_MEM): Remove definition.
26039 2017-04-21 Richard Biener <rguenther@suse.de>
26041 PR tree-optimization/79547
26042 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26043 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26044 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26045 without any constraints.
26047 2017-04-21 Richard Biener <rguenther@suse.de>
26049 PR tree-optimization/78847
26050 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26052 2017-04-21 Richard Biener <rguenther@suse.de>
26054 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26055 (build_distinct_type_copy): Likewise.
26056 (build_variant_type_copy): Likewise.
26057 * tree.c (build_qualified_type): Pass down mem-stat info.
26058 (build_distinct_type_copy): Likewise.
26059 (build_variant_type_copy): Likewise.
26061 2017-04-21 Richard Biener <rguenther@suse.de>
26063 PR tree-optimization/80237
26064 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26066 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26067 for a simplified result.
26069 2016-04-21 Richard Biener <rguenther@suse.de>
26071 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26072 sth as strict as a simple_iv but a chrec without symbols and an
26073 operand defined in the loop we are peeling (and not some subloop).
26074 (propagate_constants_for_unrolling): Propagate all constants.
26076 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26079 * config/i386/i386.c (print_reg): Remove assert for disalowed
26080 regno values, call output_operand_lossage instead.
26082 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26085 * config/i386/constraints.md (Yc): New register constraint.
26086 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26087 Use Yc constraint for alternative 2 of operand 0. Remove
26088 preferred_for_speed attribute.
26090 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26092 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26093 lastprivate clauses in SIMT case.
26095 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26097 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26099 2017-04-20 Richard Biener <rguenther@suse.de>
26101 PR tree-optimization/57796
26102 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26103 as N scalar stores.
26104 (vect_model_load_cost): Cost gathers as N scalar loads.
26106 2017-04-20 Richard Biener <rguenther@suse.de>
26108 * ggc-page.c (ggc_allocated_p): Rename to ...
26109 (safe_lookup_page_table_entry): ... this and return the lookup
26111 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26113 2017-04-20 Richard Biener <rguenther@suse.de>
26115 PR tree-optimization/80453
26116 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26117 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26118 from the conditions.
26119 (vn_phi_eq): Pass them down.
26120 (vn_phi_lookup): Record them.
26121 (vn_phi_insert): Likewise.
26123 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26125 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26126 uninitialized variable warning to avoid buffer overrun.
26128 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26131 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26132 is suppressed for '{ 0 }' in C.
26134 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26136 * BASE-VER: Set to 8.0.0.
26138 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26140 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26141 priority .init_array and .fini_array section with SECTION_NOTYPE
26144 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26146 PR middle-end/80423
26147 * tree.h (build_array_type): Add typeless_storage default argument.
26148 * tree.c (type_cache_hasher::equal): Also compare
26149 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26150 (build_array_type): Add typeless_storage argument, set
26151 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26153 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26154 (build_array_type): Likewise. Add typeless_storage argument.
26156 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26157 Jakub Jelinek <jakub@redhat.com>
26159 PR tree-optimization/80426
26160 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26161 operation on symbolic operands, also compute the overflow for the
26162 invariant part when the operation degenerates into a negation.
26164 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26167 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26168 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26171 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26173 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26176 * config/avr/avr.c (tree.h): Include it.
26177 (cgraph.h): Include it.
26178 (avr_encode_section_info): Don't warn for uninitialized progmem
26179 variable if it's just an alias.
26181 2017-04-19 Richard Biener <rguenther@suse.de>
26184 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26185 when needed by AutoPGO.
26187 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26190 * doc/lto.texi: Remove an extra 'that'.
26192 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26194 PR rtl-optimization/80429
26195 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26196 are only used in debug insns.
26198 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26199 Vladimir Makarov <vmakarov@redhat.com>
26201 * config/sparc/predicates.md (input_operand): Add comment. Return
26202 true for any memory operand when LRA is in progress.
26203 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26205 2017-04-18 Jeff Law <law@redhat.com>
26208 * mips.md ({return,simple_return}_internal): Do not overwrite
26211 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26213 PR tree-optimization/80443
26214 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26215 instead of adding 1, subtract -1 and similarly instead of subtracting
26218 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26220 PR rtl-optimization/80357
26221 * haifa-sched.c (tmp_bitmap): New variable.
26222 (model_recompute): Handle duplicate use records.
26223 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26224 (free_global_sched_pressure_data): Free it.
26226 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26229 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26230 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26231 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26232 instead of SYSTEM_HEADER_DIR.
26234 2017-04-18 Jeff Law <law@redhat.com>
26236 PR middle-end/80422
26237 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26238 predecessors after walking up the insn chain.
26240 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26243 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26244 sizetype type into debug info.
26246 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26249 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26250 unneeded test for TARGET_UPPER_REGS_SF.
26251 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26253 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26256 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26257 instead of gsi_after_labels.
26259 2017-04-18 Jeff Law <law@redhat.com>
26261 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26265 2017-04-13 Jeff Law <law@redhat.com>
26266 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26267 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26269 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26272 * config/avr/avr.c (intl.h): Include it.
26273 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26275 2017-04-18 Martin Liska <mliska@suse.cz>
26277 PR gcov-profile/78783
26278 * gcov-tool.c (gcov_output_files): Validate that destination
26279 file is either removed by the tool or by a user.
26281 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26282 Guy Benyei <guybe@mellanox.com>
26284 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26285 block, and do not negate it, the stored id is already negative.
26287 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26289 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26291 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26294 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26295 masks of options that should be turned off if the VSX vector
26296 options are turned off.
26297 (OTHER_P8_VECTOR_MASKS): Likewise.
26298 (OTHER_VSX_VECTOR_MASKS): Likewise.
26299 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26300 rs6000_disable_incompatible_switches to validate no type switches
26302 (rs6000_incompatible_switch): New function to disallow turning on
26303 other vector options if -mno-vsx, -mno-power8-vector, or
26304 -mno-power9-vector are specified.
26306 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26308 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26310 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26312 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26313 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26314 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26315 (ARG_POINTER_CFA_OFFSET): Likewise.
26317 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26319 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26320 conditions to take advantage of various optimizations.
26322 2017-04-13 Jeff Law <law@redhat.com>
26324 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26325 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26326 (zero_extendsidi2_dext): Likewise.
26328 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26331 * fold-const.c (fold_ternary_loc): Revert
26332 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26335 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26337 PR rtl-optimization/80343
26338 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26339 new scratch pseudo.
26341 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26344 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26345 to ubsan_encode_value.
26347 2017-04-13 Jeff Law <law@redhat.com>
26349 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26350 appearing in DEBUG_INSNs.
26352 2017-04-13 Martin Liska <mliska@suse.cz>
26354 PR gcov-profile/80413
26355 * gcov-io.c (gcov_write_string): Copy to buffer just when
26356 allocated size is greater than zero.
26358 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26361 * dwarf2out.c (decls_for_scope): Ignore declarations of
26362 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26364 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26367 * ipa-visibility.c (non_local_p): Fix typos.
26368 (localize_node): When localizing symbol in same comdat group,
26369 dissolve the group only when we know external symbols are going
26371 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26373 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26375 PR tree-optimization/79390
26376 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26377 order does not result in usable sequence, retry with reversed operand
26383 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26384 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26385 op0 instead of fold_convert_loc (loc, type, arg0).
26387 2017-04-12 Jeff Law <law@redhat.com>
26389 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26390 has a delay slot in the generated code.
26392 * config/cris/cris.md (cris_preferred_reload_class): Return
26393 GENNONACR_REGS rather than GENERAL_REGS.
26395 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26398 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26399 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26400 signedness of the result type.
26402 2017-04-12 Richard Biener <rguenther@suse.de>
26403 Jeff Law <law@redhat.com>
26405 PR tree-optimization/80359
26406 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26407 trim stores to TARGET_MEM_REFs.
26409 2017-04-12 Richard Biener <rguenther@suse.de>
26411 PR tree-optimization/79390
26412 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26413 threading case even more.
26415 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26418 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26419 for quad_address_p for TImode, instead of just not indexed_address.
26421 2017-04-12 Richard Biener <rguenther@suse.de>
26422 Bernd Edlinger <bernd.edlinger@hotmail.de>
26424 PR middle-end/79671
26425 * alias.c (component_uses_parent_alias_set_from): Handle
26426 TYPE_TYPELESS_STORAGE.
26427 (get_alias_set): Likewise.
26428 * tree-core.h (tree_type_common): Add typeless_storage flag.
26429 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26430 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26431 for types containing members with TYPE_TYPELESS_STORAGE.
26432 (place_field): Likewise.
26433 (layout_type): Likewise for ARRAY_TYPE.
26434 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26435 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26436 TYPE_TYPELESS_STORAGE.
26437 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26439 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26442 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26443 first argument to type.
26445 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26449 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26450 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26451 (rs6000_expand_binop_builtin): Likewise.
26452 (rs6000_expand_ternop_builtin): Likewise; also add missing
26453 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26454 vshasigma built-ins.
26455 * doc/extend.texi: Document that vec_xxpermdi's third argument
26456 must be a constant.
26458 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26460 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26461 Use shift_const cost parameter when calculating gain of STV shifts.
26463 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26465 PR rtl-optimization/70478
26466 * lra-constraints.c (process_alt_operands): Check memory for
26467 disfavoring memory insn operand.
26469 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26471 PR middle-end/80100
26472 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26473 left shift in unsigned HOST_WIDE_INT type.
26475 PR rtl-optimization/80385
26476 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26477 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26480 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26481 if they have any depend clauses.
26483 2017-04-11 Martin Liska <mliska@suse.cz>
26486 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26487 * ipa-split.c (split_function): Create a local comdat symbol
26488 if caller is in a comdat group.
26490 2017-04-11 Martin Liska <mliska@suse.cz>
26493 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26496 2017-04-11 Martin Sebor <msebor@redhat.com>
26498 PR middle-end/80364
26499 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26500 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26502 (directive::set_width, directive::set_precision, format_character):
26504 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26507 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26510 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26511 conflict, set target->arch_name instead of target->cpu_name.
26513 2017-04-11 Richard Biener <rguenther@suse.de>
26515 PR tree-optimization/80374
26516 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26517 build_zero_cst, remove fold_convertible_p check again.
26519 2017-04-11 Martin Liska <mliska@suse.cz>
26522 * ubsan.c (instrument_object_size): Do not instrument register
26525 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26528 * config/i386/i386-builtin-types.def
26529 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26530 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26531 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26532 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26533 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26534 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26535 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26536 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26537 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26538 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26539 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26540 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26541 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26542 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26543 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26544 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26545 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26546 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26547 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26548 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26549 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26550 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26551 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26552 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26553 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26554 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26555 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26556 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26557 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26558 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26559 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26560 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26561 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26562 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26563 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26564 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26565 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26566 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26567 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26568 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26569 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26570 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26571 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26572 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26573 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26574 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26576 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26577 flag to second_arg_count, handle 4 argument function type _COUNT
26578 aliases, handle second_arg_count on second argument rather than last.
26580 2017-04-10 Jeff Law <law@redhat.com>
26582 PR tree-optimization/80374
26583 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26584 record anything if we can not convert integer_zero_node to the
26587 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26590 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26591 Enhance special handling given to the TARGET_P9_MINMAX option in
26592 relation to certain other options.
26594 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26596 PR tree-optimization/80153
26597 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26598 remove POINTER_PLUS_EXPR's base part directly, rather than through
26601 2017-04-10 Richard Biener <rguenther@suse.de>
26602 Bin Cheng <bin.cheng@arm.com>
26604 PR tree-optimization/80153
26605 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26606 the first element of pointer type aff_tree. Build result expr in
26608 (add_elt_to_tree): Convert to type unconditionally. Remove other
26609 fold_convert calls.
26610 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26611 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26613 2017-04-10 Richard Biener <rguenther@suse.de>
26615 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26618 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26620 PR rtl-optimization/70478
26621 * lra-constraints.c (curr_small_class_check): New.
26622 (update_and_check_small_class_inputs): New.
26623 (process_alt_operands): Update curr_small_class_check. Disfavor
26624 alternative insn memory operands. Check available regs for small
26627 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26630 * config/mips/mips.opt (-mvirt): Update description.
26631 * doc/invoke.texi (-mvirt): Likewise.
26633 2017-04-10 Richard Biener <rguenther@suse.de>
26635 PR middle-end/80362
26636 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26637 looking for NEGATE_EXPR in -A / -B to A / B folding.
26639 2017-04-10 Martin Liska <mliska@suse.cz>
26641 PR gcov-profile/80224
26642 * gcov.c (print_usage): Fix usage string.
26643 (get_gcov_intermediate_filename): Remove.
26644 (output_gcov_file): Use both for normal and intermediate format.
26645 (generate_results): Do not initialize special file for
26646 intermediate format.
26648 2017-04-10 Richard Biener <rguenther@suse.de>
26650 PR tree-optimization/80304
26651 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26654 2017-04-10 Nathan Sidwell <nathan@acm.org>
26657 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26658 (rs6000_init_builtins): Use it.
26660 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26662 * config/arm/arm.md (<mrc>): Add mode to SET source.
26663 (<mrrc>): Likewise.
26665 2017-04-10 Richard Biener <rguenther@suse.de>
26667 PR middle-end/80344
26668 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26670 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26673 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26674 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26675 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26676 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26677 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26678 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26679 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26680 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26681 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26682 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26683 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26684 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26685 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26686 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26687 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26688 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26689 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26690 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26691 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26692 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26693 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26694 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26695 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26697 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26699 PR rtl-optimization/70478
26700 * lra-constraints.c: Reverse the last patch.
26702 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26704 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26705 Add comment for WCHAR_T.
26707 2017-04-08 Martin Liska <mliska@suse.cz>
26710 2017-04-07 Martin Liska <mliska@suse.cz>
26713 * ipa-split.c (split_function): Add function part to a same comdat
26716 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26719 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26721 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26723 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26724 (vec_pairing): ...to this.
26725 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26726 (rs6000_sched_init): Adjust for name change.
26727 (struct rs6000_sched_context): Likewise.
26728 (rs6000_init_sched_context): Likewise.
26729 (rs6000_set_sched_context): Likewise.
26731 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26737 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26739 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26740 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26741 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26743 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26745 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26747 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26749 PR rtl-optimization/70703
26750 * ira-color.c (update_conflict_hard_regno_costs): Use
26751 int64_t instead of HOST_WIDE_INT.
26753 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26755 PR rtl-optimization/70478
26756 * lra-constraints.c (process_alt_operands): Disfavor alternative
26757 insn memory operands.
26759 2017-04-07 Jeff Law <law@redhat.com>
26761 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26762 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26764 2017-04-07 Martin Liska <mliska@suse.cz>
26767 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26768 Show error message instead of an ICE.
26770 2017-04-07 Martin Liska <mliska@suse.cz>
26773 * ipa-split.c (split_function): Add function part to a same comdat
26776 2017-04-07 Richard Biener <rguenther@suse.de>
26778 PR middle-end/80341
26779 * tree.c (get_unwidened): Also handle ! for_type case for
26781 * convert.c (do_narrow): Split out from ...
26782 (convert_to_integer_1): ... here. Do not pass final truncation
26783 type to get_unwidened for TRUNC_DIV_EXPR.
26785 2017-04-07 Richard Biener <rguenther@suse.de>
26787 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26789 (aff_combination_const): Adjust.
26790 (aff_combination_scale): Likewise.
26791 (aff_combination_add_elt): Likewise.
26792 (aff_combination_add_cst): Likewise.
26793 (aff_combination_convert): Likewise.
26794 (add_elt_to_tree): Likewise. Remove unused argument.
26795 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26797 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26799 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26801 * config/arm/arm.c (arm_default_short_enums): Use
26802 ARM_DEFAULT_SHORT_ENUMS.
26803 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26805 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26808 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26809 members with redundant out-of-class redeclaration.
26811 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26814 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26815 * config/i386/i386.md (*zero_extendsidi2):
26816 Add (?*x,*x) and (?*v,*v) alternatives.
26818 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26821 * config/i386/i386.c (ix86_expand_builtin)
26822 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26823 mode from insn data. Convert operands to insn operand mode.
26824 Copy operands that don't satisfy insn predicate to a register.
26826 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26828 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26831 2017-04-06 Richard Biener <rguenther@suse.de>
26833 PR tree-optimization/80334
26834 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26835 preserve alignment of accesses.
26837 2017-04-06 Richard Biener <rguenther@suse.de>
26839 PR tree-optimization/80262
26840 * tree-sra.c (build_ref_for_offset): Preserve address-space
26842 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26843 Drop useless address-space information on MEM_REF offsets.
26845 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26847 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26849 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26851 PR rtl-optimization/70703
26852 * ira-color.c (update_conflict_hard_regno_costs): Use
26853 HOST_WIDE_INT instead of long.
26855 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26858 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26859 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26861 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26862 for x86_64 target. Handle -m3dnowa option.
26864 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26866 PR rtl-optimization/70703
26867 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26868 (update_conflict_hard_regno_costs): Use long instead of unsigned
26869 arithmetic for cost calculation.
26871 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26872 Bernd Edlinger <bernd.edlinger@hotmail.de>
26875 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26878 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26881 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26882 ptr_mode with Pmode throughout.
26883 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26884 into probe_stack_range and use DImode.
26886 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26889 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26890 call_eh_return is true.
26892 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26894 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26895 Initialize last_match_fntype_index.
26897 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26900 * tree-nvr.c: Include internal-fn.h.
26901 (pass_return_slot::execute): Ignore internal calls without
26904 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26905 Richard Biener <rguenther@suse.de>
26908 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26909 captures used multiple times, except for the last use.
26910 * generic-match-head.c: Include gimplify.h.
26912 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26914 PR tree-optimization/79390
26915 * target.h (struct noce_if_info): Declare.
26916 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26917 * target.def (noce_conversion_profitable_p): New target hook.
26918 * ifcvt.h (struct noce_if_info): New type, moved from ...
26919 * ifcvt.c (struct noce_if_info): ... here.
26920 (noce_conversion_profitable_p): Renamed to ...
26921 (default_noce_conversion_profitable_p): ... this. No longer
26923 (noce_try_store_flag_constants, noce_try_addcc,
26924 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26925 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26926 instead of noce_conversion_profitable_p.
26927 * config/i386/i386.c: Include ifcvt.h.
26928 (ix86_option_override_internal): Don't override
26929 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26930 (ix86_noce_conversion_profitable_p): New function.
26931 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26932 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26933 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26934 * doc/tm.texi: Regenerated.
26936 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26938 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26941 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26944 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26945 instructions for small multiply cores.
26947 2017-04-04 Jeff Law <law@redhat.com>
26949 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26951 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26952 that are not set by the loop over the elements.
26954 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26957 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26958 int mode, convert_modes it to mode as unsigned, otherwise use
26959 lowpart_subreg to mode rather than SImode.
26960 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26961 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26962 Use DImode instead of SImode for the shift count operand.
26963 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26966 2017-04-04 Richard Biener <rguenther@suse.de>
26968 PR middle-end/80281
26969 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26970 arithmetic done for the negate or the plus. Simplify.
26971 (A - (-B) -> A + B): Likewise.
26972 * fold-const.c (split_tree): Make sure to not negate pointers.
26974 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26976 PR rtl-optimization/60818
26977 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26978 a compare of comparisons with the thing compared if this results
26979 in a different machine mode.
26981 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26983 * alias.c (base_alias_check): Fix typo in comment.
26984 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26985 * cgraphunit.c (symbol_table::compile): Likewise.
26986 * collect2.c (maybe_run_lto_and_relink): Likewise.
26987 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26988 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26989 * config/avr/avr.c (avr_map_op_t): Likewise.
26990 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26991 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26992 * config/epiphany/epiphany.md (movcc): Likewise.
26993 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26994 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26996 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26997 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26998 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26999 * config/sh/sh.c (sh_rtx_costs): Likewise.
27000 * fold-const.c (fold_truth_andor): Likewise.
27001 * genautomata.c (collapse_flag): Likewise.
27002 * gengtype.h (struct type::u::s): Likewise.
27003 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27004 * input.c (FORMAT_AMOUNT): Likewise.
27005 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27006 (known_aggs_to_agg_replacement_list): Likewise.
27007 * ipa-inline-analysis.c: Likewise.
27008 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27009 * ipa-polymorphic-call.c
27010 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27011 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27012 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27014 * modulo-sched.c (apply_reg_moves): Likewise.
27015 * omp-expand.c (build_omp_regions_1): Likewise.
27016 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27017 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27018 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27019 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27020 * value-prof.c: Likewise.
27021 * var-tracking.c (val_reset): Likewise.
27023 2017-04-03 Richard Biener <rguenther@suse.de>
27025 PR tree-optimization/80275
27026 * fold-const.c (split_address_to_core_and_offset): Handle
27029 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27031 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27032 descriptors is at least equal to that of functions.
27034 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27036 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27038 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27041 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27042 (mov<IMOD4:mode>): New expander.
27043 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27045 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27047 PR rtl-optimization/79405
27048 * fwprop.c (propagations_left): New variable.
27049 (forward_propagate_into): Decrement it.
27050 (fwprop_init): Initialize it.
27051 (fw_prop): If the variable has reached zero, stop propagating.
27052 (fwprop_addr): Ditto.
27054 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27057 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27058 a FUNCTION_DECL, pass it as decl instead of origin to
27061 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27063 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27066 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27069 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27070 TARGET_VSX_SMALL_INTEGER.
27072 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27074 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27075 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27077 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27079 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27080 extraction from odd-numbered MSA register.
27082 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27084 PR middle-end/80173
27085 * expmed.c (store_bit_field_1): Don't attempt to create
27086 a word subreg out of hard registers wider than word if they
27087 have HARD_REGNO_NREGS of 1 for their mode.
27089 PR middle-end/80163
27090 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27091 conversions to integer types wider than word and pointer.
27094 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27095 (rtx_equal_for_cselib_p): Pass 0 to it.
27096 * cselib.c (cselib_hasher::equal): Likewise.
27097 (rtx_equal_for_cselib_1): Add depth argument. If depth
27098 is 128, don't look up VALUE locs and punt. Increment
27099 depth in recursive calls when walking VALUE locs.
27101 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27103 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27104 (make_gcov_file_name): Use the canonical path name for generating
27106 (read_line): Fix handling of files with ascii null bytes.
27108 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27110 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27111 to initialise a vector register instead
27112 of using a const_int.
27114 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27116 PR translation/80189
27117 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27118 diagnostic messages.
27120 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27123 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27124 (dfp_diex_<mode>): Update mode of operand 1.
27125 * doc/extend.texi (dxex, dxexq): Document change to return type.
27126 (diex, diexq): Document change to argument type.
27128 2017-03-30 Martin Jambor <mjambor@suse.cz>
27131 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27132 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27133 it reflects the signature changes performed at the callee side.
27134 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27135 to cgraph_build_function_type_skip_args.
27136 (build_function_decl_skip_args): Adjust call to the above function.
27138 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27141 * config/i386/sse.md
27142 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27143 register as dest whenever it is a MEM not rtx_equal_p to the
27144 corresponding dup operand, and when forcing into reg move the
27145 reg into the memory afterwards.
27146 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27147 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27148 for the force_reg mode.
27149 (avx512vl_vextractf128<mode>): Use register as dest either
27150 always when a MEM, or when it is a MEM not rtx_equal_p to the
27151 corresponding dup operand, or even not when it is a CONST_VECTOR
27152 depending on the mode and lo vs. hi.
27153 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27155 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27156 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27157 Likewise. Require that operands[2] is even.
27158 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27159 Remove extraneous parens. Require that operands[2] is a multiple
27161 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27162 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27163 disallow memory then.
27165 2017-03-30 Richard Biener <rguenther@suse.de>
27167 PR tree-optimization/77498
27168 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27169 to non-constants over backedges.
27171 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27173 PR rtl-optimization/80233
27174 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27175 as last_combined_insn. Do not test for BARRIER_P separately.
27177 2017-03-29 Andreas Schwab <schwab@suse.de>
27180 * calls.c (prepare_call_address): Convert funexp to Pmode before
27181 copying to temp reg.
27183 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27185 PR tree-optimization/80158
27186 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27187 Handle possible future case of more than one alternate
27189 (replace_rhs_if_not_dup): Likewise.
27190 (replace_one_candidate): Likewise.
27192 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27194 PR rtl-optimization/80193
27195 * ira.c (ira): Do not check allocation for LRA.
27197 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27199 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27200 (nvptx_output_simt_exit): Declare.
27201 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27202 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27203 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27204 (nvptx_declare_function_name): ...here. Emit declaration of local
27205 memory space buffer for omp_simt_enter insn.
27206 (nvptx_output_unisimt_switch): New.
27207 (nvptx_output_softstack_switch): New.
27208 (nvptx_output_simt_enter): New.
27209 (nvptx_output_simt_exit): New.
27210 * config/nvptx/nvptx.h (struct machine_function): New fields
27211 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27212 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27213 (UNSPECV_SIMT_EXIT): Ditto.
27214 (omp_simt_enter_insn): New insn.
27215 (omp_simt_enter): New expansion.
27216 (omp_simt_exit): New insn.
27217 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27219 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27220 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27221 (expand_GOMP_SIMT_EXIT): New.
27222 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27223 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27224 (GOMP_SIMT_EXIT): Ditto.
27225 * target-insns.def (omp_simt_enter): New insn.
27226 (omp_simt_exit): Ditto.
27227 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27229 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27230 (lower_rec_input_clauses): Likewise.
27231 (lower_lastprivate_clauses): Handle SIMT privatization.
27233 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27234 (ompdevlow_adjust_simt_enter): New.
27235 (find_simtpriv_var_op): New.
27236 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27237 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27239 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27240 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27241 (copy_decl_for_dup_finish): Ditto.
27243 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27245 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27248 * config/i386/i386.c (ix86_option_override_internal): Always
27249 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27251 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27253 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27255 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27257 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27258 mark new edge's irreducible flag accordign to it.
27259 (vect_do_peeling): Check loop preheader edge's irreducible flag
27260 and pass it to function slpeel_add_loop_guard.
27262 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27264 PR tree-optimization/80218
27265 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27266 Update block frequencies and counts.
27268 2017-03-28 Richard Biener <rguenther@suse.de>
27270 PR tree-optimization/78644
27271 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27272 of a simplification result we may not use it at all.
27274 2017-03-28 Richard Biener <rguenther@suse.de>
27277 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27278 without arguments, generate default definition of a SSA name.
27280 2017-03-28 Richard Biener <rguenther@suse.de>
27282 PR middle-end/80222
27283 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27284 TYPE_REF_CAN_ALIAS_ALL references.
27285 * fold-const.c (fold_indirect_ref_1): Likewise.
27287 2017-03-28 Martin Liska <mliska@suse.cz>
27290 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27291 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27293 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27294 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27296 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27297 (EXTRA_SPECS): Define.
27298 (SUBTARGET_EXTRA_SPECS): Likewise.
27299 (SUBTARGET_CPP_SPEC): Likewise.
27300 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27301 SUBTARGET_EXTRA_SPECS.
27302 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27304 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27306 * config/arc/simdext.md (vst64_insn): Update pattern.
27307 (vld32wh_insn): Likewise.
27308 (vld32wl_insn): Likewise.
27309 (vld64_insn): Likewise.
27310 (vld32_insn): Likewise.
27312 2017-03-28 Marek Polacek <polacek@redhat.com>
27315 * fold-const.c (fold_comparison): Use protected_set_expr_location
27316 instead of SET_EXPR_LOCATION.
27318 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27320 * tree.c (add_expr): Avoid name lookup warning.
27322 2017-03-27 Jeff Law <law@redhat.com>
27324 PR tree-optimization/80216
27325 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27326 function name. Limit recursion depth.
27327 (record_temporary_equivalences): Corresponding changes.
27329 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27331 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27334 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27337 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27339 * cfgcleanup.c (reg_note_cfa_p): New array.
27340 (insns_have_identical_cfa_notes): New function.
27341 (old_insns_match_p): Don't cross-jump in between /f
27342 and non-/f instructions. If both i1 and i2 are frame related,
27343 verify all CFA notes, their order and content.
27345 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27348 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27349 HImode and SImode with zero extend to DImode to one insn.
27350 (bswap<mode>2_extenddi): Likewise.
27351 (bswapsi2_extenddi): Likewise.
27352 (bswaphi2_extendsi): Likewise.
27353 (bswaphi2): Combine bswap HImode and SImode into one insn.
27354 Separate memory insns from swapping register.
27355 (bswapsi2): Likewise.
27356 (bswap<mode>2): Likewise.
27357 (bswaphi2_internal): Delete, no longer used.
27358 (bswapsi2_internal): Likewise.
27359 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27360 store, and gpr<-gpr swap insns.
27361 (bswap<mode>2_store): Likewise.
27362 (bswaphi2_reg): Register only splitter, combine with the splitter.
27363 (bswaphi2 splitter): Likewise.
27364 (bswapsi2_reg): Likewise.
27365 (bswapsi2 splitter): Likewise.
27366 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27367 the insns into load, store, and register/register insns.
27368 (bswapdi2_ldbrx): Likewise.
27369 (bswapdi2_load): Likewise.
27370 (bswapdi2_store): Likewise.
27371 (bswapdi2_reg): Likewise.
27373 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27375 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27376 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27378 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27381 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27383 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27384 special handling for target option conflicts between dform
27385 options (-mpower9-dform, -mpower9-dform-vector,
27386 -mpower9-dform-scalar) and -mno-direct-move.
27388 2017-03-27 Richard Biener <rguenther@suse.de>
27390 PR tree-optimization/80181
27391 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27393 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27395 * config/arc/predicates.md (move_double_src_operand): Replace the
27396 call to move_double_src_operand with a call to address_operand.
27398 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27400 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27401 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27402 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27404 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27406 * config/arc/predicates.md (long_immediate_loadstore_operand):
27407 Consider scaled addresses cases.
27409 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27411 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27412 restored when in interrupt.
27413 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27414 doesn't have delay slot.
27416 2017-03-27 Richard Biener <rguenther@suse.de>
27419 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27420 inlined thunk clones.
27422 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27425 * asan.c (instrument_derefs): Copy over last operand from
27426 original COMPONENT_REF to the new COMPONENT_REF with
27427 DECL_BIT_FIELD_REPRESENTATIVE.
27428 * ubsan.c (instrument_object_size): Likewise.
27430 2017-03-27 Richard Biener <rguenther@suse.de>
27432 PR tree-optimization/80170
27433 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27434 sure DR/SCEV didnt fold in constants we do not see when looking
27435 at the reference base alignment.
27437 2017-03-27 Richard Biener <rguenther@suse.de>
27439 PR middle-end/80171
27440 * gimple-fold.c (fold_ctor_reference): Properly guard against
27441 NULL return value from canonicalize_constructor_val.
27443 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27446 * config/i386/i386.c (ix86_expand_builtin)
27447 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27448 flags reg setting and flags reg using instructions.
27449 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27450 clobbering instructions to zero extend op2.
27452 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27454 * doc/install.texi (Configuration) <--with-aix-soname>:
27455 Update link to AIX ld.
27457 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27459 PR rtl-optimization/80160
27460 PR rtl-optimization/80159
27461 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27462 reg_alternate_class into account.
27464 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27467 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27468 to consider in curr_insn_transform.
27470 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27472 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27473 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27474 and emit_mode_inner.
27476 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27478 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27479 argument to the overloaded builtin variants. Use the new flag to
27480 deprecate certain builtin variants.
27481 * config/s390/s390-builtin-types.def: Add new builtin types.
27482 * config/s390/s390-builtins.h: Support new flags field for
27483 overloaded builtins.
27484 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27485 (s390_macro_to_expand): Enable vector float data type.
27486 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27487 builtins by incrementing the __VEC__ version number.
27488 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27490 (s390_resolve_overloaded_builtin): Emit error messages depending
27491 on the builtin flags.
27492 * config/s390/s390.c (s390_expand_builtin): Support additional
27493 flags argument. Change error message to match the messages
27494 emitted in s390-c.c.
27495 * config/s390/s390.md: New UNSPEC_* constants.
27496 (op_type): Add new instruction types.
27497 * config/s390/vecintrin.h: Add new builtins and test data class
27499 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27500 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27501 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27502 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27503 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27504 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27506 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27507 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27508 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27509 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27511 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27512 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27513 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27514 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27516 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27517 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27518 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27519 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27520 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27521 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27522 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27524 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27525 ("vec_scatter_element<V_HW_4:mode>_DI")
27526 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27527 ("vec_fpint<mode>", "vflls")
27528 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27529 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27530 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27531 ("*vec_cmphe<mode>_cc"): ... these.
27533 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27534 mode constant instead of magic value.
27536 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27538 * config/s390/s390.c (s390_expand_vec_compare): Support other
27539 vector floating point modes than just V2DF.
27540 (s390_expand_vcond): Likewise.
27541 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27542 (s390_cannot_change_mode_class): Prevent mode changes between TF
27543 and V1TF in vector registers.
27544 * config/s390/s390.md (DF, SF): New mode attributes.
27545 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27546 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27547 SFmode support for VRs.
27548 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27550 (VFT, VF_HW): New mode iterators.
27551 (vw, sdx): New mode attributes.
27552 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27553 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27554 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27555 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27556 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27557 also the new vector floating point modes. Renaming to ...
27559 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27560 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27561 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27562 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27563 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27564 ("vec_unordered<mode>"): ... these.
27566 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27567 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27568 ("*vec_extendv2df"): New insn definitions.
27570 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27572 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27573 ("mulditi3_2", "*muldi3_sign"): New patterns.
27574 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27575 rename the pattern definition.
27577 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27579 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27581 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27583 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27585 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27586 instruction if possible.
27587 * config/s390/vector.md (vec_halfnumelts): New mode
27589 ("*vec_vllezlf<mode>"): New pattern.
27591 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27593 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27594 ("popcountv4si2", "popcountv2di2"): Rename to ...
27595 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27596 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27598 ("popcount<mode>2_vxe"): New pattern.
27600 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27602 * common/config/s390/s390-common.c (processor_flags_table): Add
27604 * config.gcc: Add arch12.
27605 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27606 Default to arch12 for unknown CPU model numbers.
27607 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27608 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27609 PROCESSOR_max sanity check.
27610 * config/s390/s390-opts.h (enum processor_type): Add
27612 * config/s390/s390.c (processor_table): Add arch12.
27613 (s390_expand_builtin): Add check for B_VXE flag.
27614 (s390_issue_rate): Add PROCESSOR_ARCH12.
27615 (s390_get_sched_attrmask): Likewise.
27616 (s390_get_unit_mask): Likewise.
27617 (s390_sched_score): Enable z13 scheduling for arch12.
27618 (s390_sched_reorder): Likewise.
27619 (s390_sched_variable_issue): Likewise.
27620 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27622 (s390_tune_attr): Use z13 scheduling also for arch12.
27623 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27624 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27625 (TARGET_VXE_P): New macros.
27626 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27627 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27628 * config/s390/s390.opt: Add arch12 as processor_type.
27630 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27632 * config/s390/s390.md
27633 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27634 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27635 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27637 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27638 Rename expanders to ...
27640 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27641 ("fixuns_truncdddi2_emu"): ... these.
27643 ("fixuns_trunc<mode>si2_emu"): New expander.
27645 ("*fixuns_truncdfdi2_z13"): Rename to ...
27646 ("*fixuns_truncdfdi2_vx"): ... this.
27648 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27650 * config/s390/2964.md: Remove the single element vector compare
27651 instructions which are no longer used.
27652 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27654 (s390_canonicalize_comparison): Remove handling of DFmode
27656 (s390_expand_vec_compare_scalar): Remove function.
27657 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27658 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27660 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27662 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27664 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27666 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27667 will anyway by matched by mov<mode>_64dfp.
27669 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27671 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27672 vlef/vstef. Add missing operand to vleif.
27674 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27676 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27677 pair for all vector types with 64 bit elements.
27678 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27679 * config/s390/vector.md (V_HW_64): ... here.
27680 (V_128_NOSINGLE): New mode iterator.
27681 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27682 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27683 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27684 ("*vec_load_pairv2di"): Change to ...
27685 ("*vec_load_pair<mode>"): ... this one.
27687 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27689 * config/s390/constraints.md: Add comments.
27690 (jKK): Reject element sizes > 8 bytes.
27691 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27693 * config/s390/s390.md: Add the s_operand checks formerly in
27694 s390_split_ok_p to various splitters where they are still
27696 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27697 for 128 bit vectors. Plus two splitters.
27699 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27701 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27704 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27707 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27708 error if the boundary argument is not constant.
27710 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27712 PR rtl-optimization/80112
27713 * loop-doloop.c (doloop_condition_get): Don't check condition
27714 if cmp isn't SET with IF_THEN_ELSE src.
27716 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27718 PR tree-optimization/80158
27719 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27720 replacing a candidate statement, also replace it for the
27721 candidate's alternate interpretation.
27722 (replace_rhs_if_not_dup): Likewise.
27723 (replace_one_candidate): Likewise.
27725 2017-03-24 Richard Biener <rguenther@suse.de>
27727 PR tree-optimization/80167
27728 * graphite-isl-ast-to-gimple.c
27729 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27731 (translate_isl_ast_to_gimple::get_rename): Likewise.
27733 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27735 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27736 handling of certain combinations of target options, including the
27737 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27738 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27740 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27743 * config/arm/arm.md (*load_multiple): Add reload_completed to
27744 matching condition.
27746 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27747 Richard Biener <rguenth@suse.de>
27749 PR tree-optimization/79908
27750 PR tree-optimization/80136
27751 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27752 been cast away, gimplify_and_add suffices.
27754 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27756 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27758 2017-03-23 Richard Biener <rguenther@suse.de>
27760 PR tree-optimization/80032
27761 * gimplify.c (gimple_push_cleanup): Forced unconditional
27762 cleanups still have to go to the conditional_cleanups
27765 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27767 PR tree-optimization/80072
27768 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27770 (next_operand_entry_id): Change type to unsigned int.
27771 (sort_by_operand_rank): Make sure to return the right return value
27772 even if unsigned fields are bigger than INT_MAX.
27773 (struct oecount): Change cnt and id type to unsigned int.
27774 (oecount_hasher::equal): Formatting fix.
27775 (oecount_cmp): Make sure to return the right return value
27776 even if unsigned fields are bigger than INT_MAX.
27777 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27780 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27781 TREE_READONLY on result if writing it more than once.
27784 * doc/invoke.texi (-fsanitize=thread): Document that with
27785 -fnon-call-exceptions atomics are not able to throw
27789 * tsan.c: Include tree-eh.h.
27790 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27791 maybe_clean_or_replace_eh_stmt where needed.
27792 (instrument_memory_accesses): Add cfg_changed argument.
27793 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27794 if it returned true.
27795 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27797 PR rtl-optimization/63191
27798 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27799 wrapper function, moved the whole old content into ...
27800 (ix86_delegitimize_address_1): ... this. New inline function.
27801 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27802 true as last argument instead of ix86_delegitimize_address.
27804 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27806 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27807 cortexa57_branch_cost.
27809 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27811 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27813 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27816 * doc/md.texi (Constraints): Document wA constraint.
27817 * config/rs6000/constraints.md (wA): New.
27818 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27819 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27820 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27821 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27823 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27826 * gimplify.c (is_oacc_declared): New function.
27827 (oacc_default_clause): Use it to set default flags for acc declared
27828 variables inside parallel regions.
27829 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27830 declared variables.
27831 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27832 declare attribute to any decl as necessary.
27834 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27837 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27838 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27839 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27840 (arm_arch_lpae): This.
27841 * config/arm/arm.c (arm_arch7ve): Rename into ...
27842 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27843 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27846 2017-03-22 Martin Liska <mliska@suse.cz>
27849 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27850 error message instead of an ICE.
27852 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27854 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27856 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27858 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27860 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27863 2017-03-21 Martin Sebor <msebor@redhat.com>
27865 * doc/extend.texi: Use "cannot" instead of "can't."
27866 * doc/hostconfig.texi: Same.
27867 * doc/install.texi: Same.
27868 * doc/invoke.texi: Same.
27869 * doc/loop.texi: Same.
27870 * doc/md.texi: Same.
27871 * doc/objc.texi: Same.
27872 * doc/rtl.texi: Same.
27873 * doc/tm.texi: Same.
27874 * doc/tm.texi.in: Same.
27875 * doc/trouble.texi: Same.
27877 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27880 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27881 (collect_checksum_attributes): Set it.
27882 (die_checksum_ordered): Use it.
27884 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27886 PR tree-optimization/79908
27887 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27888 change: For a VA_ARG whose LHS has been cast away, use
27889 force_gimple_operand to construct the side effects.
27891 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27893 PR translation/80001
27894 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27895 more amenable to translation.
27896 (oacc_loop_auto_partitions): Likewise.
27898 2017-03-21 Marek Polacek <polacek@redhat.com>
27899 Martin Sebor <msebor@redhat.com>
27901 PR tree-optimization/80109
27902 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27903 on INTEGRAL_TYPE_P.
27905 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27906 Segher Boessenkool <segher@kernel.crashing.org>
27909 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27910 check reg_used_between_p between insn and one of succ or succ2
27911 depending on if succ is artificial insn not inserted into insn
27914 2017-03-21 Martin Liska <mliska@suse.cz>
27916 PR gcov-profile/80081
27917 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27918 * doc/gcc.texi: Include gcov-dump stuff.
27919 * doc/gcov-dump.texi: New file.
27921 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27923 PR rtl-optimization/79150
27924 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27925 conditional jump, if the jump is the last insn of the loop.
27927 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27928 Richard Biener <rguenth@suse.de>
27930 PR tree-optimization/79908
27931 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27932 been cast away, use force_gimple_operand to construct the side
27935 2017-03-21 Martin Liska <mliska@suse.cz>
27937 PR libfortran/79956
27938 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27941 2017-03-21 Brad Spengler <spender@grsecurity.net>
27944 * plugin.c (htab_hash_plugin): New function.
27945 (add_new_plugin): Use it and adjust.
27946 (parse_plugin_arg_opt): Adjust.
27947 (init_one_plugin): Likewise.
27949 2017-03-21 Richard Biener <rguenther@suse.de>
27951 PR tree-optimization/80032
27952 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27953 if set force the cleanup to happen unconditionally.
27954 (gimplify_target_expr): Push inserted clobbers with force_uncond
27955 to avoid them being removed by control-dependent DCE.
27957 2017-03-21 Richard Biener <rguenther@suse.de>
27959 PR tree-optimization/80122
27960 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27961 va_arg_pack_len when the inlined call stmt requires pack
27963 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27965 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27968 * tsan.c (instrument_builtin_call): If the memory model argument
27969 is not a constant, assume it is valid.
27972 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27975 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27977 PR rtl-optimization/79910
27978 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27979 if its dest is used by an insn before I2 (other than the combined
27980 insns themselves, which are properly handled already).
27982 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27985 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27987 * combine.c (record_used_regs): New static function.
27988 (try_combine): Handle situations where there is an additional
27989 instruction between I2 and I3 which needs to have a LOG_LINK
27993 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27995 * combine.c (try_combine): Delete redundant i1 test. Call
27996 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27998 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28001 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28002 alternatives 13/14.
28004 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28006 PR tree-optimization/80054
28007 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28008 the optimization if a PHI or any of its arguments is not dominated
28009 by the candidate's basis. Use gphi* rather than gimple* as
28011 (replace_profitable_candidates): Clean up a gimple* variable that
28012 should be a gphi* variable.
28014 2017-03-20 Martin Sebor <msebor@redhat.com>
28017 * doc/extend.texi (attribute constructor): Document present limitation.
28019 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28022 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28023 __POWER9_VECTOR__ #ifdef control, change template definition to
28024 use Power9-specific built-in function.
28025 (vec_any_eq): Likewise.
28026 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28027 to control outcomes from this test.
28028 (vector_ae_<mode>p): For VEC_F modes, likewise.
28030 2017-03-20 Ian Lance Taylor <iant@google.com>
28032 * config/i386/i386.c (ix86_function_regparm): Save an extra
28033 register for -fsplit-stack with DECL_STATIC_CHAIN.
28035 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28038 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28039 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28041 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28043 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28045 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28048 2017-03-20 Marek Polacek <polacek@redhat.com>
28051 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28053 2017-03-20 Richard Biener <rguenther@suse.de>
28055 PR tree-optimization/80113
28056 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28057 allocate extra SSA name for PHI def.
28058 (add_close_phis_to_outer_loops): Likewise.
28059 (add_close_phis_to_merge_points): Likewise.
28060 (copy_loop_close_phi_args): Likewise.
28061 (copy_cond_phi_nodes): Likewise.
28063 2017-03-20 Martin Liska <mliska@suse.cz>
28065 PR middle-end/79753
28066 * tree-chkp.c (chkp_build_returned_bound): Do not build
28067 returned bounds for a LHS that's not a BOUNDED_P type.
28069 2017-03-20 Martin Liska <mliska@suse.cz>
28073 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28074 COMPLEX_CST and VECTOR_CST.
28076 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28079 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28080 target operand. A new splitter adds the clobber statement in case
28081 the target operand is dead anyway.
28083 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28085 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28086 to age-old versions of binutils and glibc.
28088 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28090 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28092 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28094 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28096 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28098 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28099 requirement for binutils 2.13.
28101 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28103 * combine.c (try_combine): Delete redundant i1 test. Call
28104 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28106 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28108 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28109 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28111 <riscv64-*-elf>: Re-arrange section
28112 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28113 <riscv32-*-linux>: Likewise.
28114 <riscv64-*-elf>: Likewise
28115 <riscv64-*-linux>: Likewise.
28117 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28120 * aarch64.opt(verbose-cost-dump): Fix typo.
28122 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28125 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28126 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28128 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28130 * reload.c (find_reloads): When reloading a nonoffsettable address,
28131 use RELOAD_OTHER for it and its address reloads.
28133 PR rtl-optimization/79910
28134 * combine.c (record_used_regs): New static function.
28135 (try_combine): Handle situations where there is an additional
28136 instruction between I2 and I3 which needs to have a LOG_LINK
28139 2017-03-17 Jeff Law <law@redhat.com>
28141 PR tree-optimization/71437
28142 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28143 conditional in the hash table first.
28144 (vrp_dom_walker::before_dom_children): Extract condition from
28145 ASSERT_EXPR. Record condition, its inverion and any implied
28146 conditions as well.
28148 2017-03-17 Marek Polacek <polacek@redhat.com>
28149 Markus Trippelsdorf <markus@trippelsdorf.de>
28151 PR tree-optimization/80079
28152 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28155 2017-03-17 Richard Biener <rguenther@suse.de>
28157 PR middle-end/80075
28158 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28159 Properly verify the LHS before the RHS possibly claims to be
28161 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28164 2017-03-17 Martin Jambor <mjambor@suse.cz>
28166 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28167 (List of -O2 options): Likewise.
28168 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28171 2017-03-17 Tom de Vries <tom@codesourcery.com>
28173 * gcov-dump.c (print_usage): Print bug_report_url.
28175 2017-03-17 Richard Biener <rguenther@suse.de>
28177 PR middle-end/80050
28178 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28179 (parser::peek): Likewise.
28181 2017-03-17 Richard Biener <rguenther@suse.de>
28183 PR tree-optimization/80048
28184 * sese.c (free_sese_info): Properly release rename_map and
28185 copied_bb_map elements.
28187 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28189 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28190 Add linked-list forward and backlinks. Insert on
28191 construction, remove on destruction.
28192 (class pass_store_merging): Add m_stores_head field.
28193 (pass_store_merging::terminate_and_process_all_chains):
28194 Iterate over m_stores_head list.
28195 (pass_store_merging::terminate_all_aliasing_chains):
28197 (pass_store_merging::execute): Check for debug stmts first.
28198 Push new chains onto the m_stores_head stack.
28200 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28203 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28204 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28205 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28207 2017-03-16 Jeff Law <law@redhat.com>
28209 PR tree-optimization/71437
28210 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28211 member function. Implementation moved into after_dom_children
28212 member function and into the threader's thread_outgoing_edges
28214 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28215 some code into new thread_outgoing_edges.
28216 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28217 definition. Simplify marker handling (do it here). Assume we always
28218 have the available expression and the const/copies tables.
28219 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28221 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28222 * tree-vrp.c (equiv_stack): No longer file scoped.
28223 (vrp_dom_walker): New class.
28224 (vrp_dom_walker::before_dom_children): New member function.
28225 (vrp_dom_walker::after_dom_children): Likewise.
28226 (identify_jump_threads): Setup domwalker. Use it rather than
28227 walking edges in a random order by hand. Simplify setup/finalization.
28228 (finalize_jump_threads): Remove.
28229 (vrp_finalize): Do not call identify_jump_threads here.
28230 (execute_vrp): Do it here instead and call thread_through_all_blocks
28233 PR tree-optimization/71437
28234 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28236 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28238 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28239 (dom_opt_dom_walker::thread_across_edge): Remove
28240 handle_dominating_asserts argument. All callers changed.
28241 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28242 changes. Remove calls to lhs_of_dominating_assert. Other
28243 uses of handle_dominating_asserts turn into unconditional code
28244 (simplify_control_stmt_condition_1): Likewise.
28245 (simplify_control_stmt_condition): Likewise.
28246 (thread_through_normal_block, thread_across_edge): Likewise.
28247 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28248 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28249 object if it is not an SSA_NAME.
28250 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28251 before calling into the VRP specific simplifiers.
28252 (identify_jump_threads): Remove handle_dominating_asserts
28255 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28258 * tree-diagnostic.c (default_tree_printer): No longer static.
28259 * tree-diagnostic.h (default_tree_printer): New prototype.
28261 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28263 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28264 Change ins into fmov.
28266 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28268 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28269 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28270 Use h_con constraint for operand 1.
28271 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28272 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28274 2017-03-15 Jeff Law <law@redhat.com>
28276 PR tree-optimization/71437
28277 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28278 (record_temporary_equivalences): Use it.
28280 PR tree-optimization/71437
28281 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28282 tree-ssa-scopedtables.
28283 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28284 (record_conditions, record_cond, vuse_eq): Likewise.
28285 (record_edge_info): Adjust to API tweak of record_conditions.
28286 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28287 (record_temporary_equivalences, optimize_stmt): Likewise.
28288 (eliminate_redundant_computations): Likewise.
28289 (record_equivalences_from_stmt): Likewise.
28290 * tree-ssa-scopedtables.c: Include options.h and params.h.
28291 (vuse_eq): New function, moved from tree-ssa-dom.c
28292 (build_and_record_new_cond): Likewise.
28293 (record_conditions): Likewise. Accept vector of conditions rather
28294 than edge_equivalence structure for first argument.
28295 for the first argument.
28296 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28297 from tree-ssa-dom.c.
28298 (avail_exprs_stack::record_cond): Likewise.
28299 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28300 from tree-ssa-dom.c.
28301 (avail_exprs_stack): Add new member functions lookup_avail_expr
28303 (record_conditions): Declare.
28305 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28308 * lra-constraints.c (process_alt_operands): Increase reject for
28309 reloading an input/output operand.
28311 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28314 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28315 insns to convert from signed/unsigned char/short to IEEE 128-bit
28317 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28319 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28322 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28323 subreg of inner mode for values already in registers.
28325 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28327 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28328 iteration reg is used after the loop.
28330 2017-03-14 Martin Sebor <msebor@redhat.com>
28332 PR tree-optimization/79800
28333 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28334 precision in negative-positive range.
28335 (format_floating): Call non-const overload with adjusted precision.
28337 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28340 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28343 2017-03-14 Martin Sebor <msebor@redhat.com>
28345 PR middle-end/80020
28346 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28347 * builtins.def (aligned_alloc): Use it.
28350 * Makefile.in (GTFILES): Add calls.c.
28351 * calls.c: Include "gt-calls.h".
28353 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28355 PR rtl-optimization/79728
28356 * regs.h (struct target_regs): New field
28357 x_contains_allocatable_regs_of_mode.
28358 (contains_allocatable_regs_of_mode): New macro.
28359 * reginfo.c (init_reg_sets_1): Initialize it, and change
28360 contains_reg_of_mode so it includes global regs as well.
28361 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28362 rather than contains_regs_of_mode.
28364 2017-03-14 Martin Liska <mliska@suse.cz>
28366 * doc/invoke.texi: Document options that can't be combined with
28367 -fcheck-pointer-bounds.
28369 2017-03-14 Martin Liska <mliska@suse.cz>
28371 PR middle-end/79831
28372 * doc/invoke.texi (-Wchkp): Document the option.
28374 2017-03-14 Martin Liska <mliska@suse.cz>
28376 * Makefile.in: Install gcov-dump.
28378 2017-03-14 Martin Liska <mliska@suse.cz>
28380 * multiple_target.c (expand_target_clones): Bail out for
28381 an invalid attribute.
28383 2017-03-14 Richard Biener <rguenther@suse.de>
28385 * alias.c (struct alias_set_entry): Pack properly.
28386 * cfgloop.h (struct loop): Likewise.
28387 * cse.c (struct set): Likewise.
28388 * ipa-utils.c (struct searchc_env): Likewise.
28389 * loop-invariant.c (struct invariant): Likewise.
28390 * lra-remat.c (struct cand): Likewise.
28391 * recog.c (struct change_t): Likewise.
28392 * rtl.h (struct address_info): Likewise.
28393 * symbol-summary.h (function_summary): Likewise.
28394 * tree-loop-distribution.c (struct partition): Likewise.
28395 * tree-object-size.c (struct object_size_info): Likewise.
28396 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28397 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28398 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28399 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28400 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28401 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28402 (struct _stmt_vec_info): Likewise.
28404 2017-03-14 Martin Liska <mliska@suse.cz>
28407 * multiple_target.c (create_dispatcher_calls): Check that
28408 a target can create a function dispatcher.
28410 2017-03-14 Martin Liska <mliska@suse.cz>
28413 * multiple_target.c (expand_target_clones): Drop local.local
28414 flag for default implementation.
28416 2017-03-14 Richard Biener <rguenther@suse.de>
28418 PR tree-optimization/80030
28419 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28421 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28423 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28424 gcc_fallthrough() instead of __attribute__((fallthrough));
28426 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28428 * doc/gcc.texi: Remove "up" link to (DIR).
28429 * doc/gccint.texi: Ditto.
28431 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28433 * doc/install.texi (Specific) <avr>: Remove reference to
28436 2017-03-13 Jeff Law <law@redhat.com>
28438 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28439 attribute rather than comments.
28441 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28442 match_scratch operand is highest.
28444 2017-03-13 Martin Liska <mliska@suse.cz>
28446 PR middle-end/78339
28447 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28448 is a CHKP clone, use original declaration.
28450 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28452 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28453 (arc_conditional_register_usage): Use a different allocation order
28454 when optimizing for size.
28455 * common/config/arc/arc-common.c (arc_option_optimization_table):
28456 Section anchors default on when optimizing for size.
28458 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28460 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28462 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28464 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28465 * config/arc/arc.md (cpu_facility): Add cd variant.
28466 (*movqi_insn): Add code density variant.
28467 (*movhi_insn): Likewise.
28468 (*movqi_insn): Likewise.
28469 (*addsi3_mixed): Likewise.
28470 (subsi3_insn): Likewise.
28472 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28474 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28476 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28478 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28479 expressions with MINUS and UNARY ops.
28481 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28484 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28486 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28487 between vec_select and vector argument.
28488 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28489 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28490 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28491 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28492 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28493 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28495 2017-03-13 Richard Biener <rguenther@suse.de>
28498 * params.def (vect-max-peeling-for-alignment): Fix typo.
28500 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28502 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28503 issue that only occurred with binutils below 2.18.
28505 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28507 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28508 refer to binutils 2.11/2.12 minimum.
28510 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28512 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28513 ftp.kernel.org and simplify binutils requirement.
28515 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28517 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28519 (Optimize Options): Ditto. Also remove redundancy.
28521 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28523 PR translation/79848
28524 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28526 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28527 to G_ to avoid double translation.
28529 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28531 PR translation/79923
28532 * auto-profile.c (get_combined_location): Convert leading
28533 character of diagnostics to lower case and remove trailing period.
28534 (read_profile): Likewise for various diagnostics.
28535 * config/arm/arm.c (arm_option_override): Remove trailing period
28536 from various diagnostics.
28537 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28538 (msp430_expand_delay_cycles): Likewise.
28540 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28543 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28544 full command-line argument, rather than just "str".
28545 (aarch64_validate_march): Likewise.
28546 (aarch64_validate_mtune): Likewise.
28548 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28550 PR rtl-optimization/78911
28551 * lra-assigns.c (must_not_spill_p): New function.
28552 (spill_for): Use it.
28554 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28556 PR tree-optimization/79981
28557 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28558 ATOMIC_COMPARE_EXCHANGE ifn result.
28559 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28560 IFN_ATOMIC_COMPARE_EXCHANGE.
28562 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28565 * opts.c (parse_sanitizer_options): Add missing question mark to
28566 "did you mean" message.
28568 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28570 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28572 (VMULEUH_UNS): Likewise.
28573 (VMULOUB_UNS): Likewise.
28574 (VMULOUH_UNS): Likewise.
28575 * config/rs6000/rs6000.c (builtin_function_type): Remove
28576 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28578 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28581 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28582 x with result of extra_parsing_for_operand_code_0.
28583 (function_reader::extra_parsing_for_operand_code_0): Convert
28584 return type from void to rtx, returning x. When reading
28585 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28586 larger size containing struct block_symbol.
28588 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28590 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28591 -mfloat128-hardware without -m64.
28593 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28596 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28597 entries to the case statement that marks unsigned arguments to
28598 overloaded functions.
28600 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28602 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28603 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28605 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28608 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28609 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28611 2017-03-10 Martin Liska <mliska@suse.cz>
28615 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28616 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28619 2017-03-10 Olivier Hainque <hainque@adacore.com>
28621 * tree-switch-conversion (array_value_type): Start by resetting
28622 candidate type to it's main variant.
28624 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28626 PR rtl-optimization/79909
28627 * combine.c (try_combine): Use simplify_replace_rtx on individual
28628 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28629 of the whole CALL_INSN_FUNCTION_USAGE.
28631 PR tree-optimization/79972
28632 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28633 get_range_info on SSA_NAMEs. Formatting fixes.
28635 2017-03-10 Richard Biener <rguenther@suse.de>
28636 Jakub Jelinek <jakub@redhat.com>
28638 PR tree-optimization/77975
28639 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28640 edge to be constant.
28641 (get_val_for): For constant x return it. Formatting fix.
28642 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28643 would use the same bases as the current one.
28645 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28647 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28648 instead of vec_select for V1TImode.
28649 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28651 (VSX_LE_128): Add V1TI to this mode iterator.
28652 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28653 (*vsx_le_perm_store_<mode>): Likewise.
28654 (pre-reload splitter for VSX stores): Likewise.
28655 (post-reload splitter for VSX stores): Likewise.
28656 (*vsx_xxpermdi2_le_<mode>): Likewise.
28657 (*vsx_lxvd2x2_le_<mode>): Likewise.
28658 (*vsx_stxvd2x2_le_<mode>): Likewise.
28660 2017-03-09 Michael Eager <eager@eagercon.com>
28662 Correct failures with --enable-checking=yes,rtl.
28664 * config/microblaze/microblaze.c (microblaze_expand_shift):
28665 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28666 test for const0_rtx.
28667 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28668 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28670 2017-03-09 Richard Biener <rguenther@suse.de>
28672 PR tree-optimization/79977
28673 * graphite-scop-detection.c (scop_detection::merge_sese):
28674 Handle the case of extra exits to blocks dominating the entry.
28676 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28678 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28681 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28683 PR rtl-optimization/79949
28684 * lra-constraints.c (process_alt_operands): Check memory when
28685 trying to predict a cycle. Print about the overall increase.
28687 2017-03-09 Richard Biener <rguenther@suse.de>
28689 PR middle-end/79971
28690 * gimple-expr.c (useless_type_conversion_p): Preserve
28691 TYPE_SATURATING for fixed-point types.
28693 2017-03-09 Richard Biener <rguenther@suse.de>
28696 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28697 alignment of BLKmode params.
28699 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28702 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28703 (VALL_NO_V2Q): Likewise.
28705 * config/aarch64/aarch64-simd.md
28706 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28708 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28709 VALL_NO_V2Q mode iterator.
28710 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28712 2017-03-09 Martin Liska <mliska@suse.cz>
28714 PR tree-optimization/79631
28715 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28716 tree_int_cst_sign_bit just for INTEGER constants.
28718 2017-03-09 Martin Liska <mliska@suse.cz>
28722 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28725 2017-03-09 Marek Polacek <polacek@redhat.com>
28728 * tree.c (inchash::add_expr): Handle TREE_VEC.
28730 2017-03-09 Martin Liska <mliska@suse.cz>
28733 (chkp_narrow_size_and_offset): New function.
28734 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28735 (void chkp_parse_bit_field_ref): New function.
28736 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28737 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28739 2017-03-09 Martin Liska <mliska@suse.cz>
28742 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28743 (chkp_find_bounds_1): Remove gcc_unreachable.
28745 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28748 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28749 BUILT_IN_SYNC*, determine the access type from the size suffix and
28750 always build a MEM_REF with that type. Handle forgotten
28751 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28754 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28755 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28756 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28757 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28758 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28759 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28760 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28761 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28762 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28763 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28764 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28765 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28766 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28767 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28768 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28769 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28770 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28771 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28772 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28773 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28774 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28775 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28776 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28777 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28778 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28779 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28780 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28781 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28782 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28783 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28784 definitions outside of __OPTIMIZE__ guarded section.
28787 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28788 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28789 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28790 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28793 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28795 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28796 ("vfenez<mode>"): Add missing constraints.
28798 2017-03-08 Martin Sebor <msebor@redhat.com>
28801 * config/nds32/nds32.c (nds32_option_override):
28802 Fix misspelled diagnostic.
28804 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28807 * gimplify.c (gimplify_omp_for): Replace index var in outer
28808 taskloop statement with an artificial variable and add
28809 OMP_CLAUSE_PRIVATE clause for it.
28811 2017-03-08 Richard Biener <rguenther@suse.de>
28813 PR tree-optimization/79955
28814 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28815 for accesses that are completely outside of the variable.
28817 2017-03-08 Andrew Haley <aph@redhat.com>
28819 PR tree-optimization/79943
28820 * tree-ssa-loop-split.c (compute_new_first_bound): When
28821 calculating the new upper bound, (END-BEG) should be added, not
28824 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28826 * config/avr/avr.md (setmemhi): Make sure match_dup
28827 operand number comes before match_scratch.
28829 2017-03-08 Richard Biener <rguenther@suse.de>
28831 PR tree-optimization/79920
28832 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28833 with ncopies == 1 to ...
28834 (vect_transform_slp_perm_load): ... here. Properly compute
28835 all element loads by iterating VF times over the group. Do
28836 not handle ncopies (computed in a broken way) in
28837 vect_create_mask_and_perm.
28839 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28842 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28843 is a uniform vector, use uniform_vector_p return value instead of
28844 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28846 2017-03-07 Marek Polacek <polacek@redhat.com>
28848 PR middle-end/79809
28849 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28850 (alloca_call_type): Likewise.
28852 2017-03-07 Martin Liska <mliska@suse.cz>
28854 * gcov.c (process_args): Put comment to correct location.
28856 2017-03-07 Martin Liska <mliska@suse.cz>
28858 PR middle-end/68270
28859 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28860 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28861 (chkp_narrow_bounds_for_field): Likewise.
28862 (chkp_parse_array_and_component_ref): Pass one more argument to
28865 2017-03-07 Richard Biener <rguenther@suse.de>
28867 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28870 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28872 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28873 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28875 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28878 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28879 to end of description.
28880 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28882 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28884 PR rtl-optimization/79901
28885 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28887 (*avx512f_<code><mode>3<mask_name>): ... this.
28888 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28889 iterator instead of VI8_AVX2_AVX512BW.
28891 PR rtl-optimization/79901
28892 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28893 min/max expander, expand it using expand_vec_cond_expr.
28896 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28899 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28902 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28903 to void * for PCH reasons.
28904 * dwarf2out.c (output_loc_operands, output_die): Cast
28905 v.val_vec.array to unsigned char *.
28907 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28910 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28913 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28915 PR rtl-optimization/79571
28916 * lra-constraints.c (process_alt_operands): Calculate static
28917 reject and subtract it from overall when only addresses will be
28920 2017-03-06 Julia Koval <julia.koval@intel.com>
28923 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28924 incoming stack boundary to 128 for 64-bit targets.
28926 2017-03-06 Richard Biener <rguenther@suse.de>
28928 PR tree-optimization/79894
28929 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28930 to NULL after folding it.
28932 2017-03-06 Richard Biener <rguenther@suse.de>
28934 PR tree-optimization/79824
28935 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28936 check disabling peeling for gaps.
28938 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28940 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28941 attributes): Document gettimeofday.
28943 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28945 * config/s390/s390.c (s390_option_override_internal): Set
28946 PARAM_MIN_VECT_LOOP_BOUND
28948 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28950 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28951 * config/s390/s390.md: Likewise.
28953 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28956 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28957 (<avx2_avx512>_perm<mode>): Rename to ...
28958 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28960 (<avx512>_perm<mode>_mask): Rename to ...
28961 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28963 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28964 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28965 instead of VI8F_256_512.
28966 (avx512f_perm<mode>): New define_expand.
28967 (avx512f_perm<mode>_mask): Likewise.
28968 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28969 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28971 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28973 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28974 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28976 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28978 2017-03-06 Martin Liska <mliska@suse.cz>
28981 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28982 when having a SSA NAME w/o VAR_DECL assigned to it.
28984 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28986 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28987 msa_dpsub_<su>_d): Fix MODE for vec_select.
28989 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28991 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28993 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28995 2017-03-06 Richard Biener <rguenther@suse.de>
28997 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28998 * plugin.c (register_plugin_info): Likewise.
28999 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29001 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29003 * config/i386/sse.md (sse_storehps, sse_storelps,
29004 avx_<castmode><avxsizesuffix>_<castmode>,
29005 avx512f_<castmode><avxsizesuffix>_<castmode>,
29006 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29007 in condition that at least one operand is not a MEM.
29009 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29011 PR middle-end/79805
29012 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29013 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29015 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29016 gimple_call_nothrow_p flag based on whether original builtin can throw.
29017 If it can, emit following stmts on the fallthrough edge.
29018 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29019 don't create new bb if inserting just debug stmts on the edge, try to
29020 insert them on the fallthru bb or just reset debug stmts.
29022 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29025 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29026 restore recog_data (including the operand rtxes inside it) around
29027 the call to get_insn_template.
29029 2017-03-03 Martin Sebor <msebor@redhat.com>
29031 PR tree-optimization/79699
29032 * context.c (context::~context): Free MPFR caches to avoid
29033 a memory leak on program exit.
29035 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29037 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29038 Use wide_int::ulow () instead of .elt (0).
29040 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29042 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29043 (*pushxf): Limit oF constraint to 32bit targets and add oC
29044 constraint for 64bit targets.
29045 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29046 (*pushdf): Change rmF constraint to rmC.
29048 2017-03-03 Martin Liska <mliska@suse.cz>
29050 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29051 Remove unused variable.
29053 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29056 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29057 is a memory operand, increase num_memory.
29058 (ix86_expand_args_builtin): Likewise.
29060 2017-03-03 Jan Hubicka <jh@suse.cz>
29063 * ipa-devirt.c (maybe_record_node): Properly handle
29064 __cxa_pure_virtual visibility.
29066 2017-03-03 Martin Liska <mliska@suse.cz>
29068 PR tree-optimization/79803
29069 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29071 (pass_loop_prefetch::execute): Disabled optimization if an
29072 assumption about L1 cache size is not met.
29074 2017-03-03 Martin Liska <mliska@suse.cz>
29076 PR rtl-optimization/79574
29077 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29078 (hash_scan_set): Likewise.
29079 (dump_hash_table): Likewise.
29080 (hoist_code): Likewise.
29082 2017-03-03 Richard Biener <rguenther@suse.de>
29084 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29085 in place of uhigh ().
29086 (fixed_convert_from_real): Likewise.
29088 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29091 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29093 2017-03-03 Richard Biener <rguenther@suse.de>
29095 PR middle-end/79818
29096 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29097 TYPE_OVERFLOW_UNDEFINED check.
29099 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29101 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29103 (vector_ae_<mode>_p): Likewise.
29104 (vector_nez_<mode>_p): Likewise.
29105 (vector_ne_v2di_p): Likewise.
29106 (vector_ae_v2di_p): Likewise.
29107 (vector_ne_<mode>_p): Likewise.
29108 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29110 (vsx_tsqrt<mode>2_fe): Likewise.
29112 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29115 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29117 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29119 PR rtl-optimization/79780
29120 * cprop.c (one_cprop_pass): When second and further conditional trap
29121 in a single basic block is turned into an unconditional trap, turn it
29122 into a deleted note to avoid RTL verification failures.
29124 2017-03-02 Richard Biener <rguenther@suse.de>
29126 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29128 2017-03-02 Richard Biener <rguenther@suse.de>
29130 PR tree-optimization/79345
29132 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29133 param and abort the walk, returning -1 if it is hit.
29134 (walk_aliased_vdefs): Take a limit param and pass it on.
29135 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29136 defaulting to 0 and return a signed int.
29137 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29138 (check_defs): New helper.
29139 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29140 about uninitialized memory.
29141 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29142 bogus uninitialized warning.
29143 (fixed_convert_from_real): Likewise.
29145 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29147 PR tree-optimization/66768
29148 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29149 iv_use if base object can't be determined.
29151 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29153 PR tree-optimization/79345
29154 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29155 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29156 (get_pattern_stats): Initialize it.
29157 * genemit.c (gen_expand): Verify match_scratch numbers come after
29158 match_operand/match_dup numbers.
29159 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29160 match_scratch numbers.
29161 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29163 * config/s390/s390.md (trunctdsd2): Likewise.
29165 2017-03-02 Richard Biener <rguenther@suse.de>
29167 * wide-int.h (wide_int_storage::operator=): Implement in terms
29170 2017-03-02 Richard Biener <rguenther@suse.de>
29172 PR tree-optimization/79777
29173 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29174 the to insert expression to sth existing.
29176 2017-03-01 Martin Sebor <msebor@redhat.com>
29178 PR middle-end/79692
29179 * gimple-ssa-sprintf.c
29180 (directive::known_width_and_precision): New function.
29181 (format_integer): Use it.
29182 (get_mpfr_format_length): Consider the full range of precision
29183 when computing %g output with the # flag. Set the likely byte
29184 count to 3 rather than 1 when precision is indeterminate.
29185 (format_floating): Correct the lower bound of precision.
29187 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29189 * doc/invoke.texi: Document default code model for 64-bit Linux.
29191 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29194 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29195 udiv rather than div since input pattern is unsigned.
29197 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29199 * config/i386/i386.c (print_reg): Warn for values of
29200 unsupported size in integer register.
29202 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29205 * config/rs6000/predicates.md (current_file_function_operand): Do
29206 not allow self calls to be local if the function is replaceable.
29208 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29211 * config/rs6000/altivec.h (vec_ctz and others): Change the
29212 preprocessor macro that controls conditional compilation from
29213 _ARCH_PWR9 to __POWER9_VECTOR__.
29214 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29215 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29216 control (instead of _ARCH_PWR9 control) so that template
29217 definition uses power9-specific function.
29218 (vec_any_eq): Likewise.
29219 (vec_all_ne): Change macro definition to use a power9-specific
29220 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29221 _ARCH_PWR9 control).
29222 (vec_any_eq) Likewise.
29223 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29224 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29225 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29226 support for xvcmpnedp instruction.
29227 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29228 macro expansion so that Power9 implementation of vec_all_ne does
29229 not use the AltiVec predicate framework.
29230 (VCMPNEH_P): Likewise.
29231 (VCMPNEW_P): Likewise.
29232 (VCMPNED_P): Likewise.
29233 (VCMPNEFP_P): Likewise.
29234 (VCMPNEDP_P): Likewise.
29235 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29236 implementation of vec_any_eq to not use AltiVec predicate
29238 (VCMPAEH_P): Likewise.
29239 (VCMPAEW_P): Likewise.
29240 (VCMPAED_P): Likewise.
29241 (VCMPAEFP_P): Likewise.
29242 (VCMPAEDP_P): Likewise.
29243 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29244 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29245 not use the AltiVec predicate framework.
29246 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29247 of vec_any_eq to not use AltiVec predicate framework.
29248 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29249 support for predefined __POWER9_VECTOR__ macro to indicate that
29250 Power9 instruction selection is enabled.
29251 (altivec_overloaded_builtins): Remove extraneous
29252 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29253 function argument types RS6000_BTI_bool_V16QI and
29254 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29255 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29256 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29257 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29258 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29259 Power9 for implementations of vec_cmpne. Change the signature for
29260 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29261 (representing vec_all_ne) to remove the previously described first
29262 argument of type RS6000_BTI_INTSI, as this was an artifact of
29263 reliance on the AltiVec predicate framework, which is no longer
29264 used in the implementation of these functions. Add
29265 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29266 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29267 since, unlike the AltiVec predicate framework implementation, we
29268 do not share function descriptors between vec_alle and vec_anyeq.
29269 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29270 set of modes that receive special treatment even when
29271 TARGET_P9_VECTOR is true. The special treatment emits code that
29272 does not depend on Power9 instructions.
29273 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29274 define_expand to not rely on AltiVec predicate framework.
29275 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29277 (vector_ne_v2di_p): Change this define_expand to not rely on
29278 AltiVec predicate framework.
29279 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29281 (vector_ne_<mode>_p): Change this define_expand to not rely on
29282 AltiVec predicate framework.
29283 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29285 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29286 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29287 define_insn pattern.
29288 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29289 define_insn pattern because the xvcmpne<VSs>. instruction is not
29291 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29292 instruction is not supported.
29294 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29296 * config/nvptx/nvptx.c: Include intl.h.
29298 2017-03-01 Martin Jambor <mjambor@suse.cz>
29301 * ipa-prop.h (ipa_bits): Removed field known.
29302 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29303 to pointers. Adjusted their comments to warn about their sharing.
29304 (ipcp_transformation_summary): Change bits to a vector of pointers.
29305 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29306 (ipa_get_ipa_bits_for_value): Declare.
29307 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29308 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29309 (ipa_bits_hash_table): Likewise.
29310 (ipa_vr_ggc_hash_traits): Likewise.
29311 (ipa_vr_hash_table): Likewise.
29312 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29313 being pointers and vr_known being removed.
29314 (ipa_set_jf_unknown): Likewise.
29315 (ipa_get_ipa_bits_for_value): New function.
29316 (ipa_set_jfunc_bits): Likewise.
29317 (ipa_get_value_range): New overloaded functions.
29318 (ipa_set_jfunc_vr): Likewise.
29319 (ipa_compute_jump_functions_for_edge): Use the above functions to
29320 construct bits and vr parts of jump functions.
29321 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29322 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29324 (ipcp_grow_transformations_if_necessary): Also allocate
29325 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29327 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29328 them. Fix too long lines.
29329 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29330 vr_known being removed.
29331 (ipa_read_jump_function): Use new setter functions to construct bits
29332 and vr parts of jump functions or set them to NULL.
29333 (write_ipcp_transformation_info): Adjust for bits being pointers.
29334 (read_ipcp_transformation_info): Likewise.
29335 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29337 Include gt-ipa-prop.h.
29338 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29340 (ipcp_store_bits_results): Likewise.
29341 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29342 Do not write to existing jump functions but use a temporary instead.
29344 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29347 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29348 attempt to use its first operand as BIT_FIELD_REF base.
29350 2017-03-01 Richard Biener <rguenther@suse.de>
29352 PR middle-end/79721
29353 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29354 interpolating formula in wrapping arithmetic.
29355 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29357 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29359 PR tree-optimization/79734
29360 * tree-vect-generic.c (expand_vector_condition): Optimize
29361 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29362 Handle VEC_COND_EXPR where comparison has different inner width from
29363 type's inner width.
29365 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29367 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29368 markup, and similar issues. Remove @opindex entries for things
29369 that aren't options. Add missing -mmpy-option entries.
29371 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29373 PR tree-optimization/79737
29374 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29375 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29376 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29377 instead of byte_size. Formatting fix.
29378 (shift_bytes_in_array_right): Formatting fix.
29380 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29383 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29384 condition on optimize for the leaf function test.
29386 2017-02-28 Martin Liska <mliska@suse.cz>
29389 * read-rtl-function.c (function_reader::handle_unknown_directive):
29390 Bail out when one uses -flto.
29392 2017-02-28 Martin Liska <mliska@suse.cz>
29394 * common.opt: Replace space with tabular for options of <number>
29396 * config/i386/i386.opt: Show <number> value for
29397 -mlarge-data-threshold.
29398 * opts.c (print_filtered_help): Do not display number in hexadecimal
29401 2017-02-28 Martin Liska <mliska@suse.cz>
29403 * common.opt: Fix --help=option -Q for options which are of
29406 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29408 * config/i386/i386.c (print_reg): Error out for values
29409 of 8-bit size in invalid integer register.
29411 2017-02-28 Martin Sebor <msebor@redhat.com>
29413 PR tree-optimization/79691
29414 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29416 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29419 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29420 gcc_unreachable with output_operand_lossage.
29422 2017-02-28 Richard Biener <rguenther@suse.de>
29424 PR tree-optimization/79740
29425 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29427 (visit_nary_op): Insert the nary into the hashtable if we
29428 pattern-matched sth.
29429 * tree-ssa-pre.c (eliminate_insert): Robustify.
29431 2017-02-28 Richard Biener <rguenther@suse.de>
29433 PR middle-end/79731
29434 * fold-const.c (decode_field_reference): Reject out-of-bound
29437 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29439 * config/i386/i386.c: Include intl.h.
29440 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29441 instead of just cond ? "..." : "...".
29442 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29443 * coverage.c (read_counts_file): Likewise.
29444 * omp-offload.c: Include intl.h.
29445 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29446 of just cond ? "..." : "...".
29447 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29448 of just cond ? "..." : "...".
29450 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29453 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29455 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29456 'tune for' CPU name.
29457 * config/arm/arm-cpu-data.h: Regenerated.
29459 2017-02-28 Richard Biener <rguenther@suse.de>
29461 PR tree-optimization/79732
29462 * tree-inline.c (expand_call_inline): Do not shadow var.
29464 2017-02-28 Richard Biener <rguenther@suse.de>
29466 PR tree-optimization/79723
29467 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29468 address-space properly.
29470 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29472 * doc/optinfo.texi (Optimization groups): Fix option used for
29474 * doc/invoke.texi (-fopt-info): Document "omp".
29475 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29476 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29477 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29478 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29479 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29481 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29483 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29484 OPTGROUP_OMP with "omp".
29486 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29489 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29490 for arithmetic shift of unsigned V2DI.
29492 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29494 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29495 arc/linux.h headers.
29496 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29497 (LINK_SPEC): Likewise.
29498 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29499 (EXTRA_SPECS): Likewise.
29500 (STARTFILE_SPEC): Likewise.
29501 (ENDFILE_SPEC): Likewise.
29502 (LIB_SPEC): Likewise.
29503 (TARGET_SDATA_DEFAULT): Likewise.
29504 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29505 (MULTILIB_DEFAULTS): Likewise.
29506 (DWARF2_UNWIND_INFO): Likewise.
29507 * config/arc/big.h: New file.
29508 * config/arc/elf.h: Likewise.
29509 * config/arc/linux.h: Likewise.
29510 * config/arc/t-uClibc: Remove.
29512 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29514 PR tree-optimization/77536
29515 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29516 (tree_transform_and_unroll_loop): Use above function to compute the
29517 estimated niter of unrolled loop and use it when scaling profile.
29518 Also use count info rather than frequency if it's non-zero.
29519 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29520 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29521 (vect_transform_loop): Call above function.
29523 2017-02-27 Richard Biener <rguenther@suse.de>
29525 PR tree-optimization/45397
29526 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29527 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29528 (visit_nary_op): Add pattern matching for CSEing sign-changed
29529 or truncated operations with wider ones.
29531 2017-02-27 Richard Biener <rguenther@suse.de>
29533 PR tree-optimization/79690
29534 * tree-vect-stmts.c (vectorizable_store): Use vector type
29535 built from the DR with address-space.
29537 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29539 * doc/invoke.texi (Optimize Options): Refine the description
29540 of asan-use-after-return.
29542 2017-02-25 Alan Modra <amodra@gmail.com>
29544 PR rtl-optimization/79584
29545 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29546 base, not ad->base_term, the reg within base. Remove assertion
29547 that ad->base == ad->base_term. Replace gen_int_mode using
29548 bogus mode with const0_rtx.
29550 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29552 PR middle-end/79396
29553 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29554 FMA_EXPR like tcc_binary or tcc_unary.
29556 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29559 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29561 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29562 (output_loc_operands): Handle DW_OP_call_ref and
29563 DW_OP_GNU_variable_value.
29564 (struct variable_value_struct): New type.
29565 (struct variable_value_hasher): Likewise.
29566 (variable_value_hash): New variable.
29567 (string_types): Remove.
29568 (copy_loc_descr): New function.
29569 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29570 (prepend_loc_descr_to_each): New function.
29571 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29572 instead of add_loc_descr_to_each if the first argument is single
29573 location list and the second has multiple.
29574 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29575 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29576 when looking for variable value which doesn't have other location info.
29577 (loc_list_from_tree): Formatting fix.
29578 (gen_array_type_die): Simplify DW_AT_string_length handling.
29579 (adjust_string_types): Remove.
29580 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29581 string_types. Call resolve_variable_values.
29582 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29583 (resolve_addr_in_expr): Likewise. Add A argument.
29584 (copy_deref_exprloc): Remove deref argument. Adjust for the
29585 original expression being DW_OP_GNU_variable_value with optionally
29586 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29587 optionally after it.
29588 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29589 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29590 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29591 (variable_value_hasher::hash, variable_value_hasher::equal): New
29593 (resolve_variable_value_in_expr, resolve_variable_value,
29594 resolve_variable_values, note_variable_value_in_expr,
29595 note_variable_value): New functions.
29596 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29599 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29602 * opts.h (handle_generated_option): Add GENERATED_P argument.
29603 * opts-common.c (handle_option): Adjust function comment.
29604 (handle_generated_option): Add GENERATED_P argument, pass it to
29606 (control_warning_option): Pass false to handle_generated_option
29608 * opts.c (maybe_default_option): Pass true to handle_generated_option
29610 * optc-gen.awk: Likewise.
29612 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29614 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29615 a REG, look at the REG it is a SUBREG of.
29616 (splitter for cmpeqsi_t): Ditto.
29618 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29620 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29621 the special USEs with the pattern of the insn, not the insn itself.
29623 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29626 * doc/invoke.texi: Document -mload-store-pairs.
29628 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29629 Sandra Loosemore <sandra@codesourcery.com>
29631 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29632 argument isn't a CONST_INT.
29633 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29634 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29635 (nios2_validate_compare): Bypass alternate compare logic if *op2
29636 is not a CONST_INT.
29637 (ldstwm_operation_p): Return false if first_base is not a REG or
29638 if first_offset is not a CONST_INT.
29640 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29642 * config/cris/cris.md: Use correct operand in a define_peephole2.
29644 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29646 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29648 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29650 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29651 this_insn if it is an INSN or JUMP_INSN.
29652 (force_offsettable): Look at base, not at addr.
29653 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29654 on things that aren't necessarily CONST_INTs.
29656 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29658 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29659 -mfpmath=sse is the default also for x86-32 targets with SSE2
29660 instruction set when @option{-ffast-math} is enabled
29662 2017-02-24 Jeff Law <law@redhat.com>
29664 PR rtl-optimizatoin/79286
29665 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29668 2017-02-24 Richard Biener <rguenther@suse.de>
29670 PR tree-optimization/79389
29671 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29674 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29676 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29677 function comment to reflect reality.
29678 (loop_exits_before_overflow): Fix typo in function description.
29680 2017-02-24 Richard Biener <rguenther@suse.de>
29682 PR tree-optimization/79389
29683 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29684 properly that a threading opportunity exists. Detect conditional
29685 copy/constant propagation opportunities.
29687 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29689 * config/visium/visium.md (type): Add trap.
29690 (b): New mode attribute.
29691 (*btst): Rename into...
29692 (*btst<mode>): ...this and adjust.
29693 (*cbranchsi4_btst_insn): Rename into...
29694 (*cbranch<mode>4_btst_insn): ...this and adjust.
29695 (trap): New define_insn.
29697 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29699 PR tree-optimization/79389
29700 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29701 (noce_reversed_cond_code): New function.
29702 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29703 reversed_comparison_code. Formatting fix.
29704 (noce_try_store_flag): Test rev_cond != NULL in addition to
29705 reversed_comparison_code.
29706 (noce_try_store_flag_constants): Likewise.
29707 (noce_try_store_flag_mask): Likewise.
29708 (noce_try_addcc): Use rev_cond if non-NULL instead of
29709 reversed_comparison_code.
29710 (noce_try_cmove_arith): Likewise. Formatting fixes.
29711 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29712 (noce_find_if_block): Initialize rev_cond.
29713 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29714 instead of false as last argument never attempt to reverse it
29717 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29719 PR tree-optimization/79663
29720 * tree-predcom.c (combine_chains): Process refs in reverse order
29721 only for ZERO length chains, and add explaining comment.
29723 2017-02-23 Jeff Law <law@redhat.com>
29725 PR tree-optimization/79578
29726 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29727 in call to operand_equal_p.
29729 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29732 * config/i386/cpuid.h: Fix another undefined behavior.
29734 2017-02-23 Richard Biener <rguenther@suse.de>
29736 PR tree-optimization/79683
29737 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29738 vector types for data-refs.
29740 2017-02-23 Martin Liska <mliska@suse.cz>
29742 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29744 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29746 PR middle-end/79665
29747 * internal-fn.c (get_range_pos_neg): Moved to ...
29748 * tree.c (get_range_pos_neg): ... here. No longer static.
29749 * tree.h (get_range_pos_neg): New prototype.
29750 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29751 are known to be in between 0 and signed maximum inclusive, try to
29752 expand both unsigned and signed divmod and use the cheaper one from
29755 2017-02-22 Jeff Law <law@redhat.com>
29757 PR tree-optimization/79578
29758 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29759 to compare base operands.
29761 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29764 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29765 gpc_reg_operand instead of fpr_reg_operand.
29767 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29769 * config/mips/mips.c (mips_return_in_memory): Force FP
29770 vector types to be returned in memory for o32 ABI.
29772 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29774 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29775 instead of DW_TAG_member for static data member declarations and don't
29776 set no_linkage_name for static inline data members.
29777 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29780 2017-02-22 Martin Liska <mliska@suse.cz>
29782 * doc/invoke.texi: Replace inequality signs with square brackets
29785 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29788 * lra-constraints.c (simplify_operand_subreg): Handle
29789 WORD_REGISTER_OPERATIONS targets.
29791 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29794 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29795 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29796 elimination by swapping fld*.
29798 2017-02-22 Richard Biener <rguenther@suse.de>
29800 PR tree-optimization/79673
29801 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29802 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29803 irrelevant address-space qualifiers and avoiding a
29804 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29806 2017-02-22 Richard Biener <rguenther@suse.de>
29808 PR tree-optimization/79666
29809 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29810 to not symbolically negate if that may introduce undefined
29813 2017-02-22 Martin Liska <mliska@suse.cz>
29816 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29817 * data-streamer-out.c (streamer_write_gcov_count_stream):
29819 * value-prof.c (stream_out_histogram_value): Make assert more
29820 precise based on type of counter.
29822 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29825 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29826 Use nonimmediate_operand instead of memory_operand for operand 1.
29827 (float-extend standard_x87sse_constant_load splitter): Ditto.
29829 2017-02-21 Jeff Law <law@redhat.com>
29831 PR tree-optimization/79621
29832 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29833 blocks with edges to themselves.
29835 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29838 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29839 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29840 Use gimple_call_builtin_p.
29843 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29844 on temporarily removed DEBUG_INSNs.
29846 PR tree-optimization/79649
29847 * tree-loop-distribution.c (classify_partition): Give up on
29848 non-generic address space loads/stores.
29850 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29852 * doc/loop.texi (Loop manipulation): Remove nonexistent
29853 tree_ssa_loop_version from the documentation.
29854 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29856 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29859 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29860 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29861 * config/rs6000/rs6000.c: Include except.h.
29862 (rs6000_expand_split_stack_prologue): Call
29863 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29865 2017-02-21 Martin Jambor <mjambor@suse.cz>
29868 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29869 have been analyzed.
29871 2017-02-21 Martin Jambor <mjambor@suse.cz>
29873 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29874 for backward compatibility only.
29875 * doc/invoke.texi (Option Summary): Remove all references to
29876 -fipa-cp-alignment.
29878 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29882 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29884 * lra-constraints.c (curr_insn_transform): Handle
29885 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29887 2017-02-21 Martin Liska <mliska@suse.cz>
29889 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29891 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29893 PR translation/79638
29894 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29896 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29899 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29900 (arm_function_ok_for_sibcall): Return false for an indirect call by
29901 descriptor if all the argument registers are used.
29902 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29903 alignment of the function.
29905 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29907 PR tree-optimization/61441
29908 * simplify-rtx.c (simplify_const_unary_operation): For
29909 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29910 the sNaN unmodified.
29912 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29914 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29915 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29916 instead of SYSTEM_HEADER_DIR.
29918 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29919 Martin Liška <mliska@suse.cz>
29921 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29922 Fix typos and grammar, use active voice, and clarify.
29924 2017-02-20 Marek Polacek <polacek@redhat.com>
29926 PR middle-end/79537
29927 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29930 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29932 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29935 * config/i386/i386.c (ix86_expand_builtin): Handle
29936 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29937 ix86_builtins_isa[fcode].isa as a requirement of those
29938 flags and any other flag in the bitmask.
29939 (ix86_init_mmx_sse_builtins): Use 0 instead of
29940 ~OPTION_MASK_ISA_64BIT as mask.
29941 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29942 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29943 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29944 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29946 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29949 * lra-constraints.c (split_reg): Check requested split mode
29950 is supported by the register.
29952 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29954 * lra-constraints.c (simplify_operand_subreg): Remove early
29957 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29960 * lra-constraints.c (curr_insn_transform): Tighten condition
29961 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29963 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29966 * lra-constraints.c (curr_insn_transform): Handle
29967 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29969 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29972 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29974 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29976 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29979 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29982 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29984 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29987 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29989 * final.c (last_columnnum, override_columnnum): New variables.
29990 (final_start_function): Set last_columnnum, pass it to begin_prologue
29991 hook and pass 0 to dwarf2out_begin_prologue.
29992 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29993 to source_line debug hook.
29994 (notice_source_line): Compute last_columnnum and for debug_column_info
29995 return true on column changes.
29996 * debug.h (struct gcc_debug_hooks): Add column argument to
29997 source_line and begin_prologue hooks.
29998 (debug_nothing_int_charstar_int_bool): Remove prototype.
29999 (debug_nothing_int_int_charstar,
30000 debug_nothing_int_int_charstar_int_bool): New prototypes.
30001 (dwarf2out_begin_prologue): Add column argument.
30002 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30003 begin_prologue hooks.
30004 (debug_nothing_int_charstar_int_bool): Remove.
30005 (debug_nothing_int_int_charstar,
30006 debug_nothing_int_int_charstar_int_bool): New functions.
30007 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30008 through to dwarf2out_source_line.
30009 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30010 (dwarf2out_source_line): Add column argument, emit it if requested.
30011 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30013 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30014 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30015 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30016 through to dwarf2out_begin_prologue.
30017 (vmsdbgout_source_line): Add column argument, pass it through to
30018 dwarf2out_source_line.
30019 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30020 dbxout_source_line caller.
30021 (dbxout_source_line): Add column argument.
30023 * common.opt (gno-column-info, gcolumn-info): New options.
30024 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30025 (check_die): Also test for multiple DW_AT_decl_column attributes.
30026 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30027 DW_AT_decl_column if requested.
30028 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30030 (gen_variable_die): Likewise.
30031 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30032 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30035 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30036 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30037 (ix86_handle_option): Handle OPT_m3dnowa.
30038 * doc/invoke.texi (-m3dnowa): Document.
30039 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30040 -m3dnowa instead of -m3dnow -march=athlon.
30043 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30044 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30046 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30049 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30050 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30051 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30052 generator for vsx_xxpermdi_<mode>_be.
30053 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30054 force big-endian semantics.
30055 (vsx_xxpermdi_<mode>_be): New define_expand with same
30056 implementation as previous version of vsx_xxpermdi_<mode>.
30058 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30060 PR tree-optimization/79327
30061 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30062 variable, its initialization and use.
30064 2017-02-17 Julia Koval <julia.koval@intel.com>
30066 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30067 (OPTION_MASK_ISA_PKU_UNSET): New.
30068 (ix86_handle_option): Handle -mrdpid.
30069 * config/i386/cpuid.h (bit_RDPID): New.
30070 * config/i386/driver-i386.c (host_detect_local_cpu):
30071 Detect RDPID feature.
30072 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30073 * config/i386/i386-c.c (ix86_target_macros_internal):
30075 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30076 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30077 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30078 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30079 * config/i386/i386.md (define_insn "rdpid"): New.
30080 * config/i386/i386.opt Add -mrdpid.
30081 * config/i386/immintrin.h (_rdpid_u32): New.
30083 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30085 PR rtl-optimization/79541
30086 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30087 instead of transforming it into USE.
30089 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30091 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30092 If HONOR_SNANS (SFmode) force the input to a register.
30093 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30094 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30095 an frsp or similar insn.
30097 2017-02-17 Martin Liska <mliska@suse.cz>
30099 PR rtl-optimization/79577
30100 * params.def (selsched-max-sched-times): Increase minimum to 1.
30102 2017-02-17 Martin Liska <mliska@suse.cz>
30104 PR rtl-optimization/79574
30105 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30107 2017-02-17 Martin Liska <mliska@suse.cz>
30109 PR tree-optimization/79529
30110 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30111 ssa_defined_default_def_p to handle cases which are implicitly
30113 * tree-ssa.c (ssa_defined_default_def_p): New function.
30114 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30115 which are implicitly defined.
30116 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30118 2017-02-17 Richard Biener <rguenther@suse.de>
30120 PR middle-end/79576
30121 * params.def (max-ssa-name-query-depth): Limit to 10.
30123 2017-02-17 Richard Biener <rguenther@suse.de>
30125 PR tree-optimization/79552
30126 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30129 2017-02-17 Richard Biener <rguenther@suse.de>
30132 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30134 2017-02-17 Marek Polacek <polacek@redhat.com>
30136 PR middle-end/79536
30137 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30138 (fold_negate_expr): New wrapper.
30140 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30142 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30143 Correct terminology and de-emphasize pre-standard behavior.
30145 2017-02-16 Alan Modra <amodra@gmail.com>
30147 PR rtl-optimization/79286
30148 * ira.c (def_dominates_uses): New function.
30149 (update_equiv_regs): Don't create an equivalence for insns that
30150 may trap where the register def does not dominate the use.
30152 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30154 PR rtl-optimization/78127
30155 * lra.c (lra): Call lra_eliminate before finish the loop after
30158 2017-02-16 Richard Biener <rguenther@suse.de>
30160 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30162 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30163 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30164 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30165 (isl_val_int_from_wi): New function.
30166 (extract_affine_gmp): Rename to ...
30167 (extract_affine_wi): ... this, take a widest_int.
30168 (extract_affine_int): Just wrap extract_affine_wi.
30169 (add_param_constraints): Use isl_val_int_from_wi.
30170 (add_loop_constraints): Likewise, and extract_affine_wi.
30172 2017-02-15 Jeff Law <law@redhat.com>
30174 PR middle-end/79521
30175 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30176 ira_init_register_move_cost_if_necessary.
30178 2017-02-15 Martin Sebor <msebor@redhat.com>
30180 PR middle-end/32003
30181 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30182 removed in a prior commit.
30184 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30186 PR tree-optimization/79347
30187 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30188 counters during peeling.
30190 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30192 * Makefile.in (site.exp): Remove "set ISLVER".
30194 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30197 * real.c (real_from_integer): Call real_convert even for decimal.
30199 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30202 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30204 2017-02-14 Andrew Pinski <apinski@cavium.com>
30206 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30207 cores and change the partno/implementer to be correct.
30208 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30209 the 'B" as the implementer.
30210 * config/aarch64/aarch64-tune.md: Regenerate.
30212 2017-02-14 Carl Love <cel@us.ibm.com>
30214 * config/rs6000/rs6000.c: Add case statement entry to make the
30215 xvcvuxdsp built-in argument unsigned.
30216 * config/rs6000/vsx.md: Fix the source and return operand types so they
30217 match the instruction definitions from the ISA document. Fix typo
30218 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30221 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30224 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30225 member early_clobber_alts.
30226 * lra-lives.c (reg_early_clobber_p): New.
30227 (process_bb_lives): Use it.
30228 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30229 (debug_operand_data): Initialize early_clobber_alts.
30230 (setup_operand_alternative): Set up early_clobber_alts.
30231 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30232 alternatives to new_insn_reg.
30233 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30235 (lra_update_insn_regno_info): Pass the new arg.
30237 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30239 PR middle-end/79505
30240 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30241 (new_oacc_loop_raw): Don't clear already cleared fields.
30244 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30245 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30246 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30248 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30251 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30253 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30256 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30257 the extra instruction to the right place to store 128-bit constant
30260 2017-02-14 Martin Sebor <msebor@redhat.com>
30262 PR middle-end/79448
30263 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30264 warning for strings of unknown length.
30266 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30268 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30270 2017-02-14 Jeff Law <law@redhat.com>
30273 * ira-costs.c (scan_one_insn): Initialize register move costs
30274 for pseudos seen in USE/CLOBBER insns.
30276 PR tree-optimization/79095
30277 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30278 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30279 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30280 if the operands are known to be not equal, then the resulting range
30282 (intersect_ranges): If the new range is ~[0,0] and the old range is
30283 wide, then prefer ~[0,0].
30284 * tree-vrp.c (overflow_comparison_p_1): New function.
30285 (overflow_comparison_p): New function.
30286 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30287 if NAME is used in an overflow test.
30288 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30289 overflow check that can be expressed as an equality test, then adjust
30290 ops to be that equality test.
30292 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30294 * config/s390/s390-builtin-types.def: Remove flags argument.
30295 * config/s390/s390.c (s390_init_builtins): Likewise.
30297 2017-02-14 Martin Liska <mliska@suse.cz>
30299 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30300 vector. Fix trailing white spaces.
30302 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30304 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30307 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30309 PR rtl-optimization/68664
30310 * config/arm/arm.c (arm_sched_can_speculate_insn):
30311 New function. Declare prototype.
30312 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30314 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30316 PR rtl-optimization/68664
30317 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30319 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30321 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30323 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30324 max skip bytes for function, loop and jump.
30326 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30328 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30329 ABS_EXPR for gimple dump.
30331 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30334 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30336 PR tree-optimization/79408
30337 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30338 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30339 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30340 also if rhs1 is INTEGER_CST.
30342 2017-02-14 Richard Biener <rguenther@suse.de>
30344 PR middle-end/79432
30345 * tree-into-ssa.c (insert_phi_nodes): When the function can
30346 have abnormal edges rewrite SSA names with broken use-def
30347 dominance out of SSA and register them for PHI insertion.
30349 2017-02-13 Martin Sebor <msebor@redhat.com>
30351 PR middle-end/79496
30352 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30353 clearing info.nowrite flag when snprintf size argument is a range.
30355 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30357 * cprop.c (cprop_jump): Add missing space in string literal.
30358 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30359 (get_constraint_for_component_ref): Likewise.
30360 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30361 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30362 * lra-constraints.c (process_alt_operands): Likewise.
30363 * ipa-inline.c (inline_small_functions): Likewise.
30364 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30365 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30366 * trans-mem.c (diagnose_tm_1_op): Likewise.
30367 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30368 (grid_parallel_clauses_gridifiable): Likewise.
30370 * config/nvptx/mkoffload.c (process): Add space in between
30373 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30374 "MOD4_SSE_REGS" and "ALL_REGS".
30376 * spellcheck.c (test_data): Add , in between "foo" and "food".
30378 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30381 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30382 boundary crossing check and subsequent code generation agree.
30384 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30386 * config/aarch64/aarch64.c (has_memory_op): Delete.
30387 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30390 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30392 PR rtl-optimization/79388
30393 PR rtl-optimization/79450
30394 * combine.c (distribute_notes): When removing TEM_INSN for which
30395 corresponding dest has last value recorded, invalidate that last
30398 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30400 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30401 of explicit '@'. Add missing assembly comment marker on branch costs
30404 2017-02-13 Nathan Sidwell <nathan@acm.org>
30406 * gengtype-lex.l (<in_struct>): Add '/'.
30408 2017-02-13 Martin Liska <mliska@suse.cz>
30411 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30413 2017-02-13 Richard Biener <rguenther@suse.de>
30415 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30417 * configure: Re-generate.
30418 * config.in: Likewise.
30419 * graphite-dependences.c: Simplify as if
30420 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30421 * graphite-isl-ast-to-gimple.c: Likewise.
30422 * graphite-optimize-isl.c: Likewise.
30423 * graphite-poly.c: Likewise.
30424 * graphite-sese-to-poly.c: Likewise.
30425 * graphite.h: Likewise.
30426 * toplev.c: Include isl/version.h and use isl_version () for
30427 printing the ISL version.
30428 * doc/install.texi: Update ISL requirement.
30430 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30432 * doc/standards.texi (Standards): Update reference to
30435 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30437 * doc/extend.texi (Named Address Spaces): sourceware.org now
30439 * doc/install.texi (Binaries): Ditto.
30442 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30444 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30445 standard terminology "stringize"/"stringizing" throughout.
30446 * doc/cppinternals.texi: Likewise.
30448 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30450 * doc/extend.texi: Fix some spelling mistakes and typos.
30451 * doc/invoke.texi: Likewise.
30453 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30456 * params.def (inline-min-speedup) Change from 10 to 8.
30458 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30460 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30463 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30466 * ipa-inline-analysis.c (get_minimal_bb): New function.
30467 (record_modified): Use it.
30468 (remap_edge_change_prob): Handle also ancestor functions.
30470 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30472 * doc/contrib.texi (Contributors): Remove broken link into
30473 the Mauve CVS repository.
30475 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30477 PR middle-end/79454
30478 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30479 result computation whenever lhs doesn't have vector mode, not
30480 just when it has BLKmode.
30482 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30484 * doc/makefile.texi (profiledbootstrap): Refer to the
30485 installation instructions only in textual form.
30487 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30490 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30492 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30494 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30495 (Specific): Update mingw-w64 reference.
30497 (Specific): Remove broken link to Renesas RX processor.
30499 2017-02-10 Richard Biener <rguenther@suse.de>
30501 * toplev.c (process_options): Do not mention obsolete graphite
30502 options when printing sorry message about missing graphite support.
30503 Mention -floop-nest-optimize.
30505 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30507 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30508 (vtst_p16): Likewise.
30509 (vtstq_p8): Likewise.
30510 (vtstq_p16): Likewise.
30512 (vtstq_p64): Likewise.
30513 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30514 (vset_lane_p64): New.
30515 (vsetq_lane_p64): New.
30517 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30519 PR tree-optimization/79411
30520 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30521 stmt operands are SSA_NAMEs used in abnormal phis.
30522 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30525 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30528 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30531 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30533 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30535 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30537 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30541 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30542 not arbitrary TREE_CONSTANT.
30545 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30546 "omp declare target link" attribute unless is_global_var.
30547 * omp-offload.c (find_link_var_op): Likewise.
30549 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30550 Chung-Lin Tang <cltang@codesourcery.com>
30552 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30554 (gimplify_adjust_omp_clauses): Don't delete TILE.
30555 (gimplify_omp_for): Deal with TILE.
30556 * internal-fn.c (expand_GOACC_TILE): New function.
30557 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30559 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30560 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30562 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30563 avoid DIV for outermost collapse var.
30564 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30565 Remove out of date comments, fix whitespace.
30566 * omp-general.c (omp_extract_for_data): Deal with tiling.
30567 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30568 adjust OLF_DIM_BASE value.
30569 (struct omp_for_data): Add tiling field.
30570 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30571 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30572 for auto loops. Remove default auto determining, moved to
30573 oacc_loop_fixed_partitions.
30574 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30575 stmts, add e_mask field.
30576 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30577 (oacc_thread_numbers): Use oacc_dim_call.
30578 (oacc_xform_tile): New.
30579 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30580 (finish_oacc_loop): Adjust for ifns vector.
30581 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30582 add case for GOACC_TILE fns.
30583 (oacc_loop_xform_loop): Delete.
30584 (oacc_loop_process): Iterate over call list directly, and add
30585 handling for GOACC_TILE fns.
30586 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30588 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30589 vector partitioning to outer loops. Assign 2 partitions to loops
30590 when available. Add TILE handling.
30591 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30592 (execite_oacc_device_lower): Process GOACC_TILE fns,
30593 ignore unknown specs.
30594 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30595 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30596 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30598 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30600 * configure.ac (ACX_BUGURL): Update.
30601 * configure: Regenerate.
30603 2017-02-09 Richard Biener <rguenther@suse.de>
30605 PR tree-optimization/69823
30606 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30607 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30609 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30611 * config/arc/arc-c.def: Add __NPS400__ definition.
30612 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30613 (TARGET_NPS400): Define.
30615 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30617 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30619 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30620 pointer, arch_info.
30621 (arc_cpu_types): Fill the arch_info field with a pointer into the
30622 arc_arch_types table.
30623 (arc_selected_cpu): Declare.
30624 * config/arc/arc.c (arc_selected_cpu): Make global.
30625 (arc_selected_arch): Delete.
30626 (arc_base_cpu): Delete.
30627 (arc_override_options): Remove references to deleted variables,
30628 update access to arch information.
30629 (ARC_OPT): Update access to arch information.
30630 (ARC_OPTX): Likewise.
30631 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30632 (TARGET_ARC600): Update access to arch information.
30633 (TARGET_ARC601): Likewise.
30634 (TARGET_ARC700): Likewise.
30635 (TARGET_EM): Likewise.
30636 (TARGET_HS): Likewise.
30637 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30640 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30643 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30644 condition/operands for integer GE/LE/GEU/LEU operations.
30646 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30648 PR translation/79397
30649 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30652 2017-02-08 Martin Jambor <mjambor@suse.cz>
30655 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30656 whether allocation happened.
30657 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30658 nothing was allocated.
30660 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30662 PR tree-optimization/79408
30663 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30664 constant, but SSA_NAME with a known integer range, use the minimum
30665 of that range instead of op1 to determine if modulo can be replaced
30666 with its first operand.
30668 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30670 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30672 2017-02-08 Richard Biener <rguenther@suse.de>
30674 PR tree-optimization/71824
30675 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30676 Check all loops contained in the merged region.
30678 2017-02-07 Andrew Pinski <apinski@cavium.com>
30680 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30682 2017-02-07 Andrew Pinski <apinski@cavium.com>
30684 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30685 (thunderxt88): Likewise.
30686 (thunderxt81): Disable LSE and change v8.1 to v8.
30687 (thunderxt83): Likewise.
30689 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30690 Richard Biener <rguenther@suse.de>
30692 PR middle-end/79399
30693 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30694 type from int to size_t.
30695 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30697 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30699 PR rtl-optimization/79386
30700 * cprop.c (bypass_conditional_jumps): Initialize
30701 bypass_last_basic_block already before splitting bbs after
30702 unconditional traps...
30703 (bypass_conditional_jumps): ... rather than here.
30706 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30707 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30708 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30709 fix -masm=intel patterns.
30711 2017-02-07 Richard Biener <rguenther@suse.de>
30713 PR tree-optimization/79256
30714 PR middle-end/79278
30715 * builtins.c (get_object_alignment_2): Use min_align_of_type
30716 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30717 and ADJUST_FIELD_ALIGN.
30719 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30721 * doc/tm.texi: Regenerate.
30722 * stor-layout.c (layout_decl): Adjust.
30723 (update_alignment_for_field): Likewise.
30724 (place_field): Likewise.
30725 (min_align_of_type): Likewise.
30726 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30727 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30728 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30729 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30730 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30731 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30732 * config/i386/i386.c (x86_field_alignment): Likewise.
30733 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30734 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30735 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30736 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30737 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30738 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30742 2017-01-30 Richard Biener <rguenther@suse.de>
30744 PR tree-optimization/79256
30745 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30746 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30749 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30751 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30752 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30753 builtins to SImode and emit a zero-extend, if necessary.
30755 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30757 * docs/invoke.texi (RISC-V Options): Alphabetize.
30759 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30761 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30764 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30766 * config/riscv/riscv.c: New file.
30767 * common/config/riscv/riscv-common.c: Likewise.
30768 * config.gcc: Likewise.
30769 * config/riscv/constraints.md: Likewise.
30770 * config/riscv/elf.h: Likewise.
30771 * config/riscv/generic.md: Likewise.
30772 * config/riscv/linux.h: Likewise.
30773 * config/riscv/multilib-generator: Likewise.
30774 * config/riscv/peephole.md: Likewise.
30775 * config/riscv/pic.md: Likewise.
30776 * config/riscv/predicates.md: Likewise.
30777 * config/riscv/riscv-builtins.c: Likewise.
30778 * config/riscv/riscv-c.c: Likewise.
30779 * config/riscv/riscv-ftypes.def: Likewise.
30780 * config/riscv/riscv-modes.def: Likewise.
30781 * config/riscv/riscv-opts.h: Likewise.
30782 * config/riscv/riscv-protos.h: Likewise.
30783 * config/riscv/riscv.h: Likewise.
30784 * config/riscv/riscv.md: Likewise.
30785 * config/riscv/riscv.opt: Likewise.
30786 * config/riscv/sync.md: Likewise.
30787 * config/riscv/t-elf-multilib: Likewise.
30788 * config/riscv/t-linux: Likewise.
30789 * config/riscv/t-linux-multilib: Likewise.
30790 * config/riscv/t-riscv: Likewise.
30791 * configure.ac: Likewise.
30792 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30793 Waterman as RISC-V maintainers.
30794 * doc/install.texi: Add RISC-V entries.
30795 * doc/invoke.texi: Add RISC-V options section.
30796 * doc/md.texi: Add RISC-V constraints section.
30797 * configure: Regenerated.
30799 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30802 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30803 false values to be constant vectors with all 0 or all 1 bits set.
30804 (vcondu<mode><mode>): Likewise.
30805 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30807 (fpmask_comparison_operator): Update comment.
30808 (vecint_comparison_operator): New predicate.
30809 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30810 vector conditionals when the true and false values are constant
30811 vectors with all 0 bits or all 1 bits set.
30813 2017-02-06 Martin Sebor <msebor@redhat.com>
30815 PR tree-optimization/79376
30816 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30818 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30820 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30821 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30822 to simplify split condition.
30824 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30826 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30827 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30830 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30832 PR rtl-optimization/68664
30833 * target.def (can_speculate_insn): New hook.
30834 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30835 * doc/tm.texi: Regenerate.
30836 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30837 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30838 (rs6000_sched_can_speculate_insn): New function.
30840 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30842 PR tree-optimization/79284
30843 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30844 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30845 vectorizable_mask_load_store, vectorizable_operation,
30846 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30847 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30848 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30849 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30850 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30851 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30852 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30853 is_gimple_assign (stmt). Replace another such test with
30854 is_gimple_assign (stmt).
30856 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30859 * config/avr/avr.c (rtl-iter.h): Include it.
30860 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30861 (avr_legitimate_combined_insn): ...and implementation.
30863 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30865 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30866 * config/s390/s390.c (s390_const_operand_ok)
30867 (s390_canonicalize_comparison, s390_extract_part)
30868 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30869 (s390_contiguous_bitmask_p, s390_rtx_costs)
30870 (legitimize_pic_address): Likewise.
30871 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30872 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30873 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30874 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30875 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30877 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30879 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30880 REGNO($0) == REGNO($1).
30882 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30884 * config/s390/linux.h(SIZE_TYPE): Add comment.
30886 2017-02-06 Julian Brown <julian@codesourcery.com>
30887 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30888 Virendra Pathak <virendra.pathak@broadcom.com>
30890 * config/aarch64/aarch64-cores.def: Change the scheduler
30892 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30893 * config/aarch64/thunderx2t99.md: New file.
30895 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30897 * doc/standards.texi (Go Language): Update link to language
30900 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30902 * tree-eh.c (lower_resx): Sanitize profile.
30903 (cleanup_empty_eh_move_lp): Likewise.
30905 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30908 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30910 * cfgloopmanip.h (loop_version): Update prototype.
30911 * modulo-sched.c (sms_schedule): Update call of loop_version.
30912 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30913 * tree-parloops.c (gen_parallel_loop): Likewise.
30914 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30915 * tree-ssa-loop-split.c (split_loop): Likewise.
30916 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30917 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30919 2017-02-05 Martin Liska <mliska@suse.cz>
30922 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30924 (print_operand_address): Initialize a struct to zero.
30926 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30928 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30929 garbage collector only in textual form.
30931 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30933 * doc/extend.texi (x86 specific memory model extensions for
30934 transactional memory): Simplify a phrase.
30936 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30939 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30940 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30941 (atomic_storedi_1): Likewise.
30943 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30945 PR tree-optimization/79338
30946 * tree-parloops.c (gather_scalar_reductions): Don't call
30947 vect_analyze_loop_form for loop->inner before destroying loop's
30950 2017-02-03 Martin Sebor <msebor@redhat.com>
30952 PR tree-optimization/79327
30953 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30954 when precision has resulted in leading zeros.
30955 (format_integer): Adjust the likely counter to assume an unknown
30956 argument that may be zero is non-zero.
30958 2017-02-03 Jason Merrill <jason@redhat.com>
30961 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30962 avoid copying non-taken branch.
30964 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30966 PR tree-optimization/79340
30967 * tree-vect-loop.c (vectorizable_reduction): Release
30968 vec_defs elements after safe_splicing them into other vectors.
30971 PR tree-optimization/79327
30972 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30973 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30975 (format_integer): Use wide_int_to_tree instead of build_int_cst
30976 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30977 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30978 of shortest and longest sequence.
30980 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30982 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30983 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30985 2017-02-03 Walter Lee <walt@tilera.com>
30988 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30989 after initial stackframe link reg save.
30990 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30992 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30995 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30996 wu for stxssp alternative.
30998 2017-02-03 Martin Sebor <msebor@redhat.com>
31000 PR tree-optimization/79352
31001 * gimple-fold.c (get_range_strlen): Add argument.
31002 (get_range_strlen): Change return type to bool.
31003 (get_maxval_strlen): Pass in a dummy argument.
31004 * gimple-fold.h (get_range_strlen): Change return type to bool.
31005 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31006 * tree.h (array_at_struct_end_p): Add argument.
31007 * tree.c (array_at_struct_end_p): Handle it.
31009 2017-02-03 Martin Liska <mliska@suse.cz>
31012 * multiple_target.c (create_dispatcher_calls): Redirect edge
31013 from a caller of a dispatcher.
31014 (expand_target_clones): Make the clones local.
31015 (ipa_target_clone): Do both target clones and resolvers.
31016 (ipa_dispatcher_calls): Remove the pass.
31017 (pass_dispatcher_calls::gate): Likewise.
31018 (make_pass_dispatcher_calls): Likewise.
31019 * passes.def (pass_target_clone): Put as very first IPA early
31022 2017-02-03 Martin Liska <mliska@suse.cz>
31024 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31025 in case of a function with ifunc attribute.
31027 2017-02-03 Martin Liska <mliska@suse.cz>
31029 * cgraph.c (cgraph_node::dump): Dump function version info.
31030 * symtab.c (symtab_node::dump_base): Add missing new line.
31032 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31034 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31035 (ifcombine_ifandif): Use it.
31037 2017-02-03 Martin Liska <mliska@suse.cz>
31039 * doc/invoke.texi: Document default value for
31040 use-after-scope-direct-emission-threshold.
31042 2017-02-03 Martin Liska <mliska@suse.cz>
31044 PR tree-optimization/79339
31045 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31046 (format_floating): Likewise.
31048 2017-02-03 Martin Liska <mliska@suse.cz>
31051 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31053 (ipa_node_params_t::remove): Likewise.
31054 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31055 initialization from removed ipa_node_params_t::insert.
31056 (ipa_node_params::~ipa_node_params): Move from removed
31057 ipa_node_params_t::release.
31058 * symbol-summary.h (symbol_summary::m_released): New member.
31059 Do not release a summary twice. Do not allow to call finalizer
31060 for types of a summary that live in GGC memory.
31062 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31064 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31067 2017-02-02 Martin Sebor <msebor@redhat.com>
31069 PR middle-end/79275
31070 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31071 (format_string): Tighten up the range of output for non-constant
31072 strings and correct the expected range for wide non-constant strings.
31074 2017-02-02 Martin Sebor <msebor@redhat.com>
31076 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31078 PR middle-end/32003
31079 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31081 (-fdump-tree-@var): Add to index and document how to come up
31082 with pass-specific option and dump file names.
31083 (-fdump-passes): Clarify where to look for output.
31085 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31087 PR middle-end/77445
31088 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31089 statistics of the analyzed path; allow threading for speed when
31090 any of BBs along the path are optimized for speed.
31092 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31094 PR middle-end/78468
31095 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31096 settings of the virtual registers.
31099 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31101 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31102 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31105 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31107 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31108 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31110 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31112 * config/s390/s390.md: Add missing comments with the expanded
31114 * config/s390/vector.md: Likewise.
31115 * config/s390/vx-builtins.md: Likewise.
31117 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31120 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31121 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31122 conditions on a single line.
31124 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31126 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31127 __S390_VX__ to __VX__.
31129 2017-02-01 Andrew Pinski <apinski@cavium.com>
31131 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31132 stmt_info to record_stmt_cost.
31133 (vect_get_known_peeling_cost): Pass stmt_info if known to
31135 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31136 cpu_vector_cost field into
31137 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31138 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31139 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31140 splitting of scalar_stmt_cost and vec_stmt_cost.
31141 (thunderx_vector_cost): Likewise.
31142 (cortexa57_vector_cost): LIkewise.
31143 (exynosm1_vector_cost): Likewise.
31144 (xgene1_vector_cost): Likewise.
31145 (thunderx2t99_vector_cost): Improve after the splitting of the two
31147 (aarch64_builtin_vectorization_cost): Update for the splitting of
31148 scalar_stmt_cost and vec_stmt_cost.
31150 2017-02-01 Torvald Riegel <triegel@redhat.com>
31151 Richard Henderson <rth@redhat.com>
31153 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31154 conditional on existance of a fast atomic load.
31155 * optabs-query.c (can_atomic_load_p): New function.
31156 * optabs-query.h (can_atomic_load_p): Declare it.
31157 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31158 no fast atomic load is available for the particular size of access.
31159 (expand_atomic_compare_and_swap): Likewise.
31160 (expand_atomic_load): Likewise.
31161 (expand_atomic_store): Likewise.
31162 (expand_atomic_fetch_op): Likewise.
31163 * testsuite/lib/target-supports.exp
31164 (check_effective_target_sync_int_128): Remove x86 because it provides
31165 no fast atomic load.
31166 (check_effective_target_sync_int_128_runtime): Likewise.
31168 2017-02-01 Richard Biener <rguenther@suse.de>
31170 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31171 (graphite_transform_loops): Provide opt-info for optimized nests.
31172 * tree-parloop.c (parallelize_loops): Provide opt-info for
31173 parallelized loops.
31175 2017-02-01 Richard Biener <rguenther@suse.de>
31177 PR middle-end/79315
31178 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31179 was not set before.
31181 2017-02-01 Richard Biener <rguenther@suse.de>
31183 PR tree-optimization/71824
31184 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31185 Verify the loops are valid in the merged SESE region.
31186 (scop_detection::can_represent_loop_1): Check analyzing the
31187 evolution of the number of iterations in the region succeeds.
31189 2017-01-31 Ian Lance Taylor <iant@golang.org>
31191 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31192 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31194 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31196 PR preprocessor/79210
31197 * input.c (get_substring_ranges_for_loc): Replace line_width
31198 assertion with error-handling.
31200 2017-01-31 Richard Biener <rguenther@suse.de>
31202 PR tree-optimization/77318
31203 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31204 (create_pw_aff_from_tree): Take loop parameter.
31205 (add_condition_to_pbb): Pass loop of the condition to
31206 create_pw_aff_from_tree.
31208 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31210 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31211 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31213 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31217 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31219 (convert_int_to_float128): Likewise.
31220 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31221 (convert_int_to_float128): Likewise.
31222 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31223 (UNSPEC_IEEE128_CONVERT): Likewise.
31224 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31225 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31226 Use local variables for IBM extended format.
31227 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31228 (fix_trunc<mode>si2_fprs): Likewise.
31229 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31230 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31231 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31232 to know that we can now have integers of all sizes in vector
31234 (fix<uns>_<mode>di2_hw): Likewise.
31235 (float<uns>_<mode>si2_hw): Likewise.
31236 (fix_<mode>si2_hw): Likewise.
31237 (fixuns_<mode>si2_hw): Likewise.
31238 (float<uns>_<mode>di2_hw): Likewise.
31239 (float_<mode>di2_hw): Likewise.
31240 (float_<mode>si2_hw): Likewise.
31241 (floatuns_<mode>di2_hw): Likewise.
31242 (floatuns_<mode>si2_hw): Likewise.
31243 (xscvqp<su>wz_<mode>): Delete, no longer used.
31244 (xscvqp<su>dz_<mode>): Likewise.
31245 (xscv<su>dqp_<mode>): Likewise.
31246 (ieee128_mfvsrd_64bit): Likewise.
31247 (ieee128_mfvsrd_32bit): Likewise.
31248 (ieee128_mfvsrwz): Likewise.
31249 (ieee128_mtvsrw): Likewise.
31250 (ieee128_mtvsrd_64bit): Likewise.
31251 (ieee128_mtvsrd_32bit): Likewise.
31253 2017-01-31 Martin Liska <mliska@suse.cz>
31256 * ipa-prop.c (ipa_free_all_node_params): Call release method
31257 instead of ~sumbol_summary to not to trigger double times
31260 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31262 PR tree-optimization/71691
31263 * bitmap.h (class auto_bitmap): New.
31264 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31265 is_maybe_undefined instead of ssa_undefined_value_p.
31267 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31269 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31270 __S390_ARCH_LEVEL__ to __ARCH__.
31272 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31274 PR tree-optimization/79267
31275 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31276 if should_remove_lhs_p is true.
31278 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31281 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31282 (add_alignment_attribute): New.
31283 (base_type_die): Add alignment attribute.
31284 (subrange_type_die): Likewise.
31285 (modified_type_die): Likewise.
31286 (gen_array_type_die): Likewise.
31287 (gen_descr_array_type_die: Likewise.
31288 (gen_enumeration_type_die): Likewise.
31289 (gen_subprogram_die): Likewise.
31290 (gen_variable_die): Likewise.
31291 (gen_field_die): Likewise.
31292 (gen_ptr_to_mbr_type_die): Likewise.
31293 (gen_struct_or_union_type_die): Likewise.
31294 (gen_subroutine_type_die): Likewise.
31295 (gen_typedef_die): Likewise.
31296 (base_type_cmp): Compare alignment attribute.
31298 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31301 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31302 (setb_unsigned) New pattern for setb with CCUNS.
31303 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31304 subfc./subfe sequence to avoid overflow problems. Generate a
31305 shorter sequence with cmpld/setb for power9.
31306 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31307 for generating subfc. instruction.
31308 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31309 now uses this instruction.
31311 2017-01-30 Ian Lance Taylor <iant@google.com>
31314 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31315 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31317 2017-01-30 Martin Sebor <msebor@redhat.com>
31319 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31320 Move constant to the right of a relational operator.
31321 (get_mpfr_format_length, format_character, format_string): Ditto.
31322 (should_warn_p, maybe_warn): Same.
31324 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31326 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31329 * asan.c (get_translation_unit_decl): Remove function.
31330 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31332 2017-01-30 Martin Liska <mliska@suse.cz>
31334 PR gcov-profile/79259
31335 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31336 -fprofile-generate.
31338 2017-01-30 Martin Liska <mliska@suse.cz>
31341 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31342 Initialize variables with NULL value.
31344 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31347 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31349 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31351 2017-01-30 Richard Biener <rguenther@suse.de>
31353 PR tree-optimization/79276
31354 * tree-vrp.c (process_assert_insertions): Properly adjust common
31355 when removing a duplicate.
31357 2017-01-30 Richard Biener <rguenther@suse.de>
31359 PR tree-optimization/79256
31360 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31361 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31363 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31365 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31368 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31369 ("*r<noxa>sbg_<mode>_sll_bitmask")
31370 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31371 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31372 Use contiguous_bitmask_nowrap_operand.
31374 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31377 * config/rs6000/altivec.h (vec_xl): Revise #define.
31378 (vec_xst): Likewise.
31380 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31382 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31384 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31386 PR rtl-optimization/79194
31387 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31388 traps before call to bypass_conditional_jumps.
31390 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31392 PR tree-optimization/71374
31393 * lra-constraints.c (check_conflict_input_operands): New.
31394 (match_reload): Use it.
31396 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31399 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31400 account to calculate conflict_set.
31402 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31404 PR rtl-optimization/78559
31405 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31406 other_insn in combine.
31408 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31410 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31411 uint16_type_node for BT_UINT16.
31413 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31415 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31416 "RTL Tests" to menu.
31417 (GIMPLE Tests): New node.
31418 (RTL Tests): New node.
31420 2017-01-27 Richard Biener <rguenther@suse.de>
31422 PR tree-optimization/79245
31423 * tree-loop-distribution.c (distribute_loop): Apply cost
31424 modeling also to detected patterns.
31426 2017-01-27 Richard Biener <rguenther@suse.de>
31428 PR tree-optimization/71433
31429 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31430 (compare_assert_loc): New function.
31431 (process_assert_insertions): Sort and optimize assert locations
31432 to remove duplicates and push down identical assertions on
31433 edges to their destination block.
31435 2017-01-27 Richard Biener <rguenther@suse.de>
31437 PR tree-optimization/79244
31438 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31439 out SSA names even if abnormal.
31441 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31443 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31444 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31445 instead of MPFR_RNDN.
31447 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31450 * arm.c (arm_option_override): Don't call build_target_option_node
31451 until after doing all option overrides.
31452 (arm_valid_target_attribute_tree): Likewise.
31454 2017-01-27 Martin Liska <mliska@suse.cz>
31456 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31457 for {cd}tors and C++ {cd}tors.
31459 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31461 * config/s390/s390.md ("*setmem_long_and")
31462 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31464 2017-01-26 Martin Sebor <msebor@redhat.com>
31466 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31469 2017-01-26 Martin Sebor <msebor@redhat.com>
31471 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31472 HAVE_DFmode before using XFmode or DFmode.
31473 (parse_directive): Avoid using the z length modifier to avoid
31474 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31476 PR middle-end/78703
31477 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31478 to accept adjustment as an array.
31479 (get_int_range): New function.
31480 (struct directive): Make width and prec arrays.
31481 (directive::set_width, directive::set_precision): Call get_int_range.
31482 (format_integer, format_floating): Handle width and precision ranges.
31483 (format_string, parse_directive): Same.
31485 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31488 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31489 comdat_type_p set, just clone them, but keep the children in the
31493 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31494 which have direct callers with -fvar-tracking-assignments enabled
31496 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31497 inside of type units.
31499 2017-01-26 Martin Sebor <msebor@redhat.com>
31501 PR middle-end/78703
31502 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31504 (struct format_result): Replace number_chars, number_chars_min,
31505 and number_chars_max with a single member of struct result_range.
31507 (format_result::operator+=): Adjust.
31508 (struct fmtresult): Remove bounded. Handle likely and unlikely
31510 (fmtresult::adjust_for_width_or_precision): New function.
31511 (fmtresult:type_max_digits): New function.
31512 (bytes_remaining): Handle likely and unlikely counters.
31513 (min_bytes_remaining): Remove.
31514 (format_percent): Simplify.
31515 (format_integer, format_floating): Set likely and unlikely counters.
31516 (get_string_length, format_character, format_string): Same.
31517 (format_plain, should_warn_p): New function.
31518 (maybe_warn): Call should_warn_p. Update diagnostic messages
31519 and handle those for all directives, including plain strings.
31520 (format_directive): Handle likely and unlikely counters.
31521 Remove unnecessary quoting from diagnostics. Add an informational
31523 (add_bytes): Remove.
31524 (pass_sprintf_length::compute_format_length): Simplify.
31525 (try_substitute_return_value): Handle likely and unlikely counters.
31527 2017-01-26 Carl Love <cel@us.ibm.com>
31529 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31530 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31532 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31535 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31536 endianess for subregs into account.
31537 * lra-constraints.c (lra_constraints): Do risky transformations
31538 always on the first iteration.
31539 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31540 last_call_used_reg_set.
31541 (process_bb_lives): Define and use last_call_used_reg_set.
31542 * lra.c (lra): Always continue after lra_constraints on the first
31545 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31547 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31549 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31551 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31553 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31554 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31555 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31556 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31557 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31558 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31559 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31560 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31561 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31563 2017-01-26 Marek Polacek <polacek@redhat.com>
31566 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31567 for the third operand.
31569 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31571 PR middle-end/79236
31572 * omp-low.c (struct omp_context): Add simt_stmt field.
31573 (scan_omp_for): Return omp_context *.
31574 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31575 context to the _simt_ SIMD stmt.
31576 (lower_omp_for): For combined SIMD with sibling _simt_
31577 SIMD, make sure to use the same decls in _looptemp_
31578 clauses as in the sibling.
31580 2017-01-26 David Sherwood <david.sherwood@arm.com>
31582 PR middle-end/79212
31583 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31586 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31589 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31590 emit fld b; fld a; if possible.
31592 * brig-builtins.def: Update copyright years.
31593 * config/arm/arm_acle_builtins.def: Update copyright years.
31595 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31598 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31599 constraint instead of o for the stxsd instruction.
31601 2017-01-25 Carl Love <cel@us.ibm.com>
31603 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31604 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31606 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31608 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31610 2017-01-25 Richard Biener <rguenther@suse.de>
31612 PR tree-optimization/69264
31613 * target.def (vector_alignment_reachable): Improve documentation.
31614 * doc/tm.texi: Regenerate.
31615 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31617 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31618 earlier changes with respect to TYPE_USER_ALIGN.
31619 (vector_alignment_reachable_p): Likewise. Improve dumping.
31621 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31624 * config/arm/arm.md (xordi3): Force constant operand into a register
31627 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31629 * doc/invoke.texi (-fstore-merging): Correct default optimization
31630 levels at which it is enabled.
31631 (-O): Move -fstore-merging from list to...
31634 2017-01-25 Richard Biener <rguenther@suse.de>
31637 * omp-expand.c: Include debug.h.
31638 (expand_omp_taskreg): Make sure to generate early debug before
31639 outlining anything from a function.
31640 (expand_omp_target): Likewise.
31641 (grid_expand_target_grid_body): Likewise.
31643 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31646 * asan.c (get_translation_unit_decl): New function.
31647 (asan_add_global): Extract modules file name from globals
31648 TRANSLATION_UNIT_DECL name.
31650 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31653 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31654 for long calls with APCS frame and VFP.
31656 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31658 * cfg.c (original_copy_tables_initialized_p): New function.
31659 * cfg.h (original_copy_tables_initialized_p): New decl.
31660 * cfgrtl.c (relink_block_chain): Guard the call to
31661 free_original_copy_tables with a call to
31662 original_copy_tables_initialized_p.
31663 * cgraph.h (symtab_node::native_rtl_p): New decl.
31664 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31665 (symtab_node::needed_p): Don't assert for early assembly output
31666 for __RTL functions.
31667 (cgraph_node::finalize_function): Set "force_output" for __RTL
31669 (cgraph_node::analyze): Bail out early for __RTL functions.
31670 (analyze_functions): Update assertion to support __RTL functions.
31671 (cgraph_node::expand): Bail out early for __RTL functions.
31672 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31674 * function.h (struct function): Update comment for field
31676 * gimple-expr.c: Include "tree-pass.h".
31677 (gimple_has_body_p): Return false for __RTL functions.
31678 * Makefile.in (OBJS): Add run-rtl-passes.o.
31679 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31681 (gcc::pass_manager::get_clean_slate): New accessor.
31682 * passes.c: Include "insn-addr.h".
31683 (should_skip_pass_p): Add logging. Update logic for running
31684 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31685 property-provider override so it is only done for gimple passes.
31687 (skip_pass): New function.
31688 (execute_one_pass): Call skip_pass when skipping passes.
31689 * read-md.c (md_reader::read_char): Support filtering
31690 the input to a subset of line numbers.
31691 (md_reader::md_reader): Initialize fields
31692 m_first_line and m_last_line.
31693 (md_reader::read_file_fragment): New function.
31694 * read-md.h (md_reader::read_file_fragment): New decl.
31695 (md_reader::m_first_line): New field.
31696 (md_reader::m_last_line): New field.
31697 * read-rtl-function.c (function_reader::create_function): Only
31698 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31699 curr_properties. Set DECL_INITIAL to a dummy block.
31700 (read_rtl_function_body_from_file_range): New function.
31701 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31703 * run-rtl-passes.c: New file.
31704 * run-rtl-passes.h: New file.
31706 2017-01-24 Jeff Law <law@redhat.com>
31708 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31711 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31713 PR tree-optimization/79159
31714 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31715 (record_nonwrapping_iv): Improve boundary using above function if no
31716 value range information.
31718 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31719 Martin Jambor <mjambor@suse.cz>
31721 * brig-builtins.def: New file.
31722 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31723 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31724 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31725 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31726 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31727 * builtin-types.def (BT_INT8): New.
31728 (BT_INT16): Likewise.
31729 (BT_UINT8): Likewise.
31730 (BT_UINT16): Likewise.
31731 (BT_FN_ULONG): Likewise.
31732 (BT_FN_UINT_INT): Likewise.
31733 (BT_FN_UINT_ULONG): Likewise.
31734 (BT_FN_UINT_LONG): Likewise.
31735 (BT_FN_UINT_PTR): Likewise.
31736 (BT_FN_ULONG_PTR): Likewise.
31737 (BT_FN_INT8_FLOAT): Likewise.
31738 (BT_FN_INT16_FLOAT): Likewise.
31739 (BT_FN_UINT32_FLOAT): Likewise.
31740 (BT_FN_UINT16_FLOAT): Likewise.
31741 (BT_FN_UINT8_FLOAT): Likewise.
31742 (BT_FN_UINT64_FLOAT): Likewise.
31743 (BT_FN_UINT16_UINT32): Likewise.
31744 (BT_FN_UINT32_UINT16): Likewise.
31745 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31746 (BT_FN_INT_PTR_INT): Likewise.
31747 (BT_FN_UINT_PTR_UINT): Likewise.
31748 (BT_FN_LONG_PTR_LONG): Likewise.
31749 (BT_FN_ULONG_PTR_ULONG): Likewise.
31750 (BT_FN_VOID_UINT64_UINT64): Likewise.
31751 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31752 (BT_FN_INT8_INT8_INT8): Likewise.
31753 (BT_FN_INT16_INT16_INT16): Likewise.
31754 (BT_FN_INT_INT_INT): Likewise.
31755 (BT_FN_UINT_FLOAT_UINT): Likewise.
31756 (BT_FN_FLOAT_UINT_UINT): Likewise.
31757 (BT_FN_ULONG_UINT_UINT): Likewise.
31758 (BT_FN_ULONG_UINT_PTR): Likewise.
31759 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31760 (BT_FN_UINT_UINT_UINT): Likewise.
31761 (BT_FN_VOID_UINT_PTR): Likewise.
31762 (BT_FN_UINT_UINT_PTR: Likewise.
31763 (BT_FN_UINT32_UINT64_PTR): Likewise.
31764 (BT_FN_INT_INT_UINT_UINT): Likewise.
31765 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31766 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31767 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31768 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31769 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31770 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31771 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31772 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31773 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31774 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31775 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31776 * doc/frontends.texi: List BRIG FE.
31777 * doc/install.texi (Testing): Add BRIG tesring requirements.
31778 * doc/invoke.texi (Overall Options): Mention BRIG.
31779 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31781 2017-01-24 Richard Biener <rguenther@suse.de>
31783 PR translation/79208
31784 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31786 2017-01-24 Martin Jambor <mjambor@suse.cz>
31789 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31790 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31791 and known_contexts.
31793 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31795 PR middle-end/79123
31796 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31797 casts from signed to unsigned really don't have a range.
31799 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31801 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31802 GMP_RNDx for compatiblity.
31804 2017-01-24 Martin Liska <mliska@suse.cz>
31807 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31808 that would prevent us to call alloca with -1 as argument.
31810 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31812 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31813 Avoid -Wformat-security warning.
31815 2017-01-23 Andrew Pinski <apinski@cavium.com>
31817 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31820 2017-01-23 Martin Sebor <msebor@redhat.com>
31822 PR middle-end/78703
31823 * gimple-ssa-sprintf.c (warn_level): New global.
31824 (format_integer): Use it here and throughout the rest of the file.
31825 Use the same switch to compute sign as base.
31826 (maybe_warn): New function.
31827 (format_directive): Factor out warnings into maybe_warn.
31828 Add debugging output. Use warn_level.
31829 (add_bytes): Use warn_level.
31830 (pass_sprintf_length::compute_format_length): Add debugging output.
31831 (try_substitute_return_value): Same.
31832 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31834 PR middle-end/78703
31835 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31836 (struct fmtresult, format_integer, format_floating): Adjust.
31837 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31838 (get_string_length, format_string,format_directive): Same.
31839 (pass_sprintf_length::compute_format_length): Same.
31840 (try_substitute_return_value): Simplify slightly.
31842 PR middle-end/78703
31843 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31844 (fmtresult::operator+=): Outlined.
31845 (struct fmtresult): Add ctors.
31846 (struct conversion_spec): Rename...
31847 (struct directive): ...to this. Add and remove data members.
31848 (directive::set_width, directive::set_precision): New functions.
31849 (format_percent): Use fmtresult ctor.
31850 (get_width_and_precision): Remove.
31851 (format_integer): Make naming changes. Avoid computing width and
31853 (format_floating): Same. Adjust indentation.
31854 (format_character, format_none): New functions.
31855 (format_string): Moved character handling to format_character.
31856 (format_directive): Remove arguments, change return type.
31857 (parse_directive): New function.
31858 (pass_sprintf_length::compute_format_length): Move directive
31859 parsing to parse_directive.
31861 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31863 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31864 (assign_assembler_name_if_needed): ... this.
31865 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31866 (assign_assembler_name_if_needed): ... this.
31867 (free_lang_data_in_cgraph): Adjust callers.
31868 * cgraphunit.c (cgraph_node::analyze): Likewise.
31869 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31872 2017-01-23 Richard Biener <rguenther@suse.de>
31874 PR tree-optimization/79088
31875 PR tree-optimization/79188
31876 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31877 resetting loop bounds after last path deletion. Reset loop
31878 bounds of the target loop, make code match the comments.
31879 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31880 Make sure loops need no fixups.
31882 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31884 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31885 exponent support with double type for first argument.
31886 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31887 type returned by __builtin_vec_extract_sig,
31888 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31889 functions from "vector int" to "vector unsigned int" or from
31890 "vector long long int" to "vector unsigned long long int".
31891 Changed type returned by __builtin_vec_extract_exp,
31892 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31893 functions from "vector int" to "vector unsigned int" or from
31894 "vector long long int" to "vector unsigned long long int".
31895 Changed return type of __builtin_vec_test_data_class,
31896 __builtin_vec_test_data_class_sp, and
31897 __builtin_vec_test_data_class_dp from "vector int" to
31898 "vector bool int" or from "vector long long int" to "vector bool
31899 long long int" and changed second argument type from "unsigned
31900 int" to "int". Added new overloaded function forms "vector float
31901 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31902 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31903 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31904 double, vector unsigned long long int)" and "vector double
31905 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31906 long int)". Changed return type of
31907 __builtin_scalar_test_data_class and
31908 __builtin_scalar_test_data_class_sp and
31909 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31910 int" and changed second argument from "unsigned int" to "int".
31911 Changed type returned by __builtin_scalar_test_neg,
31912 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31913 from "int" to "bool int". Added new overloaded function form
31914 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31915 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31916 exponent double-precision with floating point first argument.
31917 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31918 documentation of scalar_test_data_class, scalar_test_neg,
31919 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31920 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31921 vec_test_data_class built-in functions to reflect refinements in
31922 their type signatures.
31924 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31926 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31928 (aarch64_elf_asm_destructor): Likewise.
31930 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31932 PR rtl-optimization/78634
31933 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31934 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31935 * ifcvt.c (noce_try_cmove): Add missing cost check.
31937 PR rtl-optimization/71724
31938 * combine.c (if_then_else_cond): Look for situations where it is
31939 beneficial to undo the work of one of the recursive calls.
31941 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31943 PR tree-optimization/70754
31944 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31945 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31946 combined stmt before it if not NULL.
31947 (combine_chains): Process refs reversely and compute dominance point
31950 2017-01-23 Martin Liska <mliska@suse.cz>
31952 PR tree-optimization/79196
31953 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31954 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31956 (strlen_optimize_stmt): Call the renamed function.
31958 2017-01-23 Michael Matz <matz@suse.de>
31960 PR tree-optimization/78384
31961 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31963 2017-01-23 Richard Biener <rguenther@suse.de>
31965 PR tree-optimization/79186
31966 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31967 both incoming edges before moving an assert.
31969 2017-01-23 Martin Jambor <mjambor@suse.cz>
31971 * ipa-prop.c (load_from_param_1): Removed.
31972 (load_from_unmodified_param): Bits from load_from_param_1 put back
31974 (load_from_param): Removed.
31975 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31976 with stmt. Reverted back to use of load_from_unmodified_param.
31978 2017-01-23 Martin Jambor <mjambor@suse.cz>
31981 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31982 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31983 field a pointer to garbage collected vector, mark lattices and
31984 ipcp_orig_node with GTY((skip)).
31985 (ipa_get_param_count): Adjust to descriptors being a pointer.
31986 (ipa_get_param): Likewise.
31987 (ipa_get_type): Likewise.
31988 (ipa_get_param_move_cost): Likewise.
31989 (ipa_set_param_used): Likewise.
31990 (ipa_get_controlled_uses): Likewise.
31991 (ipa_set_controlled_uses): Likewise.
31992 (ipa_is_param_used): Likewise.
31993 (ipa_node_params_t): Move into garbage collector. New methods insert
31995 (ipa_node_params_sum): Annotate wth GTY(()).
31996 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31998 (ipa_load_from_parm_agg): Adjust declaration.
31999 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32000 * ipa-profile.c (ipa_profile): Likewise.
32001 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32002 (ipa_populate_param_decls): Make descriptors parameter garbage
32004 (ipa_dump_param): Adjust to descriptors being a pointer.
32005 (ipa_alloc_node_params): Likewise.
32006 (ipa_initialize_node_params): Likewise.
32007 (load_from_param_1): Make descriptors parameter garbage collected.
32008 (load_from_unmodified_param): Likewise.
32009 (load_from_param): Likewise.
32010 (ipa_load_from_parm_agg): Likewise.
32011 (ipa_node_params::~ipa_node_params): Removed.
32012 (ipa_free_all_node_params): Remove call to delete operator.
32013 (ipa_node_params_t::insert): New.
32014 (ipa_node_params_t::remove): Likewise.
32015 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32016 copy known_csts and known_contexts vectors.
32017 (ipa_read_node_info): Adjust to descriptors being a pointer.
32018 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32020 (ipcp_transform_function): Make descriptors variable garbage
32023 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32025 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32026 * config/i386/avx512dqintrin.h: Ditto.
32027 * config/i386/avx512fintrin.h: Ditto.
32028 * config/i386/i386.c: Handle new builtins.
32029 * config/i386/i386-builtin.def: Add new builtins.
32030 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32031 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32033 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32034 Martin Liska <mliska@suse.cz>
32036 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32037 * asan.c (asan_expand_poison_ifn): Support stores and use
32038 appropriate ASAN report function.
32039 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32040 * internal-fn.def (ASAN_POISON_USE): Declare.
32041 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32042 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32043 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32044 ASAN_POISON calls w/o LHS.
32045 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32046 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32047 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32048 * gimplify.c (asan_poison_variables): Add attribute
32049 use_after_scope_memory to variables that really needs to live
32051 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32052 having the attribute.
32054 2017-01-23 Martin Liska <mliska@suse.cz>
32056 * asan.c (create_asan_shadow_var): New function.
32057 (asan_expand_poison_ifn): Likewise.
32058 * asan.h (asan_expand_poison_ifn): New declaration.
32059 * internal-fn.c (expand_ASAN_POISON): Likewise.
32060 * internal-fn.def (ASAN_POISON): New builtin.
32061 * sanopt.c (pass_sanopt::execute): Expand
32062 asan_expand_poison_ifn.
32063 * tree-inline.c (copy_decl_for_dup_finish): Make function
32065 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32066 * tree-ssa.c (is_asan_mark_p): New function.
32067 (execute_update_addresses_taken): Rewrite local variables
32068 (identified just by use-after-scope as addressable) into SSA.
32070 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32072 * doc/install.texi (Specific): opensource.apple.com uses https
32073 now. Remove trailing slash.
32075 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32077 * README.Portability: Remove note on an Irix compatibility issue.
32079 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32081 * gcov.c (INCLUDE_ALGORITHM): Define.
32082 (INCLUDE_VECTOR): Define.
32083 No longer include <vector> and <algorithm> directly.
32085 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32087 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32089 * doc/invoke.texi (Code Gen Options): Ditto.
32091 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32094 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32096 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32098 rtl-optimization/79125
32099 * cprop.c (local_cprop_pass): Handle cases where we make an
32100 unconditional trap.
32102 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32106 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32107 read from, for big endian.
32109 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32111 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32112 register pauth builtins for LP64 only.
32114 2017-01-20 Marek Polacek <polacek@redhat.com>
32117 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32120 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32122 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32124 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32125 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32126 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32128 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32131 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32132 in big-endian mode when they are not a single duplicated value.
32134 2017-01-20 Richard Biener <rguenther@suse.de>
32136 * BASE-VER: Bump to 7.0.1.
32138 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32140 * omp-low.c (omplow_simd_context): New struct. Use it...
32141 (lower_rec_simd_input_clauses): ...here and...
32142 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32143 references to idx, lane, max_vf, is_simt.
32145 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32147 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32150 2017-01-20 Martin Jambor <mjambor@suse.cz>
32152 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32153 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32155 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32156 (GTFILES): Rename hsa.c to hsa-common.c.
32157 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32158 * hsa-dump.c: Likewise.
32159 * hsa-gen.c: Likewise.
32160 * hsa-regalloc.c: Likewise.
32161 * ipa-hsa.c: Likewise.
32162 * omp-expand.c: Likewise.
32163 * omp-low.c: Likewise.
32164 * toplev.c: Likewise.
32166 2017-01-20 Marek Polacek <polacek@redhat.com>
32169 * doc/invoke.texi: Document -Wduplicated-branches.
32170 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32171 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32172 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32173 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32174 return 0 only when not OEP_LEXICOGRAPHIC.
32175 (fold_build_cleanup_point_expr): Use the expression
32176 location when building CLEANUP_POINT_EXPR.
32177 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32178 * tree.c (add_expr): Handle error_mark_node.
32180 2017-01-20 Martin Liska <mliska@suse.cz>
32183 * tree-profile.c (init_ic_make_global_vars): Do not call
32185 (gimple_init_gcov_profiler): Likewise.
32187 2017-01-20 Martin Liska <mliska@suse.cz>
32190 * cgraph.h (maybe_create_reference): Remove argument and
32192 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32194 * ipa-cp.c (create_specialized_node): Likewise.
32195 * symtab.c (symtab_node::maybe_create_reference): Handle
32196 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32198 2017-01-20 Martin Liska <mliska@suse.cz>
32200 * read-rtl-function.c (function_reader::create_function): Use
32201 build_decl instread of build_decl_stat.
32203 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32205 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32206 * config/i386/avx512dqintrin.h: Ditto.
32207 * config/i386/avx512fintrin.h: Ditto.
32208 * config/i386/i386-builtin-types.def: Add new types.
32209 * config/i386/i386.c: Handle new types.
32210 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32211 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32212 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32213 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32214 (__builtin_ia32_kshiftridi): New.
32215 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32217 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32221 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32222 define to rs6000_init_stack_protect_guard.
32223 (rs6000_init_stack_protect_guard): New function.
32225 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32226 Yunqiang Su <yunqiang.su@imgtec.com>
32228 * config.gcc (supported_defaults): Add madd4.
32229 (with_madd4): Add validation.
32230 (all_defaults): Add madd4.
32231 * config/mips/mips.opt (mmadd4): New option.
32232 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32234 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32236 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32237 (ISA_HAS_FUSED_MADD4): Likewise.
32238 * doc/invoke.texi (-mmadd4): Document the new option.
32239 * doc/install.texi (--with-madd4): Document the new option.
32241 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32243 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32244 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32245 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32246 (aarch64_init_pauth_hint_builtins): New.
32247 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32248 (aarch64_expand_builtin): Expand new builtins.
32250 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32252 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32253 * combine-stack-adj.c (no_unhandled_cfa): Handle
32254 REG_CFA_TOGGLE_RA_MANGLE.
32255 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32256 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32257 info for return address signing.
32258 (aarch64_expand_epilogue): Likewise.
32260 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32262 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32263 * config/aarch64/aarch64-protos.h
32264 (aarch64_return_address_signing_enabled): New declaration.
32265 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32267 (aarch64_expand_prologue): Sign return address before it's pushed onto
32269 (aarch64_expand_epilogue): Authenticate return address fetched from
32271 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32272 (aarch64_attributes): New function attributes for "sign-return-address".
32273 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32274 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32275 ("*do_return"): Generate combined instructions according to key index.
32276 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32277 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32279 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32280 * config/aarch64/aarch64.opt (msign-return-address=): New.
32281 * doc/extend.texi (AArch64 Function Attributes): Documents
32282 "sign-return-address=".
32283 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32285 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32287 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32288 overall option summary.
32290 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32292 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32293 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32294 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32295 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32297 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32299 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32300 -mpower9-minmax by default for -mcpu=power9.
32301 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32302 128-bit floating point.
32304 2017-01-20 Alan Modra <amodra@gmail.com>
32306 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32307 optimizing for size.
32309 2017-01-20 Alan Modra <amodra@gmail.com>
32312 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32313 for strcmp and strncmp from corresponding builtin decl.
32315 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32317 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32318 instead of i386/rtems-64.h.
32319 * config/i386/rtems-64.h: Remove.
32321 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32325 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32327 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32329 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32331 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32332 Change int to HOST_WIDE_INT.
32333 * config/aarch64/aarch64-protos.h
32334 (aarch64_simd_gen_const_vector_dup): Likewise.
32335 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32337 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32339 * langhooks-def.h (lhd_type_for_size): New decl.
32340 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32341 * langhooks.c (lhd_type_for_size): New function, taken from
32344 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32346 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32347 define_bypass for CR latency.
32348 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32349 (power9-alu2): Add define_bypass for CR latency.
32351 (power9-mul): Update insn latency.
32352 (power9-mul-compare): Update insn latency, bypass latency and remove
32355 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32357 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32359 * config/aarch64/aarch64.md
32360 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32361 aarch64_nopcrelative_literal_loads.
32362 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32364 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32366 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32367 TARGET_LOONGSON_3A.
32368 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32370 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32373 * config.gcc (supported_defaults): Add lxc1-sxc1.
32374 (with_lxc1_sxc1): Add validation.
32375 (all_defaults): Add lxc1-sxc1.
32376 * config/mips/mips.opt (mlxc1-sxc1): New option.
32377 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32379 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32380 __mips_no_lxc1_sxc1.
32381 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32382 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32383 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32385 2017-01-19 Richard Biener <rguenther@suse.de>
32387 PR tree-optimization/72488
32388 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32389 sure to restore SSA info.
32390 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32392 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32394 PR rtl-optimization/79121
32395 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32396 of the inner type when shifting an extended value.
32398 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32401 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32402 interposable aliases.
32404 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32407 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32408 Use the evmergelohi instruction.
32409 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32410 (mov_sitf_e500_subreg8_2_be): Likewise.
32411 (mov_sitf_e500_subreg12_2_le): Likewise.
32412 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32413 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32414 (mov_sitf_e500_subreg8_2_le): Likewise.
32415 (mov_sitf_e500_subreg12_2_be): Likewise.
32417 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32419 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32420 attribute from vecsimple to vecperm.
32421 (altivec_vbpermq2): Likewise.
32423 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32426 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32428 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32429 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32430 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32431 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32432 case where N arg is SIZE_MAX.
32433 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32434 (cmpstrsi): Add pattern.
32436 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32438 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32439 __builtin_vec_revb builtins.
32440 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32441 built-in functions to support generation of the ISA 3.0 XXBR<x>
32442 vector byte reverse instructions.
32443 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32444 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32445 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32446 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32447 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32448 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32449 (P9V_BUILTIN_VEC_REVB): Likewise.
32450 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32451 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32452 (p9_xxbrq_v16qi): Likewise.
32453 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32454 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32455 (p9_xxbrh_v8hi): Likewise.
32456 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32457 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32458 vec_revb built-in functions.
32460 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32462 PR rtl-optimization/78952
32463 * config/i386/i386.md (any_extract): New code iterator.
32464 (*insvqi_2): Use any_extract for source operand.
32465 (*insvqi_3): Use any_shiftrt for source operand.
32467 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32469 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32471 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32473 2017-01-18 Matthias Klose <doko@ubuntu.com>
32475 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32477 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32479 * config/rs6000/altivec.h (vec_bperm): Change #define.
32480 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32481 (altivec_vbpermq2): New define_insn.
32482 (altivec_vbpermd): Likewise.
32483 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32484 function interface.
32485 (VBPERMD): Likewise.
32486 (VBPERM): New polymorphic function interface.
32487 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32488 Add entries for P9V_BUILTIN_VEC_VBPERM.
32489 * doc/extend.texi: Add interfaces for vec_bperm.
32491 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32493 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32494 first letter of error messages.
32495 (s390_resolve_overloaded_builtin): Likewise.
32496 * config/s390/s390.c (s390_expand_builtin): Likewise.
32497 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32498 (s390_valid_target_attribute_inner_p): Likewise.
32499 * config/s390/s390.md ("tabort"): Likewise.
32501 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32503 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32504 (ISA_AVOID_DIV_HILO): New macro.
32505 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32506 (ISA_HAS_DDIV): Likewise.
32508 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32510 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32512 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32514 * doc/invoke.texi (fabi-version): Spelling fix.
32516 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32519 * doc/invoke.texi (fabi-version): Mention mangling fix for
32522 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32525 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32527 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32530 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32531 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32533 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32535 (stack_protect_test): Ditto.
32536 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32537 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32539 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32540 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32541 -mstack-protector-guard-offset=.
32542 (RS/6000 and PowerPC Options): Ditto.
32544 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32546 * config/i386/i386.h (MASK_CLASS_P): New define.
32547 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32548 there are no registers from different register sets also when
32549 mask registers are used. Update function comment.
32550 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32551 to (*k/*r) and (*k/*km) alternatives.
32553 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32555 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32556 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32557 (EH_RETURN_HANDLER_RTX): New define.
32558 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32559 Force frame pointer in EH return functions.
32560 (aarch64_expand_epilogue): Add barrier for eh_return.
32561 (aarch64_final_eh_return_addr): Remove.
32562 (aarch64_eh_return_handler_rtx): New function.
32563 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32565 (aarch64_eh_return_handler_rtx): New prototype.
32567 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32569 * config/rs6000/altivec.h (vec_rlmi): New #define.
32570 (vec_vrlnm): Likewise.
32571 (vec_rlnm): Likewise.
32572 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32573 (UNSPEC_VRLNM): Likewise.
32574 (VIlong): New mode iterator.
32575 (altivec_vrl<VI_char>mi): New define_insn.
32576 (altivec_vrl<VI_char>nm): Likewise.
32577 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32579 (VRLDNM): Likewise.
32580 (RLNM): New polymorphic function entry.
32581 (VRLWMI): New monomorphic function entry.
32582 (VRLDMI): Likewise.
32583 (RLMI): New polymorphic function entry.
32584 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32585 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32586 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32589 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32592 * dwarf2out.c (field_byte_offset): Restore the
32593 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32594 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32595 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32598 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32601 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32603 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32606 * dwarf2out.c (add_data_member_location_attribute): For constant
32607 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32608 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32609 DW_AT_byte_size attributes.
32611 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32613 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32614 after forcing to constant memory when the code model is medium.
32616 2017-01-17 Julia Koval <julia.koval@intel.com>
32619 * config/i386/avx512fintrin.h
32620 (_mm512_i32gather_ps): Change __addr type to void const*.
32621 (_mm512_mask_i32gather_ps): Ditto.
32622 (_mm512_i32gather_pd): Ditto.
32623 (_mm512_mask_i32gather_pd): Ditto.
32624 (_mm512_i64gather_ps): Ditto.
32625 (_mm512_mask_i64gather_ps): Ditto.
32626 (_mm512_i64gather_pd): Ditto.
32627 (_mm512_mask_i64gather_pd): Ditto.
32628 (_mm512_i32gather_epi32): Ditto.
32629 (_mm512_mask_i32gather_epi32): Ditto.
32630 (_mm512_i32gather_epi64): Ditto.
32631 (_mm512_mask_i32gather_epi64): Ditto.
32632 (_mm512_i64gather_epi32): Ditto.
32633 (_mm512_mask_i64gather_epi32): Ditto.
32634 (_mm512_i64gather_epi64): Ditto.
32635 (_mm512_mask_i64gather_epi64): Ditto.
32636 (_mm512_i32scatter_ps): Change __addr type to void*.
32637 (_mm512_mask_i32scatter_ps): Ditto.
32638 (_mm512_i32scatter_pd): Ditto.
32639 (_mm512_mask_i32scatter_pd): Ditto.
32640 (_mm512_i64scatter_ps): Ditto.
32641 (_mm512_mask_i64scatter_ps): Ditto.
32642 (_mm512_i64scatter_pd): Ditto.
32643 (_mm512_mask_i64scatter_pd): Ditto.
32644 (_mm512_i32scatter_epi32): Ditto.
32645 (_mm512_mask_i32scatter_epi32): Ditto.
32646 (_mm512_i32scatter_epi64): Ditto.
32647 (_mm512_mask_i32scatter_epi64): Ditto.
32648 (_mm512_i64scatter_epi32): Ditto.
32649 (_mm512_mask_i64scatter_epi32): Ditto.
32650 (_mm512_i64scatter_epi64): Ditto.
32651 (_mm512_mask_i64scatter_epi64): Ditto.
32652 * config/i386/avx512pfintrin.h
32653 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32654 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32655 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32656 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32657 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32658 (_mm512_prefetch_i32scatter_ps): Ditto.
32659 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32660 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32661 (_mm512_prefetch_i64scatter_pd): Ditto.
32662 (_mm512_prefetch_i64scatter_ps): Ditto.
32663 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32664 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32665 * config/i386/avx512vlintrin.h
32666 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32667 (_mm_mmask_i32gather_ps): Ditto.
32668 (_mm256_mmask_i32gather_pd): Ditto.
32669 (_mm_mmask_i32gather_pd): Ditto.
32670 (_mm256_mmask_i64gather_ps): Ditto.
32671 (_mm_mmask_i64gather_ps): Ditto.
32672 (_mm256_mmask_i64gather_pd): Ditto.
32673 (_mm_mmask_i64gather_pd): Ditto.
32674 (_mm256_mmask_i32gather_epi32): Ditto.
32675 (_mm_mmask_i32gather_epi32): Ditto.
32676 (_mm256_mmask_i32gather_epi64): Ditto.
32677 (_mm_mmask_i32gather_epi64): Ditto.
32678 (_mm256_mmask_i64gather_epi32): Ditto.
32679 (_mm_mmask_i64gather_epi32): Ditto.
32680 (_mm256_mmask_i64gather_epi64): Ditto.
32681 (_mm_mmask_i64gather_epi64): Ditto.
32682 (_mm256_i32scatter_ps): Change __addr type to void*.
32683 (_mm256_mask_i32scatter_ps): Ditto.
32684 (_mm_i32scatter_ps): Ditto.
32685 (_mm_mask_i32scatter_ps): Ditto.
32686 (_mm256_i32scatter_pd): Ditto.
32687 (_mm256_mask_i32scatter_pd): Ditto.
32688 (_mm_i32scatter_pd): Ditto.
32689 (_mm_mask_i32scatter_pd): Ditto.
32690 (_mm256_i64scatter_ps): Ditto.
32691 (_mm256_mask_i64scatter_ps): Ditto.
32692 (_mm_i64scatter_ps): Ditto.
32693 (_mm_mask_i64scatter_ps): Ditto.
32694 (_mm256_i64scatter_pd): Ditto.
32695 (_mm256_mask_i64scatter_pd): Ditto.
32696 (_mm_i64scatter_pd): Ditto.
32697 (_mm_mask_i64scatter_pd): Ditto.
32698 (_mm256_i32scatter_epi32): Ditto.
32699 (_mm256_mask_i32scatter_epi32): Ditto.
32700 (_mm_i32scatter_epi32): Ditto.
32701 (_mm_mask_i32scatter_epi32): Ditto.
32702 (_mm256_i32scatter_epi64): Ditto.
32703 (_mm256_mask_i32scatter_epi64): Ditto.
32704 (_mm_i32scatter_epi64): Ditto.
32705 (_mm_mask_i32scatter_epi64): Ditto.
32706 (_mm256_i64scatter_epi32): Ditto.
32707 (_mm256_mask_i64scatter_epi32): Ditto.
32708 (_mm_i64scatter_epi32): Ditto.
32709 (_mm_mask_i64scatter_epi32): Ditto.
32710 (_mm256_i64scatter_epi64): Ditto.
32711 (_mm256_mask_i64scatter_epi64): Ditto.
32712 (_mm_i64scatter_epi64): Ditto.
32713 (_mm_mask_i64scatter_epi64): Ditto.
32714 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32715 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32716 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32717 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32718 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32719 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32720 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32721 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32722 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32723 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32724 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32725 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32726 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32727 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32728 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32729 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32730 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32731 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32732 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32733 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32734 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32735 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32736 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32737 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32738 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32739 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32740 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32741 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32742 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32743 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32744 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32745 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32746 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32747 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32748 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32749 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32750 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32751 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32752 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32753 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32754 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32755 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32756 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32757 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32758 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32759 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32760 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32761 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32762 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32763 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32764 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32765 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32766 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32767 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32768 definitions accordingly.
32770 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32771 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32774 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32777 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32780 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32781 hard regs for inner regno.
32783 2017-01-17 Martin Liska <mliska@suse.cz>
32786 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32787 assumption and add comment.
32789 2017-01-17 Nathan Sidwell <nathan@acm.org>
32791 * ipa-visibility.c (localize_node): New function, broken out of ...
32792 (function_and_variable_visibility): ... here. Call it.
32794 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32796 PR middle-end/77445
32797 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32798 correctly set frequency of oudgoing edge.
32799 (duplicate_thread_path): Fix profile updating.
32801 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32804 * configure.ac: Add GCC_BASE_VER.
32805 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32806 version from BASE-VER file.
32807 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32808 (gcc.o): Depend on $(BASEVER).
32809 * common.opt (dumpfullversion): New option.
32810 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32811 * doc/invoke.texi: Document -dumpfullversion.
32812 * doc/install.texi: Document --with-gcc-major-version-only.
32813 * configure: Regenerated.
32815 2017-01-17 Richard Biener <rguenther@suse.de>
32817 PR tree-optimization/71433
32818 * tree-vrp.c (register_new_assert_for): Merge same asserts
32819 on all incoming edges.
32820 (process_assert_insertions_for): Handle insertions at the
32823 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32825 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32826 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32828 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32831 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32834 2017-01-17 Alan Modra <amodra@gmail.com>
32837 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32838 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32839 symbolic stack limit when pic.
32841 2017-01-16 Martin Sebor <msebor@redhat.com>
32843 PR tree-optimization/78608
32844 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32846 2017-01-16 Jeff Law <law@redhat.com>
32849 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32850 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32851 for several include directories that may be relative to sysroot.
32852 * config/i386/x-mingw32 (gplus_includedir): Define.
32853 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32854 (native_system_includedir): Likewise.
32855 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32856 override if TARGET_SYSTEM_ROOT is defined.
32857 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32859 PR tree-optimization/79090
32860 PR tree-optimization/33562
32861 PR tree-optimization/61912
32862 PR tree-optimization/77485
32863 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32864 and computed trims into the dump file.
32866 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32868 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32870 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32873 * gimplify.c (gimplify_init_constructor): If want_value and
32874 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32878 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32879 sequence. Formatting fixes.
32880 (doloop_optimize): Formatting fixes.
32883 * gcc.c (debug_level_greater_than_spec_func): New function.
32884 (static_spec_functions): Add debug-level-gt spec function.
32885 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32887 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32888 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32889 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32890 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32891 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32892 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32894 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32896 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32897 QImode fixups to general and mask registers only.
32899 2017-01-16 Carl Love <cel@us.ibm.com>
32901 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32902 for built-in functions
32903 vector signed char vec_nabs (vector signed char)
32904 vector signed short vec_nabs (vector signed short)
32905 vector signed int vec_nabs (vector signed int)
32906 vector signed long long vec_nabs (vector signed long long)
32907 vector float vec_nabs (vector float)
32908 vector double vec_nabs (vector double)
32909 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32911 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32912 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32913 * doc/extend.texi: Update the documentation file for the new built-in
32916 2017-01-16 Martin Sebor <msebor@redhat.com>
32918 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32921 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32923 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32924 UNSPEC_VSX__XXSPLTD to require special splat handling.
32926 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32929 * system.h: Poison strndup.
32931 2017-01-16 Alan Modra <amodra@gmail.com>
32934 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32937 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32939 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32941 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32943 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32944 call recog here. Assert that INSN_CODE (insn) is non-negative.
32946 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32949 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32951 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32952 in the currently scheduled RTL fragment.
32954 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32956 PR rtl-optimization/78751
32957 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32960 2017-01-14 Jeff Law <law@redhat.com>
32962 PR tree-optimization/79090
32963 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32964 variable length stores.
32965 (compute_trims): Delete dead assignment to *trim_tail.
32966 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32969 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32971 PR rtl-optimization/78626
32972 PR rtl-optimization/78727
32973 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32974 of a block, and split such blocks after everything else is finished.
32976 2017-01-14 Alan Modra <amodra@gmail.com>
32979 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32980 target legitimate_combined_insn.
32981 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32982 (rs6000_legitimate_combined_insn): New function.
32983 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32985 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32986 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32987 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32989 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32991 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32993 2017-01-13 Jeff Law <law@redhat.com>
32995 PR tree-optimization/33562
32996 PR tree-optimization/61912
32997 PR tree-optimization/77485
32998 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33000 (delete_dead_assignment): Likewise.
33001 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33002 statement to delete_dead_call and delete_dead_assignment.
33004 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33007 * substring-locations.c (format_warning_va): Strengthen case 1 so
33008 that both endpoints of the substring must be within the format
33009 range for just the substring to be printed.
33011 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33013 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33014 * config/i386/i386.c (ix86_target_string): Add missing options
33015 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33016 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33017 flags_other and ix86_target_other to flags2_other. Display unknown
33019 (ix86_valid_target_attribute_inner_p): Add missing options and
33020 reorder options by implied ISAs, as in ix86_target_string.
33022 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33024 * hash-table.h (hash_table::too_empty_p): New function.
33025 (hash_table::expand): Use it.
33026 (hash_table::traverse): Likewise.
33027 (hash_table::empty_slot): Use sizeof (value_type) instead of
33028 sizeof (PTR) to convert bytes to elements. Shrink the table
33029 if the current size is excessive for the current number of
33032 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33034 * ira-costs.c (record_reg_classes): Break from the inner loop
33035 early once alt_fail is known to be true. Update outer loop
33036 handling accordingly.
33038 2017-01-13 Jeff Law <law@redhat.com>
33040 * tree-ssa-dse.c (decrement_count): New function.
33041 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33042 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33043 when we know the partially dead statement is a mem* function.
33045 PR tree-optimization/61912
33046 PR tree-optimization/77485
33047 * tree-ssa-dse.c: Include expr.h.
33048 (maybe_trim_constructor_store): New function.
33049 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33051 PR tree-optimization/33562
33052 PR tree-optimization/61912
33053 PR tree-optimization/77485
33054 * doc/invoke.texi: Document new dse-max-object-size param.
33055 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33056 * tree-ssa-dse.c: Include params.h.
33057 (dse_store_status): New enum.
33058 (initialize_ao_ref_for_dse): New, partially extracted from
33060 (valid_ao_ref_for_dse, normalize_ref): New.
33061 (setup_live_bytes_from_ref, compute_trims): Likewise.
33062 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33063 (maybe_trim_partially_dead_store): Likewise.
33064 (maybe_trim_complex_store): Likewise.
33065 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33066 Track what bytes live from the original store. Return tri-state
33067 for dead, partially dead or live.
33068 (dse_dom_walker): Add constructor, destructor and new private members.
33069 (delete_dead_call, delete_dead_assignment): New extracted from
33071 (dse_optimize_stmt): Make a member of dse_dom_walker.
33072 Use initialize_ao_ref_for_dse.
33074 PR tree-optimization/33562
33075 PR tree-optimization/61912
33076 PR tree-optimization/77485
33077 * sbitmap.h (bitmap_count_bits): Prototype.
33078 (bitmap_clear_range, bitmap_set_range): Likewise.
33079 * sbitmap.c (bitmap_clear_range): New function.
33080 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33082 2017-01-13 Martin Liska <mliska@suse.cz>
33085 * function.c (set_cfun): Add new argument force.
33086 * function.h (set_cfun): Likewise.
33087 * ipa-inline-transform.c (inline_call): Use the function when
33088 strict alising from is dropped for function we inline to.
33090 2017-01-13 Richard Biener <rguenther@suse.de>
33092 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33093 for dumping GIMPLE INTEGER_CSTs.
33095 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33097 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33098 to 201112L since C++17.
33100 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33103 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33104 if -fsanitize=kernel-address is present.
33106 2017-01-13 Richard Biener <rguenther@suse.de>
33108 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33109 as _Literal ( type ) number in case usual suffixes do not
33110 preserve all information.
33112 2017-01-13 Richard Biener <rguenther@suse.de>
33114 PR tree-optimization/77283
33115 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33116 and ssa-iterators.h.
33117 (is_feasible_trace): Implement a cost model based on joiner
33120 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33123 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33124 char or short to __float128/_Float128 directly.
33126 2017-01-12 Martin Sebor <msebor@redhat.com>
33128 to -Wformat-overflow.
33129 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33130 (min_bytes_remaining): Same.
33131 (get_string_length): Same.
33132 (format_string): Same.
33133 (format_directive): Same.
33135 (pass_sprintf_length::handle_gimple_call): Same.
33137 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33139 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33140 info.nowrite calls with no lhs that can't throw. Return bool
33141 whether gsi_remove has been called or not.
33142 (pass_sprintf_length::handle_gimple_call): Return bool whether
33143 try_substitute_return_value called gsi_remove. Formatting fix.
33144 (pass_sprintf_length::execute): Don't use gsi_remove if
33145 handle_gimple_call returned true.
33148 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33149 be removed due to side-effects, don't remove following barrier nor
33150 turn the successor edge into fallthru edge.
33152 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33155 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33156 element-reversing loads and stores as not swappable.
33158 2017-01-12 Nathan Sidwell <nathan@acm.org>
33159 Nicolai Stange <nicstange@gmail.com>
33161 * combine.c (try_combine): Don't ignore result of overlap checking
33162 loop. Combine overlap & asm check into single loop.
33164 2017-01-12 Richard Biener <rguenther@suse.de>
33166 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33167 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33169 2017-01-12 Richard Biener <rguenther@suse.de>
33171 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33172 and TS_TARGET_OPTION directly derive from TS_BASE.
33173 * tree-core.h (tree_optimization_option): Derive from tree_base.
33174 (tree_target_option): Likewise.
33176 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33178 * config/i386/i386.c (memory_address_length): Increase len
33179 only when rip_relative_addr_p returns false.
33181 2017-01-11 Julia Koval <julia.koval@intel.com>
33183 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33184 (OPTION_MASK_ISA_SGX_SET): New.
33185 (ix86_handle_option): Handle OPT_msgx.
33186 * config.gcc: Added sgxintrin.h.
33187 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33188 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33189 * config/i386/i386.c (ix86_target_string): Add -msgx.
33191 (ix86_option_override_internal): Handle new options.
33192 (ix86_valid_target_attribute_inner_p): Add sgx.
33193 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33194 * config/i386/i386.opt: Add msgx.
33195 * config/i386/sgxintrin.h: New file.
33196 * config/i386/x86intrin.h: Add sgxintrin.h.
33198 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33201 * fold-const.c (maybe_nonzero_address): Return 1 for function
33203 (tree_single_nonzero_warnv_p): Don't handle function local objects
33207 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33210 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33213 * opts.c: Include "spellcheck.h"
33214 (struct string_fragment): New struct.
33215 (struct edit_distance_traits<const string_fragment &>): New
33217 (get_closest_sanitizer_option): New function.
33218 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33220 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33222 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33224 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33225 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33226 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33227 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33228 for initial die_offset if dwarf_split_debug_info.
33229 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33230 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33232 (output_skeleton_debug_sections): Formatting fix. Use
33233 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33234 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33236 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33238 * config/arm/cortex-a53.md: Add bypasses for
33239 cortex_a53_r2f_cvt.
33240 (cortex_a53_r2f): Only use for transfers.
33241 (cortex_a53_f2r): Likewise.
33242 (cortex_a53_r2f_cvt): Add reservation for conversions.
33243 (cortex_a53_f2r_cvt): Likewise.
33245 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33247 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33248 to all inlined functions, change static to extern.
33250 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33253 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33255 (arm_assemble_integer): Likewise.
33257 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33259 * config.gcc: Use new awk script to check CPU, FPU and architecture
33260 parameters for --with-... options.
33261 * config/arm/parsecpu.awk: New file
33262 * config/arm/arm-cpus.in: New file.
33263 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33265 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33267 * config/arm/t-arm: Update dependency rules.
33268 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33269 of processing .def files.
33270 * config/arm/genopt.sh: Deleted.
33271 * config/arm/gentune.sh: Deleted.
33272 * config/arm/arm-cores.def: Deleted.
33273 * config/arm/arm-arches.def: Deleted.
33274 * config/arm/arm-fpus.def: Deleted.
33275 * config/arm/arm-tune.md: Regenerated.
33276 * config/arm/arm-tables.opt: Regenerated.
33277 * config/arm/arm-cpu.h: New generated file.
33278 * config/arm/arm-cpu-data.h: New generated file.
33279 * config/arm/arm-cpu-cdata.h: New generated file.
33281 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33284 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33286 (input_varpool_node): Unpack dynamically_initialized bit.
33288 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33290 PR rtl-optimization/79032
33291 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33292 the alignment of the adjusted memory reference against that of MODE,
33293 instead of the alignment of the original memory reference.
33295 2017-01-11 Martin Jambor <mjambor@suse.cz>
33297 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33299 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33300 decorated functions.
33302 2017-01-11 Richard Biener <rguenther@suse.de>
33304 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33305 set range/nonnull info for PHI results. Do not set it on
33306 stmts marked for removal.
33308 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33310 * expr.c (store_field): In the bitfield case, fetch the return value
33311 from the registers before applying a single big-endian adjustment.
33312 Always do a final load for a BLKmode value not larger than a word.
33314 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33317 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33318 that we correctly handle column numbers greater than
33319 LINE_MAP_MAX_COLUMN_NUMBER.
33321 2017-01-10 Martin Sebor <msebor@redhat.com>
33323 PR middle-end/78245
33324 * gimple-ssa-sprintf.c (get_destination_size): Call
33325 {init,fini}object_sizes.
33326 * tree-object-size.c (addr_object_size): Adjust.
33327 (pass_through_call): Adjust.
33328 (pass_object_sizes::execute): Adjust.
33329 * tree-object-size.h (fini_object_sizes): Declare.
33331 2017-01-10 Martin Sebor <msebor@redhat.com>
33333 PR tree-optimization/78775
33334 * builtins.c (get_size_range): Move...
33335 * calls.c: ...to here.
33336 (alloc_max_size): Accept zero argument.
33337 (operand_signed_p): Remove.
33338 (maybe_warn_alloc_args_overflow): Call get_size_range.
33339 * calls.h (get_size_range): Declare.
33341 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33343 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33344 from TI's devices.csv file as of September 2016.
33345 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33347 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33349 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33350 * doc/invoke.texi: Likewise.
33351 * doc/md.texi: Likewise.
33352 * doc/objc.texi: Likewise.
33354 2017-01-10 Joshua Conner <joshconner@google.com>
33356 * config/arm/fuchsia-elf.h: New file.
33357 * config/fuchsia.h: New file.
33358 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33359 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33361 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33363 2016-01-10 Richard Biener <rguenther@suse.de>
33365 PR tree-optimization/79034
33366 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33367 Propagate out degenerate PHIs in the joiner.
33369 2017-01-10 Martin Liska <mliska@suse.cz>
33371 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33372 (sort_congruence_classes_by_decl_uid): Likewise.
33373 (sort_congruence_class_groups_by_decl_uid): Likewise.
33374 (sem_item_optimizer::merge_classes): Sort class, groups in these
33375 classes and members in the groups by DECL_UID of declarations.
33376 This would make merge operations stable.
33378 2017-01-10 Martin Liska <mliska@suse.cz>
33380 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33381 usage of m_classes_vec.
33382 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33383 (sem_item_optimizer::get_group_by_hash): Likewise.
33384 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33385 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33386 (sem_item_optimizer::verify_classes): Likewise.
33387 (sem_item_optimizer::process_cong_reduction): Likewise.
33388 (sem_item_optimizer::dump_cong_classes): Likewise.
33389 (sem_item_optimizer::merge_classes): Likewise.
33390 * ipa-icf.h (congruence_class_hash): Rename from
33391 congruence_class_group_hash. Remove declaration of m_classes_vec.
33393 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33395 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33396 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33397 * config.gcc: Add avx512vpopcntdqintrin.h.
33398 * config/i386/avx512vpopcntdqintrin.h: New.
33399 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33400 * config/i386/i386-builtin-types.def: Add new types.
33401 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33402 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33403 __builtin_ia32_vpopcountq_v8di_mask): New.
33404 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33405 __AVX512VPOPCNTDQ__.
33406 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33407 (PTA_AVX512VPOPCNTDQ): Define.
33408 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33409 TARGET_AVX512VPOPCNTDQ_P): Define.
33410 * config/i386/i386.opt: Add mavx512vpopcntdq.
33411 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33412 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33414 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33416 PR middle-end/77484
33417 * predict.def (PRED_CALL): Set to 67.
33419 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33421 * expr.c (store_field): In the bitfield case, if the value comes from
33422 a function call and is of an aggregate type returned in registers, do
33423 not modify the field mode; extract the value in all cases if the mode
33424 is BLKmode and the size is not larger than a word.
33426 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33429 * config/i386/cpuid.h: Fix undefined behavior.
33431 2017-01-04 Jeff Law <law@redhat.com>
33433 PR tree-optimization/79007
33434 PR tree-optimization/67955
33435 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33436 conservative for pt.null when flag_non_call_exceptions is on.
33438 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33440 PR translation/79019
33441 PR translation/79020
33442 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33443 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33444 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33446 * config/avr/avr.opt (maccumulate-args): Likewise.
33447 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33448 * common.opt (freport-bug): Likewise.
33449 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33450 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33451 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33452 translatable string.
33453 * config/i386/i386.c (function_value_32): Likewise.
33454 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33455 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33457 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33458 * common/config/msp430/msp430-common.c (msp430_handle_option):
33460 * symtab.c (symtab_node::verify_base): Likewise.
33461 * opts.c (set_debug_level): Likewise.
33462 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33463 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33464 missing whitespace to translatable strings.
33465 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33466 * config/sh/superh.h: Likewise.
33467 * config/i386/xopintrin.h: Likewise.
33468 * config/i386/znver1.md: Likewise.
33469 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33470 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33471 * double-int.h (struct double_int): Likewise.
33472 * double-int.c (div_and_round_double): Likewise.
33473 * wide-int.cc: Likewise.
33474 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33475 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33476 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33477 (crossjumps_occurred): ... this.
33478 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33481 PR tree-optimization/78899
33482 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33483 returning bool return struct loop *, NULL for failure and the new
33485 (versionable_outer_loop_p): Don't version outer loop if it has
33486 dont_vectorized bit set.
33487 (tree_if_conversion): When versioning outer loop, ensure
33488 tree_if_conversion is performed also on the inner loop of the
33489 non-vectorizable outer loop copy.
33490 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33491 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33492 prevent vectorization of it.
33493 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33494 the outer loop vectorization of the non-scalar version is attempted
33495 before vectorization of the inner loop in scalar version. If
33496 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33497 vectorization of its inner loop.
33498 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33499 has 2 inner loops, rename also on edges from bb whose single pred
33500 is outer_loop->header. Fix typo in function comment.
33502 2017-01-09 Martin Sebor <msebor@redhat.com>
33505 * asan.c (asan_emit_stack_protection): Increase local buffer size
33506 to avoid snprintf truncation warning.
33508 2017-01-09 Andrew Pinski <apinski@cavium.com>
33510 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33511 to reference thunderx2t99 for the tuning structure
33512 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33514 (thunderx2t99_extra_costs): This.
33515 * config/aarch64/aarch64-tune.md: Regenerate.
33516 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33517 (vulcan_addrcost_table): This.
33518 (vulcan_regmove_cost): Rename to ...
33519 (thunderx2t99_regmove_cost): This.
33520 (vulcan_vector_cost): Rename to ...
33521 (thunderx2t99_vector_cost): this.
33522 (vulcan_branch_cost): Rename to ...
33523 (thunderx2t99_branch_cost): This.
33524 (vulcan_tunings): Rename to ...
33525 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33526 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33528 2017-01-09 Martin Jambor <mjambor@suse.cz>
33532 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33533 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33534 (propagate_vr_accross_jump_function): Use the above function for all
33535 value range computations for pass-through jump functions and type
33536 converasion from explicit value range values.
33537 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33538 parameters from TYPE_ARG_TYPES.
33539 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33540 (ipa_write_node_info): Stream type of the actual argument.
33541 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33543 2017-01-09 Martin Liska <mliska@suse.cz>
33546 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33547 (lookup_compiler): Do not show error message with have_E.
33549 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33551 PR tree-optimization/78938
33552 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33553 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33554 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33555 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33558 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33560 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33563 2017-01-09 Richard Biener <rguenther@suse.de>
33565 PR tree-optimization/78997
33566 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33567 name condition properly.
33569 2017-01-09 Richard Biener <rguenther@suse.de>
33572 * dwarf2out.c (is_cxx): New overload with context.
33573 (is_naming_typedef_decl): Use it.
33575 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33577 * invoke.texi (Option Summary): Correct spacing in option lists
33578 and add line breaks to fix over-long lines.
33580 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33582 PR middle-end/17660
33584 * extend.texi (Common Variable Attributes): Add xref to GCC
33585 Internals manual to explain mode attribute keywords.
33587 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33590 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33591 and Preprocessor Options.
33592 (Options for Linking): Document -pthread here....
33593 (RS/6000 and PowerPC Options): ...not here.
33594 (Solaris 2 Options): ...or here.
33595 * doc/cppopts.texi: Document -pthread.
33597 2017-01-08 Martin Sebor <msebor@redhat.com>
33599 PR middle-end/77708
33600 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33601 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33602 New member functions.
33603 (format_directive): Used them.
33605 (pass_sprintf_length::handle_gimple_call): Same.
33606 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33607 to avoid truncation for any argument.
33608 (extract_affine_mul): Same.
33609 * tree.c (get_file_function_name): Same.
33611 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33613 PR middle-end/77484
33614 * predict.def (PRED_INDIR_CALL): Set to 86.
33616 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33618 PR preprocessor/54124
33619 * doc/cppopts.texi: Reformat -d subtable to list the full name
33620 of the options. Add cross-reference to the docs for the general
33621 compiler -d options.
33622 * doc/invoke.texi (Developer Options): Add cross-reference to the
33623 preprocessor-specific -d option documentation.
33625 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33627 PR preprocessor/13498
33628 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33629 redudant material, and reflect new command-line options.
33630 (System Headers): Likewise.
33632 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33634 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33635 -isystem, and -idirafter. Copy-edit.
33636 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33637 default for -ftrack-macro-expansion. Delete obsolete and
33638 badly-formatted implementation details about -fdebug-cpp output.
33639 * doc/cppwarnopts.texi: Copy-edit.
33641 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33644 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33645 that the transition from a max line width >= 1<<10 to narrower
33646 lines works correctly.
33648 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33650 * doc/options.texi (PerFunction): New.
33651 * opt-functions.awk (switch_flags): Map both Optimization and
33652 PerFunction to CL_OPTIMIZATION.
33653 * opth-gen.awk: Test for PerFunction flag along with
33655 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33656 it only when the latter is present. Skip those that don't in
33657 the hash function generator.
33658 * common.opt (fvar-tracking): Mark as PerFunction instead of
33660 (fvar-tracking-assignments): Likewise.
33661 (fvar-tracking-assignments-toggle): Likewise.
33662 (fvar-tracking-uninit): Likewise.
33664 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33666 PR translation/79018
33667 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33670 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33673 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33674 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33675 TARGET_LONG_JUMP_TABLE_OFFSETS.
33676 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33677 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33678 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33679 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33680 * config/m68k/m68k.md (tablejump expander): Likewise.
33681 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33682 TARGET_LONG_JUMP_TABLE_OFFSETS.
33683 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33684 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33686 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33687 David Holsgrove <david.holsgrove@xilinx.com>
33689 * common/config/microblaze/microblaze-common.c
33690 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33691 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33693 * config/microblaze/microblaze.c (microblaze_must_save_register)
33694 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33696 (microblaze_eh_return): New function.
33697 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33698 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33699 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33700 * config/microblaze/microblaze.md (eh_return): New pattern.
33702 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33704 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33705 GCC_DIAGNOSTIC_STRINGIFY): Define.
33707 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33709 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33711 * config/arm/arm.md (<mcrr>): New.
33713 * config/arm/arm.c (arm_arch5te): New.
33714 (arm_option_override): Set arm_arch5te.
33715 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33717 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33718 (arm_mcrr_qualifiers): ... this. New.
33719 (MRRC_QUALIFIERS): Define to...
33720 (arm_mrrc_qualifiers): ... this. New.
33721 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33723 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33724 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33725 (MRRCI, mrrc, MRRC): New.
33726 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33727 VUNSPEC_MRRC2): New.
33729 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33731 * config/arm/arm.md (<mcr>): New.
33733 * config/arm/arm.c (arm_coproc_builtin_available): Add
33734 support for mcr, mrc, mcr2 and mrc2.
33735 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33736 (arm_mcr_qualifiers): ... this. New.
33737 (MRC_QUALIFIERS): Define to ...
33738 (arm_mrc_qualifiers): ... this. New.
33739 (MCR_QUALIFIERS): Define to ...
33740 (arm_mcr_qualifiers): ... this. New.
33741 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33743 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33744 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33745 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33746 VUNSPEC_MRC2): New.
33748 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33750 * config/arm/arm.md (*ldc): New.
33754 * config/arm/arm.c (arm_coproc_builtin_available): Add
33755 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33756 (arm_coproc_ldc_stc_legitimate_address): New.
33757 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33758 'qualifier_const_pointer'.
33759 (LDC_QUALIFIERS): Define to...
33760 (arm_ldc_qualifiers): ... this. New.
33761 (STC_QUALIFIERS): Define to...
33762 (arm_stc_qualifiers): ... this. New.
33763 * config/arm/arm-protos.h
33764 (arm_coproc_ldc_stc_legitimate_address): New.
33765 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33766 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33767 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33768 stc2, stcl, stc2l): New.
33769 * config/arm/constraints.md (Uz): New.
33770 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33771 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33772 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33773 VUNSPEC_STC2L): New.
33775 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33777 * config/arm/arm.md (<cdp>): New.
33778 * config/arm/arm.c (neon_const_bounds): Rename this ...
33779 (arm_const_bounds): ... this.
33780 (arm_coproc_builtin_available): New.
33781 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33782 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33783 (CDP_QUALIFIERS): Define to...
33784 (arm_cdp_qualifiers): ... this. New.
33786 (arm_expand_builtin_args): Add case for 6 arguments.
33787 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33788 (arm_const_bounds): ... this.
33789 (arm_coproc_builtin_available): New.
33790 * config/arm/arm_acle.h (__arm_cdp): New.
33792 * config/arm/arm_acle_builtins.def (cdp): New.
33794 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33795 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33796 'arm_const_bounds'.
33797 * config/arm/types.md (coproc): New.
33798 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33799 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33800 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33801 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33803 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33805 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33806 (UBINOP_QUALIFIERS): New.
33808 (acle_builtin_data): New. Change comment.
33809 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33810 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33811 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33812 arm_acle_builtins.def.
33813 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33814 (arm_init_acle_builtins): New.
33815 (CRC32_BUILTIN): Remove.
33816 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33817 crc32cb, crc32ch and crc32cw.
33818 (arm_init_crc32_builtins): Remove.
33819 (arm_init_builtins): Use arm_init_acle_builtins rather
33820 than arm_init_crc32_builtins.
33821 (arm_expand_acle_builtin): New.
33822 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33823 * config/arm/arm_acle_builtins.def: New.
33825 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33827 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33828 (arm_builtin_datum): ... this.
33829 (arm_init_neon_builtin): Rename to ...
33830 (arm_init_builtin): ... this. Add a new parameters PREFIX
33831 and USE_SIG_IN_NAME.
33832 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33833 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33834 'arm_builtin_datum'.
33835 (arm_init_vfp_builtins): Likewise.
33836 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33837 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33838 (arm_expand_neon_args): Rename to ...
33839 (arm_expand_builtin_args): ... this. Rename builtin_arg
33840 enum values and differentiate between ARG_BUILTIN_MEMORY
33841 and ARG_BUILTIN_NEON_MEMORY.
33842 (arm_expand_neon_builtin_1): Rename to ...
33843 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33844 values, arm_expand_builtin_args and add bool parameter NEON.
33845 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33846 (arm_expand_vfp_builtin): Likewise.
33847 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33849 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33851 PR middle-end/77484
33852 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33853 * predict.c (tree_estimate_probability_bb): Reverse direction of
33854 polymorphic call predictor.
33856 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33858 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33859 (determine_pass_name_match): ...this new function and...
33860 (should_skip_pass_p): ...this new function.
33862 2017-01-06 Nathan Sidwell <nathan@acm.org>
33864 * ipa-visibility.c (function_and_variable_visibility): Reformat
33865 comments and long lines. Remove extrneous if.
33866 * symtab.c (symtab_node::make_decl_local): Fix code format.
33867 (symtab_node::set_section_for_node): Fix comment typo.
33869 2017-01-06 Martin Liska <mliska@suse.cz>
33872 * lra-constraints.c: Rename invariant to lra_invariant.
33873 * predict.c (set_even_probabilities): Initialize e to NULL.
33875 2017-01-05 Martin Sebor <msebor@redhat.com>
33877 PR tree-optimization/78910
33878 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33879 (format_integer): Correct off-by-one error in the handling
33880 of precision with negative numbers in signed conversions..
33882 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33884 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33886 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33888 PR tree-optimization/71016
33889 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33890 factor_out_conditional_conversion. Formatting fix.
33891 (factor_out_conditional_conversion): Add cond_stmt argument.
33892 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33893 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33896 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33898 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33899 read-rtl-function.o, and selftest-rtl.o.
33900 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33901 (selftest::aarch64_test_loading_full_dump): New function.
33902 (selftest::aarch64_run_selftests): New function.
33903 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33904 selftest::aarch64_run_selftests.
33905 * config/i386/i386.c
33906 (selftest::ix86_test_loading_dump_fragment_1): New function.
33907 (selftest::ix86_test_loading_call_insn): New function.
33908 (selftest::ix86_test_loading_full_dump): New function.
33909 (selftest::ix86_test_loading_unspec): New function.
33910 (selftest::ix86_run_selftests): Call the new functions.
33911 * emit-rtl.c (maybe_set_max_label_num): New function.
33912 * emit-rtl.h (maybe_set_max_label_num): New decl.
33913 * function.c (instantiate_decls): Guard call to
33914 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33915 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33917 * gensupport.c (gen_reader::gen_reader): Pass "false"
33918 for new "compact" param of rtx_reader.
33919 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33920 rather than an empty string for NULL strings.
33921 * read-md.c: Potentially include config.h rather than bconfig.h.
33922 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33923 (have_error): New global, copied from errors.c.
33924 (md_reader::read_name): Rename to...
33925 (md_reader::read_name_1): ...this, adding "out_loc" param,
33926 and converting "missing name or number" to returning false, rather
33928 (md_reader::read_name): Reimplement in terms of read_name_1.
33929 (md_reader::read_name_or_nil): New function.
33930 (md_reader::read_string): Handle "(nil)" by returning NULL.
33931 (md_reader::md_reader): Add new param "compact".
33932 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33933 (md_reader::read_file): New method.
33934 * read-md.h (md_reader::md_reader): Add new param "compact".
33935 (md_reader::read_file): New method.
33936 (md_reader::is_compact): New accessor.
33937 (md_reader::read_name): Convert return type from void to file_location.
33938 (md_reader::read_name_or_nil): New decl.
33939 (md_reader::read_name_1): New decl.
33940 (md_reader::m_compact): New field.
33941 (noop_reader::noop_reader): Pass "false" for new "compact" param
33943 (rtx_reader::rtx_reader): Add new "compact" param.
33944 (rtx_reader::read_rtx_operand): Make virtual and convert return
33945 type from void to rtx.
33946 (rtx_reader::read_until): New decl.
33947 (rtx_reader::handle_any_trailing_information): New virtual function.
33948 (rtx_reader::postprocess): New virtual function.
33949 (rtx_reader::finalize_string): New virtual function.
33950 (rtx_reader::m_in_call_function_usage): New field.
33951 (rtx_reader::m_reuse_rtx_by_id): New field.
33952 * read-rtl-function.c: New file.
33953 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33954 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33955 (selftest::verify_three_block_rtl_cfg): New decl.
33956 * read-rtl-function.h: New file.
33957 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33958 For host, include function.h, memmodel.h, and emit-rtl.h.
33959 (one_time_initialization): New function.
33960 (struct compact_insn_name): New struct.
33961 (compact_insn_names): New array.
33962 (find_code): Handle insn codes in compact dumps.
33963 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33964 (bind_subst_iter_and_attr): Likewise.
33965 (add_condition_to_string): Likewise.
33966 (add_condition_to_rtx): Likewise.
33967 (apply_attribute_uses): Likewise.
33968 (add_current_iterators): Likewise.
33969 (apply_iterators): Likewise.
33970 (initialize_iterators): Guard usage of apply_subst_iterator with
33971 #ifdef GENERATOR_FILE.
33972 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33973 (md_reader::read_mapping): Likewise.
33974 (add_define_attr_for_define_subst): Likewise.
33975 (add_define_subst_attr): Likewise.
33976 (read_subst_mapping): Likewise.
33977 (check_code_iterator): Likewise.
33978 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33980 (one_time_initialization): New function.
33981 (rtx_reader::read_until): New method.
33982 (read_flags): New function.
33983 (parse_reg_note_name): New function.
33984 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33985 Handle reuse_rtx ids.
33986 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33987 Add parsing support for RTL dumps, mirroring the special-cases in
33988 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33989 values, and calling handle_any_trailing_information.
33990 (rtx_reader::read_rtx_operand): Convert return type from void
33991 to rtx, returning return_rtx. Handle case 'e'. Call
33992 finalize_string on XSTR and XTMPL fields.
33993 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33994 "(nil)" values were omitted. Call the postprocess vfunc on the
33996 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33997 class ctor. Initialize m_in_call_function_usage. Call
33998 one_time_initialization.
33999 * rtl-tests.c (selftest::test_uncond_jump): Call
34000 set_new_first_and_last_insn.
34001 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34002 * selftest-rtl.c: New file.
34003 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34004 (selftest::get_insn_by_uid): New decl.
34005 * selftest-run-tests.c (selftest::run_tests): Call
34006 read_rtl_function_c_tests.
34007 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34008 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34011 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34013 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34014 operands in a special way. Assert that pos+len <= mode precision.
34016 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34018 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34019 3 argument Alias with unlimited for the negative form.
34020 (fno-vect-cost-model): Removed.
34022 2017-01-05 Martin Liska <mliska@suse.cz>
34024 * hsa-gen.c (gen_hsa_divmod): New function.
34025 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34027 2017-01-05 Martin Liska <mliska@suse.cz>
34030 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34033 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34035 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34036 small constant length operands.
34038 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34040 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34041 between loop iterations.
34043 2017-01-05 Martin Liska <mliska@suse.cz>
34046 * gimplify.c (gimplify_decl_expr): Compare to
34047 asan_poisoned_variables instread of checking flags.
34048 (gimplify_target_expr): Likewise.
34049 (gimplify_expr): Likewise.
34050 (gimplify_function_tree): Conditionally initialize
34051 asan_poisoned_variables.
34053 2017-01-04 Jeff Law <law@redhat.com>
34055 PR tree-optimizatin/78812
34056 * rtl.h (contains_mem_rtx_p): Prototype.
34057 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34058 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34059 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34060 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34061 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34063 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34065 * input.c (assert_char_at_range): Default-initialize actual_range.
34067 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34069 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34070 to match the caller.
34072 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34074 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34075 insns after final jump in test to emit dummy move.
34077 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34079 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34080 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34082 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34084 * multiple_target.c (create_dispatcher_calls): Init e_next.
34085 * tree-ssa-loop-split.c (split_loop): Init border.
34086 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34089 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34094 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34095 (altivec_register_operand): Do not return true if the operand
34096 contains a SUBREG mixing SImode and SFmode.
34097 (vsx_register_operand): Likewise.
34098 (vsx_reg_sfsubreg_ok): New predicate.
34099 (vfloat_operand): Do not return true if the operand contains a
34100 SUBREG mixing SImode and SFmode.
34101 (vint_operand): Likewise.
34102 (vlogical_operand): Likewise.
34103 (gpc_reg_operand): Likewise.
34104 (int_reg_operand): Likewise.
34105 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34106 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34107 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34109 (rs6000_emit_move_si_sf_subreg): New helper function.
34110 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34111 fixup SUBREGs involving SImode and SFmode.
34112 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34113 numbers for the new peephole2 optimization.
34114 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34115 the GLIBC math library that do AND/IOR/XOR operations on single
34116 precision floating point.
34117 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34118 target macros to say whether we need to avoid SUBREGs mixing
34120 (TARGET_ALLOW_SF_SUBREG): Likewise.
34121 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34122 (UNSPEC_SI_FROM_SF): Likewise.
34123 (iorxor): Change spacing.
34124 (and_ior_xor): New iterator for AND, IOR, and XOR.
34125 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34126 (movdi_from_sf_zero_ext): Likewise.
34127 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34128 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34129 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34130 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34131 (fms<mode>4): Likewise.
34132 (fnma<mode>4): Likewise.
34133 (fnms<mode>4): Likewise.
34134 (nfma<mode>4): Likewise.
34135 (nfms<mode>4): Likewise.
34137 2017-01-04 Marek Polacek <polacek@redhat.com>
34140 * doc/invoke.texi: Document -Wpointer-compare.
34142 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34144 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34147 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34148 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34151 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34154 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34155 documentation of the powerpc_popcntb_ok attribute.
34156 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34157 code to issue warning messages if a requested CPU configuration is
34158 not supported by the binary (assembler and loader) toolchain.
34159 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34160 made to define a built-in function that has been disabled.
34161 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34162 made to define a built-in function that has been disabled.
34163 (altivec_init_builtins): Add comment explaining why definition
34164 of the DST built-in functions is not preceded by an assertion
34165 check. Add assertions to prevent ICE if attempts are made to
34166 define an altivec predicate or an abs* built-in function that has
34168 (htm_init_builtins): Add comment explaining why definition of the
34169 htm built-in functions is not preceded by an assertion check.
34171 2017-01-04 Jeff Law <law@redhat.com>
34173 PR tree-optimizatin/67955
34174 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34175 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34176 the points-to solution does not include pt_null. Use DECL_PT_UID
34179 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34181 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34182 Use gen_int_mode instead of gen_lopwart for const_int operands.
34184 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34186 PR tree-optimization/71563
34187 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34188 out of range value - has low bits known to be zero.
34190 2017-01-04 Alan Modra <amodra@gmail.com>
34192 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34193 * configure: Regenerate.
34194 * config.in: Regenerate.
34196 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34199 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34200 a substring of the message, but strcmp with the whole message. Ifdef
34201 ENABLE_NLS, translate the message first using dgettext.
34203 2017-01-03 Jeff Law <law@redhat.com>
34205 PR tree-optimizatin/78856
34206 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34207 (mark_threaded_blocks): Remove code to truncate thread paths that
34208 cross multiple loop headers. Instead invalidate the cached loop
34209 iteration information and handle case of a thread path walking
34210 into an irreducible region.
34212 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34215 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34216 assertions. Add support for doing the signbit if the IEEE 128-bit
34217 floating point value is in a GPR.
34218 * config/rs6000/rs6000.md (Fsignbit): Delete.
34219 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34220 Update the length attribute if the value is in a GPR.
34221 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34222 the sign or zero extension instruction, since the value is always 0/1.
34223 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34226 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34227 extracting SImode to a GPR register so that we can generate a
34228 store, limit the vector to be in a traditional Altivec register
34229 for the vextuwrx instruction.
34231 2017-01-03 Ian Lance Taylor <iant@google.com>
34233 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34235 2017-01-03 Martin Sebor <msebor@redhat.com>
34237 PR tree-optimization/78696
34238 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34239 precision. Use MPFR for %f for greater fidelity. Correct handling
34241 (pass_sprintf_length::compute_format_length): Set width and precision
34242 specified by asrerisk to void_node for vararg functions.
34243 (try_substitute_return_value): Adjust dump output.
34245 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34247 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34249 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34251 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34252 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34253 -mlra/-mno-lra was passed to the compiler.
34255 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34257 PR rtl-optimization/65618
34258 * emit-rtl.c (try_split): Move initialization of "before" and
34259 "after" to just before the call to emit_insn_after_setloc.
34261 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34263 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34265 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34267 * dwarf2out.c (gen_enumeration_type_die): When
34268 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34270 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34272 PR tree-optimization/78965
34273 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34274 Change first argument from const call_info & to call_info &. For %n
34275 set info.nowrite to false.
34277 PR middle-end/78901
34278 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34279 possibly throwing calls.
34281 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34282 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34283 and fns handling, rather than in a separate case SSA_NAME.
34285 2017-01-02 Jeff Law <law@redhat.com>
34287 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34288 fixes for first_period and second_period variables.
34290 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34293 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34294 (*insvqi_1): New insn pattern.
34295 (*insvqi_1_mem_rex64): Ditto.
34296 (*insvqi_2): Ditto.
34297 (*insvqi_3): Rename from *insvqi.
34299 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34301 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34303 * doc/cfg.texi (Edges): Remove reference to Java.
34304 (Maintaining the CFG): Ditto.
34306 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34308 PR middle-end/77674
34309 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34310 transparent aliases.
34312 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34314 PR middle-end/77484
34315 * predict.def (PRED_CALL): Update hitrate.
34316 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34317 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34318 into direct/indirect/polymorphic variants.
34320 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34322 Update copyright years.
34324 * gcc.c (process_command): Update copyright notice dates.
34325 * gcov-dump.c (print_version): Ditto.
34326 * gcov.c (print_version): Ditto.
34327 * gcov-tool.c (print_version): Ditto.
34328 * gengtype.c (create_file): Ditto.
34329 * doc/cpp.texi: Bump @copying's copyright year.
34330 * doc/cppinternals.texi: Ditto.
34331 * doc/gcc.texi: Ditto.
34332 * doc/gccint.texi: Ditto.
34333 * doc/gcov.texi: Ditto.
34334 * doc/install.texi: Ditto.
34335 * doc/invoke.texi: Ditto.
34337 Copyright (C) 2017 Free Software Foundation, Inc.
34339 Copying and distribution of this file, with or without modification,
34340 are permitted in any medium without royalty provided the copyright
34341 notice and this notice are preserved.