1 2017-11-10 Jakub Jelinek <jakub@redhat.com>
4 * gimple-ssa-store-merging.c
5 (pass_store_merging::terminate_all_aliasing_chains): For
6 gimple_store_p stmts also call refs_output_dependent_p.
8 PR rtl-optimization/82913
9 * compare-elim.c (try_merge_compare): Punt if def_insn is not
12 2017-11-09 Jeff Law <law@redhat.com>
14 * vr-values.h: New file with vr_values class.
15 * tree-vrp.c: Include vr-values.h
16 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
17 data objects into the vr_values class.
18 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
19 (get_value_range): Make it a member function within vr_values class.
20 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
21 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
22 (op_with_constant_singleton_value_range): Likewise.
23 (extract_range_for_var_from_comparison_expr): Likewise.
24 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
25 (extract_range_from_binary_expr): Likewise.
26 (extract_range_from_unary_expr): Likewise.
27 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
28 (check_for_binary_op_overflow, extract_range_basic): Likewise.
29 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
30 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
31 (compare_name_with_value, compare_names): Likewise.
32 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
33 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
34 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
35 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
36 (extract_range_from_phi_node): Likewise.
37 (simplify_truth_ops_using_ranges): Likewise.
38 (simplify_div_or_mod_using_ranges): Likewise.
39 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
40 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
41 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
42 (simplify_float_conversion_using_ranges): Likewise.
43 (simplify_internal_call_using_ranges): Likewise.
44 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
45 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
46 poitner via x_vr_values for calls into gimple folder.
47 (vrp_initialize_lattice): Make this the vr_values ctor.
48 (vrp_free_lattice): Make this the vr_values dtor.
49 (set_vr_value): New function.
50 (class vrp_prop): Add vr_values data member. Add various member
51 functions as well as member functions that delegate to vr_values.
52 (check_array_ref): Make a member function within vrp_prop class.
53 (search_for_addr_array, vrp_initialize): Likewise.
54 (vrp_finalize): Likewise. Revamp to avoid direct access to
55 vr_value, values_propagated, etc.
56 (check_array_bounds): Extract vrp_prop class instance pointer from
57 walk info structure. Use it to call member functions.
58 (check_all_array_refs): Make a member function within vrp_prop class.
59 Smuggle class instance pointer via walk info structure.
60 (x_vr_values): New local static.
61 (vrp_valueize): Use x_vr_values to get class instance.
62 (vr_valueize_1): Likewise.
63 (class vrp_folder): Add vr_values data member. Add various member
64 functions as well as member functions that delegate to vr_values.
65 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
66 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
67 class instance from vr_values. Use it to call member functions.
68 (vrp_dom_walker): Add vr_values data member.
69 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
70 instance via x_vr_values.
71 (identify_jump_threads): Accept vr_values as argument. Store
72 it into the walker structure.
73 (evrp_dom_walker): Add vr_values class data member. Add various
75 (evrp_dom_walker::try_find_new_range): Use vr_values data
76 member to access the memory allocator.
77 (evrp_dom_walker::before_dom_children): Store vr_values class
78 instance into the vrp_folder class.
79 (evrp_dom_walker::push_value_range): Rework to avoid direct
80 access to num_vr_values and vr_value.
81 (evrp_dom_walker::pop_value_range): Likewise.
82 (execute_early_vrp): Remove call to vrp_initialize_lattice.
83 Use vr_values to get to dump_all_value_ranges member function.
84 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
85 and simplify_cond_using_ranges_2 via vrp_prop class instance.
86 Pass vr_values class instance down to identify_jump_threads.
87 Remove call to vrp_free_lattice.
88 (debug_all_value_ranges): Remove.
90 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
91 (vrp_folder): Likewise.
93 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
94 Get it from the existing bitmap instead.
95 (vrp_intersect_ranges_1): Likewise.
97 2017-11-09 Jakub Jelinek <jakub@redhat.com>
99 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
101 (store_immediate_info::store_immediate_info): Add bitnotp argument,
103 (imm_store_chain_info::coalesce_immediate_stores): Break group
104 if bit_not_p is different.
105 (count_multiple_uses, split_group,
106 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
107 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
108 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
109 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
110 to store_immediate_info ctor.
112 2017-11-09 Jim Wilson <jimw@sifive.com>
114 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
115 (scan_prog_file): Likewise.
117 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
119 * bb-reorder.c (max_entry_frequency): Remove.
120 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
121 connect_traces, push_to_next_round_p): Remove prototypes.
122 (find_traces_1_round): Use counts only.
123 (push_to_next_round_p): Likewise.
124 (find_traces): Likewise.
125 (rotate_loop): Likewise.
126 (find_traces_1_round): Likewise.
127 (connect_traces): Likewise.
128 (edge_order): Likewise.
130 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
132 * config/arm/arm.c (output_return_instruction): Add comments to
133 indicate requirement for cmse_nonsecure_entry return to account
134 for the size of clearing instruction output here.
135 (thumb_exit): Likewise.
136 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
137 return in hardfloat mode.
139 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
141 * config/rs6000/rs6000.c (machine_function): Add a bool,
142 "toc_is_wrapped_separately".
143 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
144 if it wasn't explicitly set or unset, we are optimizing for speed, and
145 doing separate shrink-wrapping.
146 (rs6000_get_separate_components): Enable the TOC component if
147 saving the TOC register in the prologue.
148 (rs6000_components_for_bb): Handle the TOC component.
149 (rs6000_emit_prologue_components): Store the TOC register where needed.
150 (rs6000_set_handled_components): Mark TOC as handled, if handled.
151 (rs6000_emit_prologue): Don't save the TOC if that is already done.
153 2017-11-09 Martin Jambor <mjambor@suse.cz>
155 * ipa-param-manipulation.c: New file.
156 * ipa-param-manipulation.h: Likewise.
157 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
158 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
159 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
160 (ipa_parm_adjustment): Likewise.
161 (ipa_parm_adjustment_vec): Likewise.
162 (ipa_get_vector_of_formal_parms): Moved declaration to
163 ipa-param-manipulation.h.
164 (ipa_get_vector_of_formal_parm_types): Likewise.
165 (ipa_modify_formal_parameters): Likewise.
166 (ipa_modify_call_arguments): Likewise.
167 (ipa_combine_adjustments): Likewise.
168 (ipa_dump_param_adjustments): Likewise.
169 (ipa_modify_expr): Likewise.
170 (ipa_get_adjustment_candidate): Likewise.
171 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
172 ipa-param-manipulation.c.
173 (ipa_get_vector_of_formal_parm_types): Likewise.
174 (ipa_modify_formal_parameters): Likewise.
175 (ipa_modify_call_arguments): Likewise.
176 (ipa_modify_expr): Likewise.
177 (get_ssa_base_param): Likewise.
178 (ipa_get_adjustment_candidate): Likewise.
179 (index_in_adjustments_multiple_times_p): Likewise.
180 (ipa_combine_adjustments): Likewise.
181 (ipa_dump_param_adjustments): Likewise.
182 * tree-sra.c: Also include ipa-param-manipulation.h
183 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
186 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
187 Alan Hayward <alan.hayward@arm.com>
188 David Sherwood <david.sherwood@arm.com>
190 * doc/sourcebuild.texi (vect_masked_store): Document.
192 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
193 Alan Hayward <alan.hayward@arm.com>
194 David Sherwood <david.sherwood@arm.com>
196 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
198 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
199 Alan Hayward <alan.hayward@arm.com>
200 David Sherwood <david.sherwood@arm.com>
202 * doc/sourcebuild.texi (vect_variable_length): Document.
204 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
205 Alan Hayward <alan.hayward@arm.com>
206 David Sherwood <david.sherwood@arm.com>
208 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
210 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
211 Alan Hayward <alan.hayward@arm.com>
212 David Sherwood <david.sherwood@arm.com>
214 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
216 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
217 Alan Hayward <alan.hayward@arm.com>
218 David Sherwood <david.sherwood@arm.com>
220 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
221 previously undocumented selectors.
222 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
224 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
226 * doc/rtl.texi (const_vector): Say that elements can be
228 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
229 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
230 (gen_vec_duplicate): Use it instead of CONSTANT_P.
231 * optabs.c (expand_vector_broadcast): Likewise.
233 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
234 Alan Hayward <alan.hayward@arm.com>
235 David Sherwood <david.sherwood@arm.com>
237 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
238 scaled index even if the unscaled address was invalid.
239 Don't increase the complexity of using a scale in that case.
241 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
242 Alan Hayward <alan.hayward@arm.com>
243 David Sherwood <david.sherwood@arm.com>
245 * doc/rtl.texi: Rewrite the subreg rules so that they partition
246 the inner register into REGMODE_NATURAL_SIZE bytes rather than
247 UNITS_PER_WORD bytes.
248 * emit-rtl.c (validate_subreg): Divide subregs into blocks
249 based on REGMODE_NATURAL_SIZE of the inner mode.
250 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
251 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
252 * expmed.c (lowpart_bit_field_p): Divide the value up into
253 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
254 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
255 whether something is likely to occupy more than one register.
257 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
260 * ipa-inline-transform.c (update_noncloned_frequencies): Use
261 profile_count::adjust_for_ipa_scaling.
262 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
263 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
265 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
267 2017-11-09 Jakub Jelinek <jakub@redhat.com>
269 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
270 (split_group): Add total_orig and total_new arguments, estimate the
271 number of statements related to the store group without store merging
272 and with store merging.
273 (imm_store_chain_info::output_merged_store): Adjust split_group
274 callers, punt if estimated number of statements with store merging
275 is not smaller than estimated number of statements without it.
277 (handled_load): Remove has_single_use checks.
278 (pass_store_merging::process_store): Likewise.
280 2017-11-09 Richard Biener <rguenther@suse.de>
282 PR tree-optimization/82902
283 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
285 2017-11-09 Martin Liska <mliska@suse.cz>
288 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
291 2017-11-09 Martin Liska <mliska@suse.cz>
293 PR tree-optimization/82669
294 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
296 2017-11-09 Martin Liska <mliska@suse.cz>
298 PR gcov-profile/48463
299 * coverage.c (coverage_begin_function): Output also end locus
300 of a function and information whether the function is
302 * gcov-dump.c (tag_function): Parse and print the information.
303 * gcov.c (INCLUDE_MAP): Add include.
304 (INCLUDE_SET): Likewise.
305 (struct line_info): Move earlier in the source file because
306 of vector<line_info> in function_info structure.
307 (line_info::line_info): Likewise.
308 (line_info::has_block): Likewise.
309 (struct source_info): Add new member index.
310 (source_info::get_functions_at_location): New function.
311 (function_info::group_line_p): New function.
312 (output_intermediate_line): New function.
313 (output_intermediate_file): Use the mentioned function.
314 (struct function_start): New.
315 (struct function_start_pair_hash): Likewise.
316 (process_file): Add code that identifies group functions.
317 Assign lines either to global or function scope.
318 (generate_results): Skip artificial functions.
319 (find_source): Assign index for each source file.
320 (read_graph_file): Read new flag artificial and end_line.
321 (add_line_counts): Assign it either to global of function scope.
322 (accumulate_line_counts): Isolate core of the function to
323 accumulate_line_info and call it for both function and global
325 (accumulate_line_info): New function.
326 (output_line_beginning): Fix GNU coding style.
327 (print_source_line): New function.
328 (output_line_details): Likewise.
329 (output_function_details): Likewise.
330 (output_lines): Iterate both source (global) scope and function
332 (struct function_line_start_cmp): New class.
333 * doc/gcov.texi: Reflect changes in documentation.
335 2017-11-09 Jakub Jelinek <jakub@redhat.com>
338 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
339 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
340 and similarly for not instead of neg.
342 2017-11-08 Andi Kleen <ak@linux.intel.com>
344 * config/i386/i386.opt: Add -mforce-indirect-call.
345 * config/i386/predicates.md: Check for flag_force_indirect_call.
346 * doc/invoke.texi: Document -mforce-indirect-call
348 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
350 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
352 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
353 config/riscv/riscv.c (predict.h): New include.
354 (riscv_slow_unaligned_access_p): No longer static.
355 (riscv_block_move_straight): Add require.
356 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
358 2017-11-08 Jakub Jelinek <jakub@redhat.com>
361 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
362 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
363 nonimmediate_operand predicate for operand 1 instead of
366 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
368 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
370 * config/aarch64/constraints.md (Uml): New constraint.
371 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
374 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
376 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
377 of two vec_duplicates into a vec_concat.
379 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
381 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
382 Simplify vec_merge of vec_duplicate and vec_concat.
383 * config/aarch64/constraints.md (Utq): New constraint.
384 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
387 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
389 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
390 Simplify vec_merge of vec_duplicate and const_vector.
391 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
393 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
394 mode iterator. Update predicate on operand 1 to
395 handle non-const_vec constants. Delete constraints.
396 (*aarch64_combinez_be<mode>): Likewise for operand 2.
398 2017-11-08 Jakub Jelinek <jakub@redhat.com>
400 PR tree-optimization/78821
401 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
403 (store_operand_info::store_operand_info): Initialize it to false.
404 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
405 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
406 store in the group, and if chain_info is non-NULL, to ignore altogether
408 (compatible_load_p): Fail if bit_not_p does not match.
409 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
410 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
411 (pass_store_merging::process_store): Adjust
412 terminate_all_aliasing_chains calls to pass NULL in all current spots,
413 call terminate_all_aliasing_chains newly when adding a store into
414 a chain with non-NULL chain_info.
416 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
418 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
420 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
422 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
424 (aarch64_layout_frame): Initialise emit_frame_chain.
425 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
426 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
428 2017-11-08 Martin Liska <mliska@suse.cz>
430 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
431 of gimple_call_internal_p.
433 2017-11-07 Tom de Vries <tom@codesourcery.com>
435 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
437 2017-11-07 Tom de Vries <tom@codesourcery.com>
439 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
442 2017-11-08 Martin Liska <mliska@suse.cz>
445 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
447 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
449 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
451 2017-11-07 Jakub Jelinek <jakub@redhat.com>
454 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
455 (*cmp<mode>_ccz_1): New insn with $k alternative.
458 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
459 RTX_COMM_COMPARE as commutative as well.
460 (ix86_binary_operator_ok): Formatting fix.
461 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
462 *<code><mode>3<mask_name><round_saeonly_name>,
463 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
464 *<s>mul<mode>3_highpart<mask_name>,
465 *vec_widen_umult_even_v16si<mask_name>,
466 *vec_widen_umult_even_v8si<mask_name>,
467 *vec_widen_umult_even_v4si<mask_name>,
468 *vec_widen_smult_even_v16si<mask_name>,
469 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
470 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
471 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
472 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
473 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
474 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
475 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
476 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
477 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
478 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
479 ix86_binary_operator_ok. Formatting fixes.
480 (*<plusminus_insn><mode>3<mask_name><round_name>,
481 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
484 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
486 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
487 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
488 eq and ne if TARGET_ISEL.
489 (cmp): New code_iterator.
490 (UNS, UNSU_, UNSIK): New code_attrs.
491 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
492 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
494 ("eq<mode>3"): ... this.
495 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
497 ("ne<mode>3"): ... this.
499 2017-11-07 Julia Koval <julia.koval@intel.com>
502 * common/config/i386/i386-common.c
503 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
504 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
505 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
506 * config/i386/i386.opt: Ditto.
507 * config/i386/i386.c (ix86_target_string): Ditto.
508 (ix86_option_override_internal): Ditto.
509 (ix86_init_mpx_builtins): Move MPX to args2.
510 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
511 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
512 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
513 __builtin_ia32_vgf2p8affineinvqb_v32qi,
514 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
515 __builtin_ia32_vgf2p8affineinvqb_v16qi,
516 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
518 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
521 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
522 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
523 (zero-extendsidi peephole2): Remove peephole.
525 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
528 * stor-layout.c: Include attribs.h.
529 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
530 explicit lookup of "aligned" attribute.
532 2017-11-07 Andrew Waterman <andrew@sifive.com>
534 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
535 (riscv_expand_block_move): Likewise.
536 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
538 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
539 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
540 * config/riscv/riscv.c (riscv_block_move_straight): New function.
541 (riscv_adjust_block_mem): Likewise.
542 (riscv_block_move_loop): Likewise.
543 (riscv_expand_block_move): Likewise.
544 gcc/config/riscv/riscv.md (movmemsi): New pattern.
546 2017-11-07 Michael Clark <michaeljclark@mac.com>
548 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
549 (MUSL_DYNAMIC_LINKER): Likewise.
551 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
553 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
554 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
557 2017-11-07 Richard Biener <rguenther@suse.de>
559 * match.pd: Fix build.
561 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
562 Jackson Woodruff <jackson.woodruff@arm.com>
564 PR tree-optimization/71026
565 * match.pd: Canonicalize negate in division.
567 2017-11-07 Sudakshina Das <sudi.das@arm.com>
570 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
572 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
574 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
575 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
577 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
579 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
580 non-scalar integral types.
581 * match.pd (negate_expr_p): Handle MINUS_EXPR.
582 (-(A-B), -(~A)): New transformations.
584 2017-11-07 Tom de Vries <tom@codesourcery.com>
586 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
587 semicolon after "do {} while (0)".
588 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
589 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
590 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
591 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
592 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
594 2017-11-07 Tom de Vries <tom@codesourcery.com>
596 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
597 after "do {} while (0)".
598 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
599 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
600 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
601 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
602 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
604 2017-11-07 Tom de Vries <tom@codesourcery.com>
607 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
609 (arm_rtx_costs_internal): Add missing semicolon after
610 HANDLE_NARROW_SHIFT_ARITH call.
612 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
614 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
615 disable isel if it was not set explicitly.
617 2017-11-06 James Bowman <james.bowman@ftdichip.com>
619 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
621 (add_type_attribute) likewise.
623 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
625 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
626 of struct ix86_frame.
627 (ix86_initial_elimination_offset): Likewise.
628 (ix86_expand_split_stack_prologue): Likewise.
630 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
632 * tree-vrp.h (enum value_range_type): Update stale comment.
634 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
635 Alan Hayward <alan.hayward@arm.com>
636 David Sherwood <david.sherwood@arm.com>
638 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
639 (aarch64_expand_vec_perm_const): Take the number of units too.
640 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
641 (aarch64_expand_vec_perm_const): Likewise.
642 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
643 (vec_perm<mode>): Update accordingly.
645 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
646 Alan Hayward <alan.hayward@arm.com>
647 David Sherwood <david.sherwood@arm.com>
649 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
650 Take the number of units too.
651 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
652 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
653 but check for a vector mode before rather than after the call.
654 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
655 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
656 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
657 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
658 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
659 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
660 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
661 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
662 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
663 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
664 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
665 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
666 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
667 (aarch64_sqdmull2_n<mode>): Update accordingly.
669 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
670 Alan Hayward <alan.hayward@arm.com>
671 David Sherwood <david.sherwood@arm.com>
673 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
674 the number of units too.
675 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
676 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
677 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
678 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
679 (vec_store_lanesxi<mode>): Update accordingly.
681 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
682 Alan Hayward <alan.hayward@arm.com>
683 David Sherwood <david.sherwood@arm.com>
685 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
686 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
687 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
688 of units rather than the mode.
689 * config/aarch64/iterators.md (nunits): New mode attribute.
690 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
691 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
692 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
693 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
694 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
695 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
696 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
697 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
698 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
699 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
700 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
701 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
702 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
703 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
704 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
705 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
706 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
707 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
708 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
709 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
710 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
711 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
712 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
713 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
714 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
715 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
716 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
717 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
718 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
719 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
720 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
721 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
722 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
723 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
724 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
725 (aarch64_simd_vec_setv2di): Likewise.
727 2017-11-06 Carl Love <cel@us.ibm.com>
729 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
731 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
732 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
734 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
736 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
737 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
738 Add power 8 macro expansions.
739 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
740 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
741 power 8 instructions. (VSX_XXBR): Add iterator.
743 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
745 * config/arm/arm.md (predicable_short_it): Change default to "no",
746 improve documentation, remove uses that are identical to the default.
747 (enabled_for_depr_it): Rename to enabled_for_short_it.
748 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
749 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
750 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
751 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
752 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
754 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
757 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
758 float128 helper macros, which are no longer used after deleting
759 the old 'q' built-in functions, and moving the round to odd
760 built-in functions to being special built-in functions.
761 (BU_FLOAT128_2): Likewise.
762 (BU_FLOAT128_1_HW): Likewise.
763 (BU_FLOAT128_2_HW): Likewise.
764 (BU_FLOAT128_3_HW): Likewise.
765 (FABSQ): Delete old 'q' built-in functions.
766 (COPYSIGNQ): Likewise.
767 (SQRTF128_ODD): Move round to odd built-in functions to be
768 special built-in functions, so that we can handle
769 -mabi=ieeelongdouble.
770 (TRUNCF128_ODD): Likewise.
771 (ADDF128_ODD): Likewise.
772 (SUBF128_ODD): Likewise.
773 (MULF128_ODD): Likewise.
774 (DIVF128_ODD): Likewise.
775 (FMAF128_ODD): Likewise.
776 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
777 built-in names to 'f128'.
778 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
779 old 'q' built-in functions, as the machine independent code for
780 'f128' built-in functions handles this.
781 (rs6000_expand_builtin): Add expansion for float128 round to odd
782 functions, keying off on -mabi=ieeelongdouble of whether to use
783 the KFmode or TFmode variant.
784 (rs6000_init_builtins): Initialize the _Float128 round to odd
786 * doc/extend.texi (PowerPC Built-in Functions): Document the old
787 _Float128 'q' built-in functions are now mapped into the new
788 'f128' built-in functions.
790 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
792 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
793 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
795 2017-11-06 David Malcolm <dmalcolm@redhat.com>
798 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
799 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
800 * toplev.c: Include "ipa-fnsummary.h".
801 (toplev::finalize): Call ipa_fnsummary_c_finalize.
803 2017-11-06 Jakub Jelinek <jakub@redhat.com>
805 PR tree-optimization/82838
806 * gimple-ssa-store-merging.c
807 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
808 on a separate gimple_seq which is then appended to seq.
810 2017-11-06 Jeff Law <law@redhat.com>
813 * config/i386/i386.c (PROBE_INTERVAL): Remove.
814 (get_probe_interval): New functions.
815 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
816 (ix86_adjust_stack_and_probe): Likewise.
817 (output_adjust_stack_and_probe): Likewise.
818 (ix86_emit_probe_stack_range): Likewise.
819 (ix86_expand_prologue): Likewise.
821 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
823 PR tree-optimization/82816
824 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
825 if the modes of the two types are the same.
826 (convert_plusminus_to_widen): Likewise.
828 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
830 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
832 (usadv16qi): New define_expand.
833 (usadv8hi): New define_expand.
835 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
838 * ipa-inline-transform.c (update_noncloned_frequencies): Always
840 (inline_transform): Likewise.
841 * predict.c (counts_to_freqs): Remove useless conditional.
842 * profile-count.h (profile_count::apply_scale): Move sanity check.
843 * tree-inline.c (copy_bb): Always scale.
844 (copy_cfg_body): Likewise.
846 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
849 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
851 (*cmp_ite0): Add enabled_for_depr_it attribute.
852 (*cmp_ite1): Likewise.
854 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
856 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
859 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
861 * tree-vrp.c (vrp_int_const_binop): Return true on success and
862 return the value by pointer.
863 (extract_range_from_multiplicative_op_1): Update accordingly.
864 Return as soon as an operation fails.
866 2017-11-05 Tom de Vries <tom@codesourcery.com>
869 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
870 (DEF_SANITIZER_BUILTIN): ... here.
871 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
872 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
874 2017-11-05 Tom de Vries <tom@codesourcery.com>
877 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
879 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
880 ASM_OUTPUT_BEFORE_CASE_LABEL call.
881 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
883 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
884 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
885 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
887 2017-11-05 Tom de Vries <tom@codesourcery.com>
890 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
893 2017-11-04 Michael Clark <michaeljclark@mac.com>
895 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
896 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
898 (*addsi3_extended): Likewise.
899 (*addsi3_extended2): Likewise.
900 (<optab>si3): Likewise.
901 (<optab>di3): Likewise.
902 (<optab><mode>3): Likewise.
903 (<*optabe>si3_internal): Likewise.
904 (zero_extendqi<SUPERQI:mode>2): Likewise.
905 (*add<mode>hi3): Likewise.
906 (*xor<mode>hi3): Likewise.
907 (<optab>di3): Likewise.
908 (*<optab>si3_extend): Likewise.
909 (*sge<u>_<X:mode><GPR:mode>): Likewise.
910 (*slt<u>_<X:mode><GPR:mode>): Likewise.
911 (*sle<u>_<X:mode><GPR:mode>): Likewise.
913 2017-11-04 Andrew Waterman <andrew@sifive.com>
915 * config/riscv/riscv.c (riscv_option_override): Conditionally set
916 TARGET_STRICT_ALIGN based upon -mtune argument.
918 2017-11-04 Andrew Waterman <andrew@sifive.com>
920 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
922 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
924 config/i386/i386.c (choose_basereg): Use optional scratch
925 register and add assertion.
926 (x86_emit_outlined_ms2sysv_save): Use scratch register when
927 needed, and don't allocate stack.
928 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
929 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
930 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
932 2017-11-03 Jeff Law <law@redhat.com>
934 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
935 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
936 to probe at the start of a noreturn function.
938 2017-11-03 Jakub Jelinek <jakub@redhat.com>
940 PR tree-optimization/78821
941 * gimple-ssa-store-merging.c: Update the file comment.
942 (MAX_STORE_ALIAS_CHECKS): Define.
943 (struct store_operand_info): New type.
944 (store_operand_info::store_operand_info): New constructor.
945 (struct store_immediate_info): Add rhs_code and ops data members.
946 (store_immediate_info::store_immediate_info): Add rhscode, op0r
947 and op1r arguments to the ctor, initialize corresponding data members.
948 (struct merged_store_group): Add load_align_base and load_align
950 (merged_store_group::merged_store_group): Initialize them.
951 (merged_store_group::do_merge): Update them.
952 (merged_store_group::apply_stores): Pick the constant for
953 encode_tree_to_bitpos from one of the two operands, or skip
954 encode_tree_to_bitpos if neither operand is a constant.
955 (class pass_store_merging): Add process_store method decl. Remove
956 bool argument from terminate_all_aliasing_chains method decl.
957 (pass_store_merging::terminate_all_aliasing_chains): Remove
958 var_offset_p argument and corresponding handling.
959 (stmts_may_clobber_ref_p): New function.
960 (compatible_load_p): New function.
961 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
962 if there is overlap and rhs_code is not INTEGER_CST. For
963 non-overlapping stores terminate group if rhs is not mergeable.
964 (get_alias_type_for_stmts): Change first argument from
965 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
966 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
967 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
969 (get_location_for_stmts): Change first argument from
970 auto_vec<gimple *> & to vec<gimple *> &.
971 (struct split_store): Remove orig_stmts data member, add orig_stores.
972 (split_store::split_store): Create orig_stores rather than orig_stmts.
973 (find_constituent_stmts): Renamed to ...
974 (find_constituent_stores): ... this. Change second argument from
975 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
976 to info structures rather than the statements.
977 (split_group): Rename ALLOW_UNALIGNED argument to
978 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
979 it. Adjust find_constituent_stores caller.
980 (imm_store_chain_info::output_merged_store): Handle rhs_code other
981 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
982 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
983 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
984 (mem_valid_for_store_merging): New function.
985 (handled_load): New function.
986 (pass_store_merging::process_store): New method.
987 (pass_store_merging::execute): Use process_store method. Adjust
988 terminate_all_aliasing_chains caller.
990 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
992 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
993 Return true for more constants, symbols and label references.
994 (aarch64_valid_floating_const): Remove unused function.
996 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
999 * config/aarch64/aarch64.c (aarch64_layout_frame):
1000 Undo forcing of LR at bottom of frame.
1002 2017-11-03 Jeff Law <law@redhat.com>
1005 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1006 for int_registers_saved.
1008 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1009 extracted from tree-ssa-dom.c.
1010 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1011 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1012 (record_equivalences_from_incoming_edge): Add additional argument
1013 to single_pred_edge_ignoring_loop_edges call.
1014 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1015 (uncprop_dom_walker::before_dom_children): Add additional argument
1016 to single_pred_edge_ignoring_loop_edges call.
1017 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1018 single_pred_edge_ignoring_loop_edges rather than open coding.
1019 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1021 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1023 * match.pd (-(-A)): Rewrite.
1025 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1027 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1028 (rs6000_emit_int_cmove): New declaration.
1029 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1030 (rs6000_emit_sISEL): Delete.
1031 (rs6000_emit_int_cmove): Make non-static.
1032 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1033 instead of rs6000_emit_sISEL.
1035 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1037 * asan.c (create_cond_insert_point): Maintain profile.
1038 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1040 * basic-block.h (struct basic_block_def): Remove frequency.
1041 (EDGE_FREQUENCY): Use to_frequency
1042 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1044 (find_traces): Update to use to_frequency.
1045 (find_traces_1_round): Likewise; use only IPA counts.
1046 (bb_to_key): Likewise.
1047 (connect_traces): Use IPA counts only.
1048 (copy_bb_p): Update to use to_frequency.
1049 (fix_up_crossing_landing_pad): Likewise.
1050 (sanitize_hot_paths): Likewise.
1051 * bt-load.c (basic_block_freq): Likewise.
1052 * cfg.c (init_flow): Set count_max to uninitialized.
1053 (check_bb_profile): Remove frequencies; check counts.
1054 (dump_bb_info): Do not dump frequencies.
1055 (update_bb_profile_for_threading): Update counts only.
1056 (scale_bbs_frequencies_int): Likewise.
1057 (MAX_SAFE_MULTIPLIER): Remove.
1058 (scale_bbs_frequencies_gcov_type): Update counts only.
1059 (scale_bbs_frequencies_profile_count): Update counts only.
1060 (scale_bbs_frequencies): Update counts only.
1061 * cfg.h (struct control_flow_graph): Add count-max.
1062 (update_bb_profile_for_threading): Update prototype.
1063 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1064 (find_many_sub_basic_blocks): Likewise.
1065 * cfgcleanup.c (try_forward_edges): Likewise.
1066 (try_crossjump_to_edge): Likewise.
1067 * cfgexpand.c (expand_gimple_cond): Likewise.
1068 (expand_gimple_tailcall): Likewise.
1069 (construct_init_block): Likewise.
1070 (construct_exit_block): Likewise.
1071 * cfghooks.c (verify_flow_info): Check consistency of counts.
1072 (dump_bb_for_graph): Do not dump frequencies.
1073 (split_block_1): Do not update frequencies.
1074 (split_edge): Do not update frequencies.
1075 (make_forwarder_block): Do not update frequencies.
1076 (duplicate_block): Do not update frequencies.
1077 (account_profile_record): Do not update frequencies.
1078 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1079 for global heuristics.
1080 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1081 (expected_loop_iterations_unbounded): Use counts only.
1082 * cfgloopmanip.c (scale_loop_profile): Simplify.
1083 (create_empty_loop_on_edge): Simplify
1085 (duplicate_loop_to_header_edge): Simplify
1086 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1087 (update_br_prob_note): Take care of removing note when profile
1089 (relink_block_chain): Do not dump frequency.
1090 (rtl_account_profile_record): Use to_frequency.
1091 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1092 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1093 (cgraph_update_edges_for_call_stmt_node): Likewise.
1094 (cgraph_edge::verify_count_and_frequency): Update.
1095 (cgraph_node::verify_node): Temporarily disable frequency verification.
1096 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1097 to_cgraph_frequency.
1098 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1099 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1101 (cgraph_node::expand_thunk): Update profile.
1102 * except.c (dw2_build_landing_pads): Do not update frequency.
1103 * final.c (compute_alignments): Use to_frequency.
1104 (dump_basic_block_info): Do not dump frequency.
1105 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1106 (dump_gimple_bb_header): Do not dump frequency.
1107 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1109 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1110 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1111 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1112 (init_before_recovery): Do not update frequency.
1113 (sched_create_recovery_edges): Do not update frequency.
1114 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1115 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1116 (ipa_cp_c_finalize): Set max_count to uninitialized.
1117 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1118 (param_change_prob): Use counts.
1119 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1121 * ipa-split.c (consider_split): Use to_frequency.
1122 (split_function): Use to_frequency.
1123 * ira-build.c (loop_compare_func): Likewise.
1124 (mark_loops_for_removal): Likewise.
1125 (mark_all_loops_for_removal): Likewise.
1126 * loop-doloop.c (doloop_modify): Do not update frequency.
1127 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1129 * lto-streamer-in.c (input_function): Update count_max.
1130 * omp-expand.c (expand_omp_taskreg): Update count_max.
1131 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1132 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1133 (maybe_hot_count_p): Use ipa counts only.
1134 (maybe_hot_bb_p): Simplify.
1135 (maybe_hot_edge_p): Simplify.
1136 (probably_never_executed): Do not take frequency argument.
1137 (probably_never_executed_bb_p): Do not pass frequency.
1138 (probably_never_executed_edge_p): Likewise.
1139 (combine_predictions_for_bb): Check that profile is nonzero.
1140 (propagate_freq): Do not set frequency.
1141 (drop_profile): Simplify.
1142 (counts_to_freqs): Simplify.
1143 (expensive_function_p): Use to_frequency.
1144 (propagate_unlikely_bbs_forward): Simplify.
1145 (determine_unlikely_bbs): Simplify.
1146 (estimate_bb_frequencies): Add hack to silence graphite issues.
1147 (compute_function_frequency): Use ipa counts.
1148 (pass_profile::execute): Update.
1149 (rebuild_frequencies): Use counts only.
1150 (force_edge_cold): Use counts only.
1151 * profile-count.c (profile_count::dump): Dump new count types.
1152 (profile_count::differs_from_p): Check compatiblity.
1153 (profile_count::to_frequency): New function.
1154 (profile_count::to_cgraph_frequency): New function.
1155 * profile-count.h (struct function): Declare.
1156 (enum profile_quality): Add profile_guessed_local and
1157 profile_guessed_global0.
1158 (class profile_proability): Decrease number of bits to 29;
1159 update from_reg_br_prob_note and to_reg_br_prob_note.
1160 (class profile_count: Update comment; decrease number of bits
1161 to 61. Check compatibility.
1162 (profile_count::compatible_p): New private member function.
1163 (profile_count::ipa_p): New member function.
1164 (profile_count::operator<): Handle global zero correctly.
1165 (profile_count::operator>): Handle global zero correctly.
1166 (profile_count::operator<=): Handle global zero correctly.
1167 (profile_count::operator>=): Handle global zero correctly.
1168 (profile_count::nonzero_p): New member function.
1169 (profile_count::force_nonzero): New member function.
1170 (profile_count::max): New member function.
1171 (profile_count::apply_scale): Handle IPA scalling.
1172 (profile_count::guessed_local): New member function.
1173 (profile_count::global0): New member function.
1174 (profile_count::ipa): New member function.
1175 (profile_count::to_frequency): Declare.
1176 (profile_count::to_cgraph_frequency): Declare.
1177 * profile.c (OVERLAP_BASE): Delete.
1178 (compute_frequency_overlap): Delete.
1179 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1180 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1181 * sched-ebb.c (rank): Use counts only.
1182 * shrink-wrap.c (handle_simple_exit): Use counts only.
1183 (try_shrink_wrapping): Use counts only.
1184 (place_prologue_for_one_component): Use counts only.
1185 * tracer.c (find_best_predecessor): Use to_frequency.
1186 (find_trace): Use to_frequency.
1187 (tail_duplicate): Use to_frequency.
1188 * trans-mem.c (expand_transaction): Do not update frequency.
1189 * tree-call-cdce.c: Do not update frequency.
1190 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1191 (gimple_merge_blocks): Likewise.
1192 (gimple_split_edge): Likewise.
1193 (gimple_duplicate_sese_region): Likewise.
1194 (gimple_duplicate_sese_tail): Likewise.
1195 (move_sese_region_to_fn): Likewise.
1196 (gimple_account_profile_record): Likewise.
1197 (insert_cond_bb): Likewise.
1198 * tree-complex.c (expand_complex_div_wide): Likewise.
1199 * tree-eh.c (lower_resx): Update profile.
1200 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1202 (initialize_cfun): Do not initialize frequencies
1203 (freqs_to_counts): Delete.
1204 (copy_cfg_body): Ignore count parameter.
1205 (copy_body): Update.
1206 (expand_call_inline): Update count_max.
1207 (optimize_inline_calls): Update count_max.
1208 (tree_function_versioning): Update count_max.
1209 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1210 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1212 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1213 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1214 (try_peel_loop): Likewise.
1215 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1217 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1218 (tree_transform_and_unroll_loop): Do not use frequencies
1219 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1220 Use reliable prediction only.
1221 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1222 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1223 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1224 probability scaling.
1225 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1226 not update frequency
1227 (any_remaining_duplicated_blocks): Likewise.
1228 (update_profile): Likewise.
1229 (estimated_freqs_path): Delete.
1230 (freqs_to_counts_path): Delete.
1231 (clear_counts_path): Delete.
1232 (ssa_fix_duplicate_block_edges): Likewise.
1233 (duplicate_thread_path): Likewise.
1234 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1235 * tree-tailcall.c (decrease_profile): Do not update frequency.
1236 (eliminate_tail_call): Likewise.
1237 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1238 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1239 (optimize_mask_stores): Likewise.
1240 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1241 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1242 (ubsan_expand_ptr_ifn): Update profile.
1243 * value-prof.c (gimple_ic): Simplify.
1244 * value-prof.h (gimple_ic): Update prototype.
1245 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1246 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1248 (want_inline_self_recursive_call_p): Likewise.
1249 (resolve_noninline_speculation): Only cummulate defined counts.
1250 (inline_small_functions): Use nonzero_p.
1251 (ipa_inline): Do not access freed node.
1253 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1255 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1256 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1258 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1260 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1261 non-legitimate address.
1263 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1265 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1266 (*lt0_<mode>di, *lt0_<mode>si): New.
1268 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1270 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1271 TARGET_PAIRED_FLOAT.
1273 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1274 Jim Wilson <jim.wilson@linaro.org>
1276 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1277 * config/aarch64/aarch64-tune.md: Regenerated.
1278 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1279 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1281 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1283 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1284 (arc_expand_prologue): Restore blink for millicode.
1285 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1287 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1290 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1291 gen_vec_duplicate after forcing the scalar into a register.
1293 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1295 * combine (try_combine): Print the insns input to try_combine to the
1298 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1301 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1302 Remove second argument from aarch64_process_target_attr call.
1303 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1305 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1307 (aarch64_handle_attr_arch): Remove second argument.
1308 (aarch64_handle_attr_cpu): Ditto.
1309 (aarch64_handle_attr_tune): Ditto.
1310 (aarch64_handle_attr_isa_flags): Ditto.
1311 (aarch64_process_one_target_attr): Ditto.
1312 (aarch64_process_target_attr): Ditto.
1313 (aarch64_option_valid_attribute_p): Remove second argument.
1314 on aarch64_process_target_attr call.
1316 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1318 * diagnostic.c: Include "selftest-diagnostic.h".
1319 (selftest::assert_location_text): New function.
1320 (selftest::test_diagnostic_get_location_text): New function.
1321 (selftest::diagnostic_c_tests): Call it.
1323 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1325 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1326 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1327 (class selftest::test_diagnostic_context): Move to...
1328 * selftest-diagnostic.c: New file.
1329 * selftest-diagnostic.h: New file.
1331 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1333 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1334 offset range for FT32B.
1335 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1336 * config/ft32/ft32.md: Add TARGET_NOPM.
1337 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1338 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1340 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1342 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1344 2017-11-02 Jeff Law <law@redhat.com>
1346 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1347 virtual keyword on FINAL OVERRIDE members.
1349 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1350 virtuals together. Add virtual destructor.
1351 (substitute_and_fold_engine): Similarly.
1353 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1355 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1357 2017-11-02 Richard Biener <rguenther@suse.de>
1359 PR tree-optimization/82795
1360 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1362 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1364 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1366 (gcc_GAS_CHECK_FEATURE): Remove.
1367 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1368 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1369 * configure: Regenerate.
1371 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1373 * config/arc/arc.c (hwloop_optimize): Account for empty
1376 2017-11-02 Richard Biener <rguenther@suse.de>
1379 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1380 Truncate ARRAY_REF index and element size.
1382 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1384 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1386 2017-11-01 Jeff Law <law@redhat.com>
1388 * tree-ssa-ccp.c (ccp_folder): New class derived from
1389 substitute_and_fold_engine.
1390 (ccp_folder::get_value): New member function.
1391 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1392 (ccp_fold_stmt): Remove prototype.
1393 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1394 * tree-ssa-copy.c (copy_folder): New class derived from
1395 substitute_and_fold_engine.
1396 (copy_folder::get_value): Renamed from get_value.
1397 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1398 * tree-vrp.c (vrp_folder): New class derived from
1399 substitute_and_fold_engine.
1400 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1401 (vrp_folder::get_value): New member function.
1402 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1403 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1404 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1405 provide a class interface to folder/substitute routines.
1406 (ssa_prop_fold_stmt_fn): Remove typedef.
1407 (ssa_prop_get_value_fn): Likewise.
1408 (subsitute_and_fold): Remove prototype.
1409 (replace_uses_in): Likewise.
1410 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1411 Renamed from replace_uses_in. Call the virtual member function
1412 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1413 (substitute_and_fold_dom_walker): Remove initialization of
1414 data member entries for calbacks. Add substitute_and_fold_engine
1415 member and initialize it.
1416 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1417 member functions for get_value, replace_phi_args_in c
1418 replace_uses_in, and fold_stmt calls.
1419 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1420 substitute_and_fold. Remove assert. Update ctor call.
1422 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1423 (ssa_prop_visit_phi_fn): Likewise.
1424 (class ssa_propagation_engine): New class to provide an interface
1426 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1428 (ssa_prop_visit_phi): Likewise.
1429 (ssa_propagation_engine::simulate_stmt): Moved into class.
1430 Call visit_phi/visit_stmt from the class rather than via
1431 file scoped static variables.
1432 (ssa_propagation_engine::simulate_block): Moved into class.
1433 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1434 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1435 set file scoped statics for the visit_stmt/visit_phi callbacks.
1436 * tree-complex.c (complex_propagate): New class derived from
1437 ssa_propagation_engine.
1438 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1439 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1440 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1442 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1443 ssa_propagation_engine.
1444 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1445 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1446 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1447 * tree-ssa-copy.c (copy_prop): New class derived from
1448 ssa_propagation_engine.
1449 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1450 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1451 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1452 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1453 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1454 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1455 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1457 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1459 PR rtl-optimization/82778
1460 PR rtl-optimization/82597
1461 * compare-elim.c (struct comparison): Add in_a_setter field.
1462 (find_comparison_dom_walker::before_dom_children): Remove killed
1463 bitmap and df_simulate_find_defs call, instead walk the defs.
1464 Compute last_setter and initialize in_a_setter. Merge definitions
1465 with first initialization for a few variables.
1466 (try_validate_parallel): Use insn_invalid_p instead of
1467 recog_memoized. Return insn rather than just the pattern.
1468 (try_merge_compare): Fix up comment. Don't uselessly test if
1469 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1471 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1474 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1475 Alan Hayward <alan.hayward@arm.com>
1476 David Sherwood <david.sherwood@arm.com>
1478 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1479 aarch64_hard_regno_nregs to get the number of registers
1482 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1483 Alan Hayward <alan.hayward@arm.com>
1484 David Sherwood <david.sherwood@arm.com>
1486 * config/aarch64/constraints.md (Upl): Rename to...
1488 * config/aarch64/aarch64.md
1489 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1492 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1493 Alan Hayward <alan.hayward@arm.com>
1494 David Sherwood <david.sherwood@arm.com>
1496 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1497 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1500 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1501 Alan Hayward <alan.hayward@arm.com>
1502 David Sherwood <david.sherwood@arm.com>
1504 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1505 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1506 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1508 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1509 of a single element.
1510 * config/aarch64/iterators.md: Add a comment above the permute
1511 unspecs to say that they are generated directly by
1512 aarch64_expand_vec_perm_const.
1513 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1515 2017-11-01 Nathan Sidwell <nathan@acm.org>
1517 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1519 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1521 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1522 and medany code models, and describe what they do.
1524 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1526 Revert accidental duplicate:
1528 * combine.c (can_change_dest_mode): Reject changes in
1529 REGMODE_NATURAL_SIZE.
1531 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1533 PR rtl-optimization/64682
1534 PR rtl-optimization/69567
1535 PR rtl-optimization/69737
1536 PR rtl-optimization/82683
1537 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1538 register mentioned in the note, drop the note, unless it came from I3,
1539 in which case it should go to I3 again.
1541 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1543 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1544 and return false if not.
1545 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1547 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1549 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1550 range is known to be empty.
1552 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1553 Alan Hayward <alan.hayward@arm.com>
1554 David Sherwood <david.sherwood@arm.com>
1556 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1557 and CONST_VECTOR_NUNITS instead of computing the number of units from
1558 the byte sizes of the vector and element.
1559 (simplify_binary_operation_1): Likewise.
1560 (simplify_const_binary_operation): Likewise.
1561 (simplify_ternary_operation): Likewise.
1563 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1564 Alan Hayward <alan.hayward@arm.com>
1565 David Sherwood <david.sherwood@arm.com>
1567 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1568 (int_mem_offset): ...this new function.
1569 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1570 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1571 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1574 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1575 Alan Hayward <alan.hayward@arm.com>
1576 David Sherwood <david.sherwood@arm.com>
1578 * lower-subreg.c (interesting_mode_p): New function.
1579 (compute_costs, find_decomposable_subregs, decompose_register)
1580 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1581 (resolve_clobber, dump_choices): Use it.
1583 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1584 Alan Hayward <alan.hayward@arm.com>
1585 David Sherwood <david.sherwood@arm.com>
1587 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
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 * alias.c (find_base_value, find_base_term): Only process integer
1594 truncations. Check the precision rather than the size.
1596 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1597 Alan Hayward <alan.hayward@arm.com>
1598 David Sherwood <david.sherwood@arm.com>
1600 * machmode.h (is_narrower_int_mode): New function
1601 * optabs.c (expand_float, expand_fix): Use it.
1602 * dwarf2out.c (rotate_loc_descriptor): Likewise.
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 * rtl.h (narrower_subreg_mode): New function.
1609 * ira-color.c (update_costs_from_allocno): Use it.
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 * optabs-query.h (convert_optab_p): New function, split out from...
1616 (convert_optab_handler): ...here.
1617 (widening_optab_handler): Delete.
1618 (find_widening_optab_handler): Remove permit_non_widening parameter.
1619 (find_widening_optab_handler_and_mode): Likewise. Provide an
1620 override that operates on mode class wrappers.
1621 * optabs-query.c (widening_optab_handler): Delete.
1622 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1623 parameter. Assert that the two modes are the same class and that
1624 the "from" mode is narrower than the "to" mode. Use
1625 convert_optab_handler instead of widening_optab_handler.
1626 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1627 instead of widening_optab_handler.
1628 * expr.c (expand_expr_real_2): Update calls to
1629 find_widening_optab_handler.
1630 * optabs.c (expand_widen_pattern_expr): Likewise.
1631 (expand_binop_directly): Take the insn_code as a parameter.
1632 (expand_binop): Only call find_widening_optab_handler for
1633 conversion optabs; use optab_handler otherwise. Update calls
1634 to find_widening_optab_handler and expand_binop_directly.
1635 Use convert_optab_handler instead of widening_optab_handler.
1636 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1637 find_widening_optab_handler and use scalar_mode rather than
1639 (convert_plusminus_to_widen): Likewise.
1641 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1642 Alan Hayward <alan.hayward@arm.com>
1643 David Sherwood <david.sherwood@arm.com>
1645 * machmode.h (fixed_size_mode): New class.
1646 * rtl.h (get_pool_mode): Return fixed_size_mode.
1647 * gengtype.c (main): Add fixed_size_mode.
1648 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1649 (get_raw_arg_mode): Likewise.
1650 * doc/tm.texi: Regenerate.
1651 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1652 * targhooks.c (default_get_reg_raw_mode): Likewise.
1653 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1654 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1655 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1656 (msp430_get_raw_result_mode): Likewise.
1657 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1658 * dbxout.c (dbxout_parms): Require fixed-size modes.
1659 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1660 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1661 * omp-low.c (lower_oacc_reductions): Likewise.
1662 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1663 (simplify_subreg): Update accordingly.
1664 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1665 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1666 that aren't fixed-size.
1667 (get_pool_mode): Return a fixed_size_mode.
1668 (output_constant_pool_2): Take a fixed_size_mode.
1670 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1671 Alan Hayward <alan.hayward@arm.com>
1672 David Sherwood <david.sherwood@arm.com>
1674 * doc/rtl.texi (vec_series): Document.
1675 (const): Say that the operand can be a vec_series.
1676 * rtl.def (VEC_SERIES): New rtx code.
1677 * rtl.h (const_vec_series_p_1): Declare.
1678 (const_vec_series_p): New function.
1679 * emit-rtl.h (gen_const_vec_series): Declare.
1680 (gen_vec_series): Likewise.
1681 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1682 (gen_vec_series): Likewise.
1683 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1684 * simplify-rtx.c (simplify_unary_operation): Handle negations
1686 (simplify_binary_operation_series): New function.
1687 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1688 (test_vector_ops_series): New function.
1689 (test_vector_ops): Call it.
1690 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1691 gen_const_vec_series.
1692 (altivec_lvsr): Likewise.
1693 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1695 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1696 Alan Hayward <alan.hayward@arm.com>
1697 David Sherwood <david.sherwood@arm.com>
1699 * doc/rtl.texi (const): Update description of address constants.
1700 Say that vector constants are allowed too.
1701 * common.md (E, F): Use CONSTANT_P instead of checking for
1703 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1704 checking for CONST_VECTOR.
1705 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1707 * expr.c (expand_expr_real_2): Check for vector modes instead
1708 of checking for CONST_VECTOR.
1709 * rtl.h (const_vec_p): New function.
1710 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1711 (unwrap_const_vec_duplicate): Handle them here too.
1713 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1714 David Malcolm <dmalcolm@redhat.com>
1715 Alan Hayward <alan.hayward@arm.com>
1716 David Sherwood <david.sherwood@arm.com>
1718 * rtl.h (vec_duplicate_p): New function.
1719 * selftest-rtl.c (assert_rtx_eq_at): New function.
1720 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1721 (assert_rtx_eq_at): Declare.
1722 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1723 * selftest-run-tests.c (selftest::run_tests): Call it.
1724 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1725 (simplify_unary_operation_1): Recursively handle vector duplicates.
1726 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1728 (simplify_subreg): Handle subregs of vector duplicates.
1729 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1730 (selftest::simplify_rtx_c_tests): New functions.
1732 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1733 Alan Hayward <alan.hayward@arm.com>
1734 David Sherwood <david.sherwood@arm.com>
1736 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1737 (gen_vec_duplicate): Likewise.
1738 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1740 (gen_const_vector): ...here.
1741 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1742 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1743 whose elements are all equal.
1744 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1745 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1746 (simplify_relational_operation): Likewise.
1747 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1749 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1750 (aarch64_expand_vector_init): Likewise.
1751 * config/arm/arm.c (neon_vdup_constant): Likewise.
1752 (neon_expand_vector_init): Likewise.
1753 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1754 (arm_block_set_unaligned_vect): Likewise.
1755 (arm_block_set_aligned_vect): Likewise.
1756 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1757 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1758 (expand_vec_perm_even_odd_pack): Likewise.
1759 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1760 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1761 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1762 gen_const_vec_duplicate.
1763 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1764 * config/mips/mips.c (mips_gen_const_int_vector): Use
1765 gen_const_vec_duplicate.
1766 (mips_expand_vector_init): Use CONST0_RTX.
1767 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1768 (define_split): Use gen_const_vec_duplicate.
1769 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1770 (define_split): Use gen_const_vec_duplicate.
1771 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1772 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1773 * config/spu/spu.c (spu_const): Likewise.
1775 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1776 Alan Hayward <alan.hayward@arm.com>
1777 David Sherwood <david.sherwood@arm.com>
1779 * combine.c (can_change_dest_mode): Reject changes in
1780 REGMODE_NATURAL_SIZE.
1782 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1784 * configure.ac (--enable-libssp): New.
1785 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1786 trying to determine target-specific default. Adjust indentation.
1787 * configure: Regenerated.
1788 * doc/install.texi (Configuration): Expand --disable-libssp
1791 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1793 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1796 2017-10-31 Martin Jambor <mjambor@suse.cz>
1799 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1801 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1803 * auto-profile.c (autofdo_source_profile::read): Use
1804 UNKNOWN_LOCATION rather than 0.
1805 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1806 (warning_at): ...this overload.
1807 (warning_at_rich_loc_n): Rename to...
1808 (warning_n): ...this overload.
1809 (error_at_rich_loc): Rename to...
1810 (error_at): ...this overload.
1811 (pedwarn_at_rich_loc): Rename to...
1812 (pedwarn): ...this overload.
1813 (permerror_at_rich_loc): Rename to...
1814 (permerror): ...this overload.
1815 (inform_at_rich_loc): Rename to...
1816 (inform): ...this overload.
1817 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1818 (diagnostic_n_impl_richloc): Rename to...
1819 (diagnostic_n_impl): ...this rich_location *-based decl.
1820 (inform_at_rich_loc): Rename to...
1821 (inform): ...this, and add an assertion.
1822 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1823 (warning_at_rich_loc): Rename to...
1824 (warning_at): ...this, and add an assertion.
1825 (warning_at_rich_loc_n): Rename to...
1826 (warning_n): ...this, and add an assertion.
1827 (warning_n): Update location_t-based implementation for removal of
1828 location_t-based diagnostic_n_impl.
1829 (pedwarn_at_rich_loc): Rename to...
1830 (pedwarn): ...this, and add an assertion.
1831 (permerror_at_rich_loc): Rename to...
1832 (permerror): ...this, and add an assertion.
1833 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1834 (error_at_rich_loc): Rename to...
1835 (error_at): ...this, and add an assertion.
1836 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1837 (driver::do_spec_on_infiles): Likewise.
1838 * substring-locations.c (format_warning_va): Update for renaming
1839 of inform_at_rich_loc.
1841 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1843 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1844 _Float<N>X built-in functions so that the variant without the
1845 "__builtin_" prefix is only enabled for the GNU C and Objective C
1846 languages when they are in non-strict ANSI/ISO mode.
1847 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1848 * target.def (floatn_builtin_p): Add a target hook to control
1849 whether _Float<N> and _Float<N>X built-in functions without the
1850 "__builtin_" prefix are enabled, and return true for C and
1851 Objective C in the default hook. Include langhooks.h in
1853 * targhooks.h (default_floatn_builtin_p): Likewise.
1854 * targhooks.c (default_floatn_builtin_p): Likewise.
1855 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1856 floatn_builtin_p target hook.
1857 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1859 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1860 Eric Botcazou <ebotcazou@adacore.com>
1862 PR rtl-optimization/81803
1863 * lra-constraints.c (curr_insn_transform): Also reload the whole
1864 register for a strict subreg no wider than a word if this is for
1865 a WORD_REGISTER_OPERATIONS target.
1867 2017-10-31 Jason Merrill <jason@redhat.com>
1869 * gdbinit.in: Skip over inlines from timevar.h.
1871 2017-10-31 Martin Liska <mliska@suse.cz>
1873 * doc/gcov.texi: Document new option.
1874 * gcov.c (print_usage): Likewise print it.
1875 (process_args): Support the argument.
1876 (format_count): New function.
1877 (format_gcov): Use the function.
1879 2017-10-31 Martin Liska <mliska@suse.cz>
1881 * gcov.c (struct name_map): do not use typedef.
1882 Define operator== and operator<.
1883 (name_search): Remove.
1884 (name_sort): Remove.
1885 (main): Do not allocate names.
1886 (process_file): Add vertical space.
1887 (generate_results): Use std::find.
1888 (release_structures): Do not release memory.
1889 (find_source): Use std::find.
1891 2017-10-31 Martin Liska <mliska@suse.cz>
1893 * gcov.c (struct line_info): Remove it's typedef.
1894 (line_info::line_info): Add proper ctor.
1895 (line_info::has_block): Do not use a typedef.
1896 (struct source_info): Do not use typedef.
1897 (circuit): Likewise.
1898 (get_cycles_count): Likewise.
1899 (output_intermediate_file): Iterate via vector iterator.
1900 (add_line_counts): Use std::vector methods.
1901 (accumulate_line_counts): Likewise.
1902 (output_lines): Likewise.
1904 2017-10-31 Martin Liska <mliska@suse.cz>
1906 * gcov.c (struct source_info): Remove typedef.
1907 (source_info::source_info): Add proper ctor.
1908 (accumulate_line_counts): Use struct, not it's typedef.
1909 (output_gcov_file): Likewise.
1910 (output_lines): Likewise.
1911 (main): Do not allocate an array.
1912 (output_intermediate_file): Use size of vector container.
1913 (process_file): Resize the vector.
1914 (generate_results): Do not preallocate, use newly added vector
1916 (release_structures): Do not release sources.
1917 (find_source): Use vector methods.
1918 (add_line_counts): Do not use typedef.
1920 2017-10-31 Martin Liska <mliska@suse.cz>
1922 * doc/gcov.texi: Document that.
1923 * gcov.c (add_line_counts): Mark lines with a non-executed
1925 (output_line_beginning): Handle such lines.
1926 (output_lines): Pass new argument.
1927 (output_intermediate_file): Print it in intermediate format.
1929 2017-10-31 Martin Liska <mliska@suse.cz>
1931 * color-macros.h: New file.
1932 * diagnostic-color.c: Factor out color related to macros to
1934 * doc/gcov.texi: Document -k option.
1935 * gcov.c (INCLUDE_STRING): Include string.h.
1936 (print_usage): Add -k option.
1937 (process_args): Parse it.
1938 (pad_count_string): New function.
1939 (output_line_beginning): Likewise.
1940 (DEFAULT_LINE_START): New macro.
1941 (output_lines): Support color output.
1943 2017-10-31 Martin Liska <mliska@suse.cz>
1945 PR gcov-profile/82633
1946 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1947 their interaction with GCOV infrastructure.
1948 * configure.ac: Add -fkeep-{inline,static}-functions to
1950 * configure: Regenerate.
1952 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1955 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1957 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1960 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1961 into a register if it does not fit into an immediate offset field.
1963 2017-10-31 Olivier Hainque <hainque@adacore.com>
1965 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1967 2017-10-31 Julia Koval <julia.koval@intel.com>
1969 * config.gcc: Add gfniintrin.h.
1970 * config/i386/gfniintrin.h: New.
1971 * config/i386/i386-builtin-types.def
1972 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1973 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1974 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1975 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1976 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1977 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1978 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1979 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1980 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1981 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1982 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1983 * config/i386/immintrin.h: Include gfniintrin.h.
1984 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1986 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1988 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1990 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1992 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1993 (arm_ashldi3_1bit): Remove pattern.
1994 (ashrdi3): Remove shift by 1 expansion.
1995 (arm_ashrdi3_1bit): Remove pattern.
1996 (lshrdi3): Remove shift by 1 expansion.
1997 (arm_lshrdi3_1bit): Remove pattern.
1998 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1999 cost of ashldi3 by 1.
2000 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2001 (<shift>di3_neon): Likewise.
2003 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2005 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2006 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2007 and (*aarch64_simd_mov<VQ:mode>).
2008 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2009 pattern alternative.
2010 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2011 attributes to match pattern alternative.
2013 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2015 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2016 * config/rs6000/emmintrin.h: New file.
2017 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2019 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2021 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2022 * (movdi_vfp_cortexa8): Remove pattern.
2024 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2026 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2028 (Specific, alpha*-dec-osf5.1): Remove.
2029 (Specific, mips-sgi-irix5): Remove.
2030 (Specific, mips-sgi-irix6): Remove.
2032 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2035 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2036 arguments to clear_bit_region_be.
2038 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2040 * gimplify.c: Include memmodel.h.
2042 2017-10-30 Martin Jambor <mjambor@suse.cz>
2044 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2045 condition whether loop should be executed at all.
2047 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2049 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2050 gimple folding of vec_madd() intrinsics.
2051 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2052 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2053 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2055 2017-10-30 Richard Biener <rguenther@suse.de>
2057 PR tree-optimization/82762
2059 2017-10-23 Richard Biener <rguenther@suse.de>
2061 PR tree-optimization/82129
2063 2017-08-01 Richard Biener <rguenther@suse.de>
2065 PR tree-optimization/81181
2066 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2067 (compute_antic): ... end of iteration here.
2069 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2071 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2072 -std=iso9899:2017 and -std=gnu17.
2073 * doc/standards.texi (C Language): Document C17 support.
2074 * doc/cpp.texi (Overview): Mention -std=c17.
2075 (Standard Predefined Macros): Document C11 and C17 values of
2076 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2077 * doc/extend.texi (Inline): Do not list individual options for
2078 standards newer than C99.
2079 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2081 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2084 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2086 * asan.c (asan_finish_file): Align asan globals array by shadow
2089 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2092 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2093 (struct store_immediate_info): Add bitregion_start and bitregion_end
2095 (store_immediate_info::store_immediate_info): Add brs and bre
2096 arguments and initialize bitregion_{start,end} from those.
2097 (struct merged_store_group): Add bitregion_start, bitregion_end,
2098 align_base and mask fields. Drop unnecessary struct keyword from
2099 struct store_immediate_info. Add do_merge method.
2100 (clear_bit_region_be): Use memset instead of loop storing zeros.
2101 (merged_store_group::do_merge): New method.
2102 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2103 stores as long as the surrounding bitregions have no gaps.
2104 (merged_store_group::merge_overlapping): Use do_merge.
2105 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2106 is byte aligned, rather than requiring that start and width are
2107 byte aligned. Drop unnecessary struct keyword from
2108 struct store_immediate_info. Allocate and populate also mask array.
2109 Make start of the arrays relative to bitregion_start rather than
2110 start and size them according to bitregion_{end,start} difference.
2111 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2112 struct store_immediate_info.
2113 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2114 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2115 struct keyword from struct store_immediate_info.
2116 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2117 between stores as long as the surrounding bitregions have no gaps.
2119 (struct split_store): Add orig non-static data member.
2120 (split_store::split_store): Initialize orig to false.
2121 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2122 if there is exactly a single original stmt. Change stmts argument
2123 to pointer from reference, if NULL, don't push anything to it. Add
2124 first argument, use it to optimize skipping over orig stmts that
2125 are known to be before bitpos already. Simplify.
2126 (split_group): Return unsigned int count how many stores are or
2127 would be needed rather than a bool. Add allow_unaligned argument.
2128 Change split_stores argument from reference to pointer, if NULL,
2129 only do a dry run computing how many stores would be produced.
2130 Rewritten algorithm to use both alignment and misalign if
2131 !allow_unaligned and handle bitfield stores with gaps.
2132 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2133 from bitregion_start instead of start. Compute allow_unaligned
2134 here, if true, do 2 split_group dry runs to compute which one
2135 produces fewer stores and prefer aligned if equal. Punt if
2136 new count is bigger or equal than original before emitting any
2137 statements, rather than during that. Remove no longer needed
2138 new_ssa_names tracking. Replace num_stmts with
2139 split_stores.length (). Use 32-bit stack allocated entries
2140 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2141 if possible. Handle bitfields with gaps.
2142 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2143 Compute bitregion_{start,end} for the stores and construct
2144 store_immediate_info with that. Formatting fixes.
2146 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2149 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2150 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2152 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2154 * gimplify.c: Include tm_p.h.
2156 * common.opt (gcoff): Re-add as ignored option.
2157 (gcoff1, gcoff2, gcoff3): Likewise.
2159 * Makefile.in (OBJS): Delete sdbout.o.
2160 (GTFILES): Delete $(srcdir)/sdbout.c.
2161 * debug.h: Delete sdb_debug_hooks.
2162 * final.c: Delete sdbout.h include.
2163 (final_scan_insn): Delete SDB_DEBUG check.
2164 (rest_of_clean_state): Likewise.
2165 * output.h: Delete sdb_begin_function_line.
2168 * toplev.c: Delete sdbout.h include.
2169 (process_options): Delete SDB_DEBUG check.
2170 * tree-core.h (tree_type_common): Delete pointer field of
2172 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2173 TYPE_SYMTAB_POINTER.
2174 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2175 (TYPE_SYMTAB_IS_POINTER): Delete.
2176 (TYPE_SYMTAB_IS_DIE): Renumber.
2177 * xcoffout.c: Refer to former sdbout.c file.
2178 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2180 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2181 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2182 (Debugging Options): Delete -gcoff.
2183 (-gstabs): Delete SDB reference.
2185 (-gcoff@var{level}): Delete.
2186 * doc/passes.texi (Debugging information output): Delete SDB and
2187 sdbout.c references.
2188 * doc/tm.texi: Regenerate.
2189 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2190 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2192 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2193 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2194 (SDB_DEBUGGING_INFO): Delete.
2195 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2196 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2197 * target.def (output_source_filename): Delete COFF reference.
2199 * common.opt (gcoff): Delete.
2200 (gxcoff+): Update Negative chain.
2201 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2203 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2204 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2205 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2207 (expand_function_start): Change sdb reference to past tense.
2208 (expand_function_end): Change sdb reference to past tense.
2209 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2210 * opts.c (debug_type_names): Delete coff entry.
2211 (common_handle_option): Delete OPT_gcoff case.
2212 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2214 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2215 * config/cris/cris.h: Delete SDB reference in comment.
2216 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2217 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2218 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2219 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2221 (ix86_expand_prologue): Likewise.
2222 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2223 * config/ia64/ia64.h: Likewise.
2224 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2225 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2227 * config/mmix/mmix.h: Likewise.
2228 * config/nds32/nds32.c: Likewise.
2229 * config/stormy/storym16.h: Likewise.
2230 * config/visium/visium.h: Likewise.
2231 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2233 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2235 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2236 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2237 FRAME_POINTER_REGNUM point at high end of local var area.
2239 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2241 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2242 Move comment around. Do not reset best_edge for a copiable
2243 destination if the copy would cause a partition change.
2244 (better_edge_p): Remove redundant check.
2246 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2248 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2250 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2252 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2253 for math functions that have _Float<N> and _Float<N>X variants.
2254 (mathfn_built_in_2): Add support for math functions that have
2255 _Float<N> and _Float<N>X variants.
2256 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2257 (expand_builtin_mathfn_ternary): Add support for fma with
2258 _Float<N> and _Float<N>X variants.
2259 (expand_builtin): Likewise.
2260 (fold_builtin_3): Likewise.
2261 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2262 create math function _Float<N> and _Float<N>X variants as external
2264 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2265 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2266 the __builtin_ prefix and if not strict ansi, without the prefix.
2267 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2268 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2269 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2270 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2271 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2272 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2273 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2274 function signatures for fma _Float<N> and _Float<N>X variants.
2275 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2276 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2277 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2278 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2279 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2280 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2281 * gencfn-macros.c (print_case_cfn): Add support for math functions
2282 that have _Float<N> and _Float<N>X variants.
2283 (print_define_operator_list): Likewise.
2284 (fltfn_suffixes): Likewise.
2286 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2287 for math functions that have _Float<N> and _Float<N>X variants.
2288 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2289 and _Float<N>X variants.
2290 (COPYSIGN): Likewise.
2293 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2294 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2296 (integer_valued_read_call_p): Likewise.
2297 * fold-const-call.c (fold_const_call_ss): Likewise.
2298 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2299 _Float<N> and _Float<N>X variants.
2300 (fold_const_call_ssss): Add support for fma _Float<N> and
2301 _Float<N>X variants.
2302 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2303 support for copysign and fma _Float<N> and _Float<N>X variants.
2304 (backprop::process_builtin_call_use): Likewise.
2305 * tree-call-cdce.c (can_test_argument_range); Add support for
2306 sqrt _Float<N> and _Float<N>X variants.
2307 (edom_only_function): Likewise.
2308 (get_no_error_domain): Likewise.
2309 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2310 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2311 copysign _Float<N> and _Float<N>X variants.
2312 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2313 handled by machine independent code.
2314 (FMAF128): Likewise.
2315 * doc/cpp.texi (Common Predefined Macros): Document defining
2316 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2317 fma _Float<N> and _Float<N>X variants.
2319 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2322 * config/i386/i386-modes.def (CCFPU): Remove definition.
2323 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2324 (ix86_cc_modes_compatible): Ditto.
2325 (ix86_expand_carry_flag_compare): Ditto.
2326 (ix86_expand_int_movcc): Ditto.
2327 (ix86_expand_int_addcc): Ditto.
2328 (ix86_reverse_condition): Ditto.
2329 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2330 Return true/false for unordered/ordered fp comparisons.
2331 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2332 (ix86_prepare_fp_compare_args): Update for rename.
2333 (ix86_expand_fp_compare): Update for rename. Generate unordered
2334 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2335 (ix86_expand_sse_compare_and_jump): Ditto.
2336 * config/i386/predicates.md (fcmov_comparison_operator):
2337 Remove CCFPU mode handling.
2338 (ix86_comparison_operator): Ditto.
2339 (ix86_carry_flag_operator): Ditto.
2340 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2341 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2342 (*cmpu<mode>_cc_i387): Ditto.
2343 (FPCMP): Remove mode iterator.
2344 (unord): Remove mode attribute.
2345 (unord_subst): New define_subst transformation
2346 (unord): New define_subst attribute.
2348 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2349 (*cmpi<unord>xf_i387): Ditto.
2350 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2351 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2352 using unord_subst transformation.
2353 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2354 (round_saeonly): Also handle CCFP mode.
2355 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2356 Remove UNSPEC_SAHF unspec handling.
2358 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2360 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2362 2017-10-27 Jeff Law <law@redhat.com>
2364 * gimple-ssa-sprintf.c: Include domwalk.h.
2365 (class sprintf_dom_walker): New class, derived from dom_walker.
2366 (sprintf_dom_walker::before_dom_children): New function.
2367 (struct call_info): Moved into sprintf_dom_walker class
2368 (compute_formath_length, handle_gimple_call): Likewise.
2369 (sprintf_length::execute): Call the dominator walker rather
2370 than walking the statements.
2372 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2373 gimple statement locations.
2374 (check_array_bounds): Corresponding changes. Get the statement's
2375 location directly from wi->stmt.
2377 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2380 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2382 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2384 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2385 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2387 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2390 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2391 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2392 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2393 maybe_get_pool_constant.
2394 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2397 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2399 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2400 2.26 caveat. Update gas and gld versions.
2401 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2404 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2406 * cgraph.h (set_malloc_flag): Declare.
2407 * cgraph.c (set_malloc_flag_1): New function.
2408 (set_malloc_flag): Likewise.
2409 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2410 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2412 (read_ipa_call_summary): Add support for reading is_return_callee.
2413 (write_ipa_call_summary): Stream is_return_callee.
2414 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2415 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2416 ipa-prop.h, ipa-fnsummary.h.
2417 (pure_const_names): Change to static.
2418 (malloc_state_e): Define.
2419 (malloc_state_names): Define.
2420 (funct_state_d): Add field malloc_state.
2421 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2422 (check_retval_uses): New function.
2423 (malloc_candidate_p): Likewise.
2424 (analyze_function): Add support for malloc attribute.
2425 (pure_const_write_summary): Stream malloc_state.
2426 (pure_const_read_summary): Add support for reading malloc_state.
2427 (dump_malloc_lattice): New function.
2428 (propagate_malloc): New function.
2429 (warn_function_malloc): New function.
2430 (ipa_pure_const::execute): Call propagate_malloc and
2431 ipa_free_fn_summary.
2432 (pass_local_pure_const::execute): Add support for malloc attribute.
2433 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2434 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2436 2017-10-27 Martin Liska <mliska@suse.cz>
2438 PR gcov-profile/82457
2439 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2440 for fork-like functions to be properly instrumented.
2442 2017-10-27 Richard Biener <rguenther@suse.de>
2445 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2446 info when we redirected EH.
2448 2017-10-26 Michael Collison <michael.collison@arm.com>
2450 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2452 (<optab>_trunchf<GPI:mode>2: New pattern.
2453 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2454 * config/aarch64/iterators.md (wv): New mode attribute.
2455 (vf, VF): New mode attributes.
2456 (vgp, VGP): New mode attributes.
2457 (s): Update attribute with SImode and DImode prefixes.
2459 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2461 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2462 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2463 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2464 (nios2_option_overide): Initialize it. Don't allow R0-relative
2465 addressing with PIC.
2466 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2467 (nios2_symbolic_constant_p): Likewise.
2468 (nios2_legitimate_address_p): Likewise.
2469 (nios2_r0rel_section_name_p): New.
2470 (nios2_symbol_ref_in_r0rel_data_p): New.
2471 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2472 (r0rel_constant_p): New.
2473 (nios2_print_operand_address): Handle r0rel_constant_p.
2474 (nios2_cdx_narrow_form_p): Likewise.
2475 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2476 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2477 (Nios II Options): Document -mr0rel-sec.
2479 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2481 * config/nios2/nios2.c: Include xregex.h.
2482 (nios2_gprel_sec_regex): New.
2483 (nios2_option_overide): Initialize it. Don't allow GP-relative
2484 addressing with PIC.
2485 (nios2_small_section_name_p): Check for regex match.
2486 * config/nios2/nios2.opt (mgprel-sec=): New option.
2487 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2488 (Nios II Options): Document -mgprel-sec.
2490 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2492 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2494 2017-10-26 Tom de Vries <tom@codesourcery.com>
2496 PR tree-optimization/82707
2497 * gimple.c (gimple_copy): Fix unsharing of
2498 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2500 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2502 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2503 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2504 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2505 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2506 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2507 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2508 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2509 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2510 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2511 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2512 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2513 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2514 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2515 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2516 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2517 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2518 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2520 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2522 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2524 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2525 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2526 warning to rs6000.c. Remove the Undocumented flag, since it has
2528 (-mabi=ibmlongdouble): Likewise.
2529 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2530 already set, set the default format for long double.
2531 (rs6000_debug_reg_global): Print whether long double is IBM or
2533 (rs6000_option_override_internal): Rework setting long double
2534 format. Only warn if the user is changing the long double default
2535 and they did not use -Wno-psabi.
2536 * doc/invoke.texi (PowerPC options): Update the documentation for
2537 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2539 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2540 Alan Hayward <alan.hayward@arm.com>
2541 David Sherwood <david.sherwood@arm.com>
2543 * rtl.h (wider_subreg_mode): New function.
2544 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2545 rather than an unsigned int *.
2546 * ira-color.c (regno_max_ref_width): Replace with...
2547 (regno_max_ref_mode): ...this new variable.
2548 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2549 Use wider_subreg_mode.
2550 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2551 rather than an unsigned int *.
2552 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2553 (process_alt_operands): Likewise.
2554 (invariant_p): Likewise.
2555 * lra-spills.c (assign_mem_slot): Likewise.
2556 (add_pseudo_to_slot): Likewise.
2557 * lra.c (collect_non_operand_hard_regs): Likewise.
2558 (add_regs_to_insn_regno_info): Likewise.
2559 * reload1.c (regno_max_ref_width): Replace with...
2560 (regno_max_ref_mode): ...this new variable.
2561 (reload): Update accordingly. Update call to
2562 ira_sort_regnos_for_alter_reg.
2563 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2564 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2565 (scan_paradoxical_subregs): Likewise.
2567 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2569 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2570 (aarch64_frame): Add emit_frame_chain boolean.
2571 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2572 Move eh_return case to aarch64_layout_frame.
2573 (aarch64_layout_frame): Initialize emit_frame_chain.
2574 (aarch64_expand_prologue): Use emit_frame_chain.
2576 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2578 * config/aarch64/aarch64.c (aarch64_layout_frame):
2579 Ensure LR is always stored at the bottom of the callee-saves.
2580 Remove rarely used frame layout which saves callee-saves at top of
2581 frame, so the store of LR can be used as a valid probe in all cases.
2583 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2585 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2586 Improve unaligned TImode/TFmode base/offset split.
2588 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2589 Alan Hayward <alan.hayward@arm.com>
2590 David Sherwood <david.sherwood@arm.com>
2592 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2593 * combine.c (find_single_use_1): Likewise.
2594 (expand_field_assignment): Likewise.
2595 (move_deaths): Likewise.
2596 * lra-constraints.c (simplify_operand_subreg): Likewise.
2597 (curr_insn_transform): Likewise.
2598 * lra.c (collect_non_operand_hard_regs): Likewise.
2599 (add_regs_to_insn_regno_info): Likewise.
2600 * rtlanal.c (reg_referenced_p): Likewise.
2601 (covers_regno_no_parallel_p): Likewise.
2603 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2605 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2606 Don't print any bits outside the precision of the value.
2607 * wide-int.cc (test_printing): Add some new tests.
2609 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2611 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2612 supports -xbrace_comment option.
2613 * configure: Regenerate.
2614 * config.in: Regenerate.
2615 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2616 (ASM_CPU_SPEC): Use it.
2618 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2620 * target.def (static_rtx_alignment): New hook.
2621 * targhooks.h (default_static_rtx_alignment): Declare.
2622 * targhooks.c (default_static_rtx_alignment): New function.
2623 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2624 * doc/tm.texi: Regenerate.
2625 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2626 instead of targetm.constant_alignment. Remove call to
2628 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2629 (cris_preferred_mininum_alignment): New function, split out from...
2630 (cris_constant_alignment): ...here.
2631 (cris_static_rtx_alignment): New function.
2632 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2634 (ix86_constant_alignment): ...here.
2635 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2636 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2637 (mmix_static_rtx_alignment): New function.
2638 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2639 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2641 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2644 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2645 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2647 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2649 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2650 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2651 mark as a sign-extending load.
2652 (local_pic_load_u): Define.
2654 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2657 * fold-const.c (operand_equal_for_comparison_p): Also return true
2658 if ARG0 is a simple variant of ARG1 with narrower precision.
2659 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2661 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2663 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2665 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2666 scatter_static, scatter_per_elt.
2667 * x86-tune-costs.h: Add new cost entries.
2669 2017-10-25 Richard Biener <rguenther@suse.de>
2671 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2672 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2673 class pass_fre): Move to ...
2674 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2675 class pass_fre): ... here and adjust for statistics.
2677 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2680 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2681 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2682 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2683 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2686 2017-10-25 Richard Biener <rguenther@suse.de>
2688 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2689 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2690 eliminate_push_avail, eliminate_insert): Move inside...
2691 (class eliminate_dom_walker): ... this class in preparation
2693 (fini_eliminate): Remove by merging with ...
2694 (eliminate): ... this function. Adjust for class changes.
2695 (pass_pre::execute): Remove fini_eliminate call.
2696 (pass_fre::execute): Likewise.
2698 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2701 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2702 (VPERMI2, VPERMI2I): New mode iterators.
2703 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2704 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2706 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2707 mode iterator. Remove 3 old define_insn patterns.
2708 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2709 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2710 VPERMI2 mode iterator, remove the other two expanders.
2711 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2712 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2713 instructions, remove the other two patterns.
2714 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2715 mode iterator, remove the other two patterns.
2716 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2717 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2718 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2719 and adjust argument order accordingly.
2720 (ix86_expand_vec_perm): Adjust caller.
2721 (expand_vec_perm_1): Likewise.
2722 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2723 (expand_vec_perm_vpermt2_vpshub2): ... this.
2724 (ix86_expand_vec_perm_const_1): Adjust caller.
2725 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2728 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2729 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2730 (vec_shl_<mode>): Remove unused expander.
2731 (avx512bw_<shift_insn><mode>3): New define_insn.
2732 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2733 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2735 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2738 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2741 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2743 PR rtl-optimization/82396
2744 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2745 (autopref_multipass_init): Simplify initialization.
2746 (autopref_rank_data): Simplify sort order.
2747 * gcc/sched-int.h (autopref_multipass_data_): Remove
2748 multi_mem_insn_p, min_offset and max_offset.
2750 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2753 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2754 Check special value of flag_omit_frame_pointer.
2755 (aarch64_can_eliminate): Likewise.
2756 (aarch64_override_options_after_change_1): Simplify handling of
2757 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2759 2017-10-24 Richard Biener <rguenther@suse.de>
2761 PR tree-optimization/82697
2762 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2763 zero for conditional load and unconditional store.
2765 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2767 * doc/install.texi: Document bootstrap-cet.
2769 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2772 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2773 ENDBR instruction at function entrance if function is only
2776 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2779 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2780 patterns to better describe from which operation the CF is computed.
2781 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2782 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2783 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2784 is 0, use _0 suffixed expanders instead of emitting a comparison
2787 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2789 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2790 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2792 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2795 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2796 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2797 * loop-iv.c (iv_get_reaching_def): Likewise.
2798 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2799 variable is promoted and the partition contains undefined values.
2801 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2803 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2805 (nios2_address_cost): Define.
2806 (nios2_legitimize_address): Recognize (exp + constant) directly.
2807 (TARGET_ADDRESS_COST): Define.
2809 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2811 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2812 (nios2_symbolic_memory_operand_p): Declare.
2813 (nios2_split_large_constant): Declare.
2814 (nios2_split_symbolic_memory_operand): Declare.
2815 * config/nios2/nios2.c: Adjust includes.
2816 (nios2_symbolic_constant_allowed): New.
2817 (nios2_symbolic_constant_p): New.
2818 (nios2_plus_symbolic_constant_p): New.
2819 (nios2_valid_addr_expr_p): Recognize addresses involving
2821 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2822 (nios2_symbolic_memory_operand_p): New.
2823 (nios2_large_constant_p): New.
2824 (nios2_split_large_constant): New.
2825 (nios2_split_plus_large_constant): New.
2826 (nios2_split_symbolic_memory_operand): New.
2827 (nios2_legitimize_address): Code refactoring. Handle addresses
2828 involving symbolic constants.
2829 (nios2_emit_move_sequence): Likewise.
2830 (nios2_print_operand): Improve error output.
2831 (nios2_print_operand_address): Handle LO_SUM.
2832 (nios2_cdx_narrow_form_p): Likewise.
2833 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2834 operands involving symbolic constants.
2835 (movhi_internal, movsi_internal): Likewise.
2836 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2837 (extendhisi2, extendqi<mode>2): Likewise.
2839 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2841 * tree-pass.h (PROP_rtl_split_insns): Define.
2842 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2844 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2846 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2848 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2851 * target.def (const_not_ok_for_debug_p): Default to
2852 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2853 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2854 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2855 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2856 which targetm.const_not_ok_for_debug_p returned true.
2857 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2859 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2861 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2862 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2863 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2865 (ix86_const_not_ok_for_debug_p): New function.
2866 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2867 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2869 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2872 * system.h: Conditionally include "unique-ptr.h" if
2873 INCLUDE_UNIQUE_PTR is defined.
2874 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2875 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2877 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2879 * config/rl78/rl78.md: New define_expand "subdi3".
2881 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2884 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2885 DF_REF_INSN if DF_REF_INSN_INFO is false.
2887 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2889 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2890 xmm_move instead of sse_move.
2891 (sse_store_index): New function.
2892 (ix86_register_move_cost): Be more sensible about mismatch stall;
2893 model AVX moves correctly; make difference between sse->integer and
2895 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2896 moves; make difference between SSE and AVX.
2897 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2898 and zmm_move. Increase size of sse load and store tables;
2899 add unaligned load and store tables; add ssemmx_to_integer.
2900 * x86-tune-costs.h: Update all entries according to real
2901 move latencies from Agner Fog's manual and chip documentation.
2903 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2906 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2907 * config/i386/constraints.md (Wf): New constraint.
2908 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2909 (cmp<dwi>_doubleword): Removed.
2910 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2911 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2912 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2913 expand with cmp<dwi>_doubleword. For LTU and GEU use
2914 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2916 * common.opt (gcolumn-info): Enable by default.
2917 * doc/invoke.texi (gcolumn-info): Document new default.
2919 2017-10-23 Richard Biener <rguenther@suse.de>
2921 PR tree-optimization/82672
2922 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2923 Fold the stmt if we propagated into it.
2925 2017-10-23 Richard Biener <rguenther@suse.de>
2927 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2928 (bitmap_remove_expr_from_set): ... this. All callers call this
2929 for non-constant values.
2930 (bitmap_set_subtract): Rename to...
2931 (bitmap_set_subtract_expressions): ... this. Adjust and
2933 (bitmap_set_contains_value): Remove superfluous check.
2934 (bitmap_set_replace_value): Inline into single caller ...
2935 (bitmap_value_replace_in_set): ... here and simplify.
2936 (dependent_clean): Merge into ...
2937 (clean): ... this using an overload. Adjust.
2938 (prune_clobbered_mems): Adjust.
2939 (compute_antic_aux): Likewise.
2940 (compute_partial_antic_aux): Likewise.
2942 2017-10-23 Richard Biener <rguenther@suse.de>
2944 PR tree-optimization/82129
2946 2017-08-01 Richard Biener <rguenther@suse.de>
2948 PR tree-optimization/81181
2949 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2950 (compute_antic): ... end of iteration here.
2952 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2954 * target.def (starting_frame_offset): New hook.
2955 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2956 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2957 * doc/tm.texi.in: Regenerate.
2958 * hooks.h (hook_hwi_void_0): Declare.
2959 * hooks.c (hook_hwi_void_0): New function.
2960 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2961 STARTING_FRAME_OFFSET.
2962 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2963 * reload1.c (reload): Likewise.
2964 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2965 instead of STARTING_FRAME_OFFSET.
2966 * function.c (try_fit_stack_local): Likewise.
2967 (assign_stack_local_1): Likewise
2968 (instantiate_virtual_regs): Likewise.
2969 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2970 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2971 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2972 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2973 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2974 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2975 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2976 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2977 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2978 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2979 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2980 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2981 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2982 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2983 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2984 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2985 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2986 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2987 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2988 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2989 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2990 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2991 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2992 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2993 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2994 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2995 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2996 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2997 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2998 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2999 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3000 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3001 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3002 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3003 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3004 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3005 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3006 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3007 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3008 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3009 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3010 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3011 return a HOST_WIDE_INT.
3012 (avr_builtin_setjmp_frame_value): Use it instead of
3013 STARTING_FRAME_OFFSET.
3014 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3015 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3016 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3018 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3019 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3020 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3021 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3022 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3023 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3024 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3025 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3026 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3027 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3028 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3029 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3031 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3032 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3033 * config/mips/mips.c (mips_compute_frame_info): Refer to
3034 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3035 (mips_starting_frame_offset): New function.
3036 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3037 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3038 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3039 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3040 and return a HOST_WIDE_INT.
3041 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3042 (mmix_initial_elimination_offset): Refer to
3043 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3044 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3045 * config/pa/pa.c (pa_starting_frame_offset): New function.
3046 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3047 (pa_expand_prologue): Likewise.
3048 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3049 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3050 !FRAME_GROWS_DOWNWARD handling to...
3051 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3052 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3053 !FRAME_GROWS_DOWNWARD handling to...
3054 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3055 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3056 !FRAME_GROWS_DOWNWARD handling to...
3057 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3058 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3060 (rs6000_starting_frame_offset): New function.
3061 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3062 !FRAME_GROWS_DOWNWARD handling to...
3063 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3064 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3065 !FRAME_GROWS_DOWNWARD handling to...
3066 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3067 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3068 !FRAME_GROWS_DOWNWARD handling to...
3069 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3070 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3071 (rs6000_starting_frame_offset): New function.
3072 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3073 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3074 * config/vax/vax.c (vax_starting_frame_offset): New function.
3075 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3076 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3077 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3078 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3079 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3080 * system.h (STARTING_FRAME_OFFSET): Poison.
3082 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3084 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3085 SCALAR_TYPE_MODE instead of TYPE_MODE.
3087 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3088 Alan Hayward <alan.hayward@arm.com>
3089 David Sherwood <david.sherwood@arm.com>
3091 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3093 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3094 Alan Hayward <alan.hayward@arm.com>
3095 David Sherwood <david.sherwood@arm.com>
3097 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3099 2017-10-23 Richard Biener <rguenther@suse.de>
3101 PR tree-optimization/82129
3102 * tree-ssa-pre.c (bitmap_set_and): Remove.
3103 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3104 canonicalizing expressions in the set to those with lowest
3105 ID rather than taking that from the first edge.
3107 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3109 * combine.c (rtx_equal_for_field_assignment_p): Use
3110 byte_lowpart_offset.
3112 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3113 Alan Hayward <alan.hayward@arm.com>
3114 David Sherwood <david.sherwood@arm.com>
3116 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3117 to a SUBREG_PROMOTED_VAR.
3119 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3120 Alan Hayward <alan.hayward@arm.com>
3121 David Sherwood <david.sherwood@arm.com>
3123 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3124 (expand_debug_source_expr): Likewise.
3125 * combine.c (combine_simplify_rtx): Likewise.
3126 * cse.c (fold_rtx): Likewise.
3127 * optabs.c (expand_float): Likewise.
3128 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3129 (simplify_binary_operation_1): Likewise.
3131 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3132 Alan Hayward <alan.hayward@arm.com>
3133 David Sherwood <david.sherwood@arm.com>
3135 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3136 (record_promoted_value): Likewise.
3137 * expr.c (expand_expr_real_2): Likewise.
3138 * ree.c (update_reg_equal_equiv_notes): Likewise.
3139 (combine_set_extension): Likewise.
3140 * rtlanal.c (low_bitmask_len): Likewise.
3141 * simplify-rtx.c (neg_const_int): Likewise.
3142 (simplify_binary_operation_1): Likewise.
3144 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3145 Alan Hayward <alan.hayward@arm.com>
3146 David Sherwood <david.sherwood@arm.com>
3148 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3149 * regcprop.c (maybe_mode_change): Likewise.
3150 * reload1.c (alter_reg): Likewise.
3152 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3154 * inchash.h (inchash::hash::add_wide_int): New function.
3155 * lto-streamer-out.c (hash_tree): Use it.
3157 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3159 * inchash.h (inchash::hash::add_wide_int): Rename to...
3160 (inchash::hash::add_hwi): ...this.
3161 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3162 (polymorphic_call_target_hasher::hash): Likewise.
3163 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3164 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3165 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3166 * lto-streamer-out.c (hash_tree): Likewise.
3167 * optc-save-gen.awk: Likewise.
3168 * tree.c (add_expr): Likewise.
3170 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3173 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3174 for ordered inequality comparisons even with TARGET_IEEE_FP.
3176 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3179 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3180 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3181 Expand with cmp<dwi>_doubleword.
3183 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3185 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3186 List CET intrinsics.
3187 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3188 specific to -fcf-protection option.
3190 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3192 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3193 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3194 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3195 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3196 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3197 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3198 (extra_objs): Add cet.o for Linux/x86 targets.
3199 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3200 * config/i386/cet.c: New file.
3201 * config/i386/cetintrin.h: Likewise.
3202 * config/i386/t-cet: Likewise.
3203 * config/i386/cpuid.h (bit_SHSTK): New.
3204 (bit_IBT): Likewise.
3205 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3206 pass IBT and SHSTK bits.
3207 * config/i386/i386-builtin-types.def
3208 (VOID_FTYPE_UNSIGNED_PVOID): New.
3209 (VOID_FTYPE_UINT64_PVOID): Likewise.
3210 * config/i386/i386-builtin.def: Add CET intrinsics.
3211 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3212 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3213 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3214 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3216 * config/i386/i386.c (rest_of_insert_endbranch): New.
3217 (pass_data_insert_endbranch): Likewise.
3218 (pass_insert_endbranch): Likewise.
3219 (make_pass_insert_endbranch): Likewise.
3220 (ix86_notrack_prefixed_insn_p): Likewise.
3221 (ix86_target_string): Add -mibt, -mshstk flags.
3222 (ix86_option_override_internal): Add flag_cf_protection
3224 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3225 (ix86_print_operand): Add 'notrack' prefix output.
3226 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3227 (ix86_expand_builtin): Expand CET intrinsics.
3228 (x86_output_mi_thunk): Add 'endbranch' instruction.
3229 * config/i386/i386.h (TARGET_IBT): New.
3230 (TARGET_IBT_P): Likewise.
3231 (TARGET_SHSTK): Likewise.
3232 (TARGET_SHSTK_P): Likewise.
3233 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3234 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3235 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3236 (builtin_setjmp_setup): New pattern.
3237 (builtin_longjmp): Likewise.
3238 (rdssp<mode>): Likewise.
3239 (incssp<mode>): Likewise.
3240 (saveprevssp): Likewise.
3241 (rstorssp): Likewise.
3242 (wrss<mode>): Likewise.
3243 (wruss<mode>): Likewise.
3244 (setssbsy): Likewise.
3245 (clrssbsy): Likewise.
3246 (nop_endbr): Likewise.
3247 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3249 * config/i386/immintrin.h: Include <cetintrin.h>.
3250 * config/i386/linux-common.h
3251 (file_end_indicate_exec_stack_and_cet): New prototype.
3252 (TARGET_ASM_FILE_END): New.
3254 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3256 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3257 latencies instead of having separate table; make difference between
3258 integer and float costs.
3259 * i386.h (processor_costs): Remove scalar_stmt_cost,
3260 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3261 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3263 * x86-tune-costs.h: Remove entries which has been removed in
3264 procesor_costs from all tables; make cond_taken_branch_cost
3265 and cond_not_taken_branch_cost COST_N_INSNS based.
3267 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3269 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3271 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3273 * config/i386/i386.md (isa): Remove fma_avx512f.
3274 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3275 <avx512>_fmadd_<mode>_mask3<round_name>,
3276 <avx512>_fmsub_<mode>_mask<round_name>,
3277 <avx512>_fmsub_<mode>_mask3<round_name>,
3278 <avx512>_fnmadd_<mode>_mask<round_name>,
3279 <avx512>_fnmadd_<mode>_mask3<round_name>,
3280 <avx512>_fnmsub_<mode>_mask<round_name>,
3281 <avx512>_fnmsub_<mode>_mask3<round_name>,
3282 <avx512>_fmaddsub_<mode>_mask<round_name>,
3283 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3284 <avx512>_fmsubadd_<mode>_mask<round_name>,
3285 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3286 (*vec_widen_umult_even_v16si<mask_name>,
3287 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3288 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3290 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3292 * extend.texi: Add 'nocf_check' documentation.
3293 * gimple.texi: Add second parameter to
3294 gimple_build_call_from_tree.
3295 * invoke.texi: Add -fcf-protection documentation.
3296 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3298 2017-10-20 Richard Biener <rguenther@suse.de>
3300 PR tree-optimization/82473
3301 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3302 the largest input type.
3304 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3306 * c-attribs.c (handle_nocf_check_attribute): New function.
3307 (c_common_attribute_table): Add 'nocf_check' handling.
3308 * gimple-parser.c: Add second argument NULL to
3309 gimple_build_call_from_tree.
3310 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3311 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3313 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3314 * common.opt: Add fcf-protection flag.
3315 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3316 * flag-types.h: Add enum cf_protection_level.
3317 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3318 Add 'nocf_check' attribute propagation to gimple call.
3319 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3320 (gimple_build_call_from_tree): Update prototype.
3321 (gimple_call_nocf_check_p): New function.
3322 (gimple_call_set_nocf_check): Likewise.
3323 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3324 * ipa-icf.c: Add nocf_check attribute in statement hash.
3325 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3326 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3327 * toplev.c (process_options): Add flag_cf_protection handling.
3329 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3331 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3333 2017-10-20 Richard Biener <rguenther@suse.de>
3335 PR tree-optimization/82603
3336 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3337 remove false predicated stores.
3339 2017-10-20 Richard Biener <rguenther@suse.de>
3341 * graphite-isl-ast-to-gimple.c
3342 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3343 Remove return value and simplify, dump copied stmt after lhs
3345 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3346 Reduce dump verbosity.
3347 (gsi_insert_earliest): Likewise.
3348 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3349 * graphite.c (print_global_statistics): Adjust dumping.
3350 (print_graphite_scop_statistics): Likewise.
3351 (print_graphite_statistics): Do not dump loops here.
3352 (graphite_transform_loops): But here.
3354 2017-10-20 Nicolas Roche <roche@adacore.com>
3356 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3357 * configure: Regenerate.
3359 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3362 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3363 functions when optimizing replace GIMPLE_RETURN stmts with
3364 calls to __builtin_unreachable ().
3367 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3368 for -fsanitize=thread link of executables.
3369 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3370 link of executables.
3373 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3375 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3376 define_insns for logical vector shifts to use VI248_AVX512BW
3377 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3378 condition, useless isa and prefix attributes. Change the first
3379 2 of these define_insns to ...
3380 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3381 define_insn for avx512vl.
3382 (<shift_insn><mode>3): ... and this, new define_insn without
3383 masking for non-avx512vl.
3386 * config/i386/sse.md (*andnot<mode>3,
3387 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3388 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3389 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3390 not applied use empty suffix even for TARGET_AVX512VL.
3391 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3392 is applied, supply evex,evex or evex,evex,evex instead of just
3395 2017-10-20 Julia Koval <julia.koval@intel.com>
3397 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3398 (OPTION_MASK_ISA_GFNI_UNSET): New.
3399 (ix86_handle_option): Handle OPT_mgfni.
3400 * config/i386/cpuid.h (bit_GFNI): New.
3401 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3402 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3403 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3404 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3405 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3406 * config/i386/i386.opt: Add mgfni.
3408 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3410 * config/msp430/msp430.c (msp430_option_override): Disable
3411 -fdelete-null-pointer-checks.
3412 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3414 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3416 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3417 of x87 and SSE instructions.
3419 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3421 * asan.c (create_cond_insert_point): Do not update edge count.
3422 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3423 (afdo_propagate_circuit): Likewise.
3424 (afdo_calculate_branch_prob): Likewise.
3425 (afdo_annotate_cfg): Likewise.
3426 * basic-block.h (struct edge_def): Remove count.
3427 (edge_def::count): New accessor.
3428 * bb-reorder.c (rotate_loop): Update.
3429 (find_traces_1_round): Update.
3430 (connect_traces): Update.
3431 (sanitize_hot_paths): Update.
3432 * cfg.c (unchecked_make_edge): Update.
3433 (make_single_succ_edge): Update.
3434 (check_bb_profile): Update.
3435 (dump_edge_info): Update.
3436 (update_bb_profile_for_threading): Update.
3437 (scale_bbs_frequencies_int): Update.
3438 (scale_bbs_frequencies_gcov_type): Update.
3439 (scale_bbs_frequencies_profile_count): Update.
3440 (scale_bbs_frequencies): Update.
3441 * cfganal.c (connect_infinite_loops_to_exit): Update.
3442 * cfgbuild.c (compute_outgoing_frequencies): Update.
3443 (find_many_sub_basic_blocks): Update.
3444 * cfgcleanup.c (try_forward_edges): Update.
3445 (try_crossjump_to_edge): Update
3446 * cfgexpand.c (expand_gimple_cond): Update
3447 (expand_gimple_tailcall): Update
3448 (construct_exit_block): Update
3449 * cfghooks.c (verify_flow_info): Update
3450 (redirect_edge_succ_nodup): Update
3451 (split_edge): Update
3452 (make_forwarder_block): Update
3453 (duplicate_block): Update
3454 (account_profile_record): Update
3455 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3456 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3457 * cfgloopmanip.c (scale_loop_profile): Update.
3459 (lv_adjust_loop_entry_edge): Update.
3460 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3461 (force_nonfallthru_and_redirect): Update.
3462 (purge_dead_edges): Update.
3463 (rtl_flow_call_edges_add): Update.
3464 * cgraphunit.c (init_lowered_empty_function): Update.
3465 (cgraph_node::expand_thunk): Update.
3466 * gimple-pretty-print.c (dump_probability): Update.
3467 (dump_edge_probability): Update.
3468 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3469 * haifa-sched.c (sched_create_recovery_edges): Update.
3470 * hsa-gen.c (convert_switch_statements): Update.
3471 * ifcvt.c (dead_or_predicable): Update.
3472 * ipa-inline-transform.c (inline_transform): Update.
3473 * ipa-split.c (split_function): Update.
3474 * ipa-utils.c (ipa_merge_profiles): Update.
3475 * loop-doloop.c (add_test): Update.
3476 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3477 * lto-streamer-in.c (input_cfg): Update.
3478 (input_function): Update.
3479 * lto-streamer-out.c (output_cfg): Update.
3480 * modulo-sched.c (sms_schedule): Update.
3481 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3482 * predict.c (maybe_hot_edge_p): Update.
3483 (unlikely_executed_edge_p): Update.
3484 (probably_never_executed_edge_p): Update.
3485 (dump_prediction): Update.
3486 (drop_profile): Update.
3487 (propagate_unlikely_bbs_forward): Update.
3488 (determine_unlikely_bbs): Update.
3489 (force_edge_cold): Update.
3490 * profile.c (compute_branch_probabilities): Update.
3491 * reg-stack.c (better_edge): Update.
3492 * shrink-wrap.c (handle_simple_exit): Update.
3493 * tracer.c (better_p): Update.
3494 * trans-mem.c (expand_transaction): Update.
3495 (split_bb_make_tm_edge): Update.
3496 * tree-call-cdce.c: Update.
3497 * tree-cfg.c (gimple_find_sub_bbs): Update.
3498 (gimple_split_edge): Update.
3499 (gimple_duplicate_sese_region): Update.
3500 (gimple_duplicate_sese_tail): Update.
3501 (gimple_flow_call_edges_add): Update.
3502 (insert_cond_bb): Update.
3503 (execute_fixup_cfg): Update.
3504 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3505 * tree-complex.c (expand_complex_div_wide): Update.
3506 * tree-eh.c (lower_resx): Update.
3507 (unsplit_eh): Update.
3508 (cleanup_empty_eh_move_lp): Update.
3509 * tree-inline.c (copy_edges_for_bb): Update.
3510 (freqs_to_counts): Update.
3511 (copy_cfg_body): Update.
3512 * tree-ssa-dce.c (remove_dead_stmt): Update.
3513 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3514 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3515 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3516 (unloop_loops): Update.
3517 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3518 * tree-ssa-loop-split.c (connect_loops): Update.
3519 (split_loop): Update.
3520 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3521 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3522 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3523 * tree-ssa-reassoc.c (branch_fixup): Update.
3524 * tree-ssa-tail-merge.c (replace_block_by): Update.
3525 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3526 (compute_path_counts): Update.
3527 (update_profile): Update.
3528 (recompute_probabilities): Update.
3529 (update_joiner_offpath_counts): Update.
3530 (estimated_freqs_path): Update.
3531 (freqs_to_counts_path): Update.
3532 (clear_counts_path): Update.
3533 (ssa_fix_duplicate_block_edges): Update.
3534 (duplicate_thread_path): Update.
3535 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3536 (case_bit_test_cmp): Update.
3537 (collect_switch_conv_info): Update.
3538 (gen_inbound_check): Update.
3539 (do_jump_if_equal): Update.
3540 (emit_cmp_and_jump_insns): Update.
3541 * tree-tailcall.c (decrease_profile): Update.
3542 (eliminate_tail_call): Update.
3543 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3544 (vect_do_peeling): Update.
3545 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3546 * ubsan.c (ubsan_expand_null_ifn): Update.
3547 (ubsan_expand_ptr_ifn): Update.
3548 * value-prof.c (gimple_divmod_fixed_value): Update.
3549 (gimple_mod_pow2): Update.
3550 (gimple_mod_subtract): Update.
3551 (gimple_ic): Update.
3552 (gimple_stringop_fixed_value): Update.
3554 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3557 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3559 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3561 PR rtl-optimization/82395
3562 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3563 based on non_spilled_static_chain_regno_p.
3565 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3567 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3568 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3569 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3571 2017-10-19 Martin Sebor <msebor@redhat.com>
3573 PR tree-optimization/82596
3574 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3576 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3578 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3579 (get_mem_refs_of_builtin_call): Likewise.
3580 * builtins.c (expand_builtin_apply): Adjust call to
3581 allocate_dynamic_stack_space.
3582 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3583 the third argument to allocate_dynamic_stack_space, otherwise -1.
3584 (expand_builtin): Deal with all alloca variants.
3585 (is_inexpensive_builtin): Likewise.
3586 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3587 * calls.c (special_function_p): Deal with all alloca variants.
3588 (initialize_argument_information): Adjust call to
3589 allocate_dynamic_stack_space.
3590 (expand_call): Likewise.
3591 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3592 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3593 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3594 use it for the stack usage computation.
3595 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3596 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3597 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3598 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3599 (in_loop_p): Remove first argument and useless check.
3600 (pass_walloca::execute): Remove useless test and adjust call to above.
3601 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3602 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3603 (gimplify_call_expr): Deal with all alloca variants.
3604 * hsa-gen.c (gen_hsa_alloca): Likewise.
3605 (gen_hsa_insns_for_call): Likewise.
3606 * ipa-pure-const.c (special_builtin_state): Likewise.
3607 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3608 * tree-object-size.c (alloc_object_size): Likewise.
3609 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3610 (call_may_clobber_ref_p_1): Likewise.
3611 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3612 (ccp_fold_stmt): Likewise.
3613 (optimize_stack_restore): Likewise.
3614 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3615 (mark_all_reaching_defs_necessary_1): Likewise.
3616 (propagate_necessity): Likewise.
3617 (eliminate_unnecessary_stmts): Likewise.
3618 * tree.c (build_common_builtin_nodes): Build
3619 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3620 (build_alloca_call_expr): New function.
3621 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3622 (CASE_BUILT_IN_ALLOCA): Likewise.
3623 (build_alloca_call_expr): Declare.
3624 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3626 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3629 * dwarf2out.c (new_die_raw): New static inline function.
3630 (new_die): Use it to create the DIE.
3631 (add_AT_external_die_ref): Likewise.
3632 (clone_die): Likewise.
3633 (clone_as_declaration): Likewise.
3634 (dwarf2out_vms_debug_main_pointer): Likewise.
3635 (base_type_die): Likewise. Remove early return for corner cases.
3636 Do not call add_pubtype on the DIE here.
3637 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3638 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3639 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3640 native order exists for base types, attach the DIE manually and call
3641 add_pubtype on it. Do not equate a reverse order DIE to the type.
3643 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3645 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3646 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3648 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3650 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3652 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3653 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3654 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3655 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3656 * opts.c (sanitizer_opts): Add builtin.
3657 * ubsan.c (instrument_builtin): New function.
3658 (pass_ubsan::execute): Call it.
3659 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3660 * doc/invoke.texi: Document -fsanitize=builtin.
3662 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3663 builtins, store max (log2 (align), 0) into uchar field instead of
3664 align into uptr field.
3665 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3666 store uchar 0 field instead of uptr 0 field.
3667 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3668 instead of passing one address of struct with 2 locations pass
3669 two addresses of structs with 1 location each.
3670 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3671 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3672 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3673 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3674 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3675 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3676 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3677 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3679 2017-10-19 Martin Liska <mliska@suse.cz>
3682 * file-find.c (remove_prefix): Remove.
3683 * file-find.h (remove_prefix): Likewise.
3684 * gcc-ar.c: Remove smartness of lookup.
3686 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3688 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3689 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3690 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3692 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3695 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3696 (setcc + and to xor + setcc): New peephole2.
3698 2017-10-19 Tom de Vries <tom@codesourcery.com>
3700 * doc/sourcebuild.texi (Test Directives, Variants of
3701 dg-require-support): Add dg-require-stack-size.
3703 2017-10-19 Martin Liska <mliska@suse.cz>
3706 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3707 that have a large alignment.
3708 (gimplify_target_expr): Likewise.
3710 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3712 PR rtl-optimization/82602
3713 * ira.c (rtx_moveable_p): Return false for volatile asm.
3715 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3718 * config/i386/i386-modes.def (CCGZ): New CC mode.
3719 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3720 * config/i386/predicates.md (ix86_comparison_operator):
3722 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3723 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3724 with double-word subtraction.
3725 (put_condition_code): Handle CCGZmode.
3727 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3729 * wide-int.cc (debug (const wide_int &)): New.
3730 (debug (const wide_int *)): New.
3731 (debug (const widest_int &)): New.
3732 (debug (const widest_int *)): New.
3734 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3737 * lra-constraints.c (curr_insn_transform): Use non-input operand
3738 instead of output one for matched reload.
3740 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3742 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3743 (tree-ssa-loop-ivopts.h): New header file.
3744 (struct builtin_info): New fields.
3745 (classify_builtin_1): Compute and record base and offset parts for
3746 memset builtin partition by calling strip_offset.
3747 (offset_cmp, fuse_memset_builtins): New functions.
3748 (finalize_partitions): Fuse adjacent memset partitions by calling
3750 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3751 Expose the interface.
3752 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3754 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3756 PR tree-optimization/82574
3757 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3758 that data reference must be executed exactly once per iteration
3759 against the outermost loop in nest.
3760 (classify_partition): Update call to above function.
3762 2017-10-18 Richard Biener <rguenther@suse.de>
3764 PR tree-optimization/82591
3765 * graphite.c (graphite_transform_loops): Move code gen message
3767 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3768 Here. Handle scop_to_isl_ast failing.
3769 (scop_to_isl_ast): Limit the number of ISL operations.
3771 2017-10-18 Richard Biener <rguenther@suse.de>
3773 * graphite-isl-ast-to-gimple.c
3774 (translate_isl_ast_to_gimple::set_rename): Simplify.
3775 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3776 (graphite_copy_stmts_from_block): ... here.
3777 (copy_bb_and_scalar_dependences): Simplify.
3778 (add_parameters_to_ivs_params): Canonicalize.
3779 (generate_entry_out_of_ssa_copies): Simplify.
3780 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3781 by passing in ISL dimension.
3782 (parameter_index_in_region_1): Rename to ...
3783 (parameter_index_in_region): ... this.
3784 (extract_affine): Adjust assert, pass down parameter index.
3785 (add_param_constraints): Use range-info when available.
3786 (build_scop_context): Adjust.
3787 * sese.c (new_sese_info): Adjust.
3788 (free_sese_info): Likewise.
3789 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3790 Remove unused typedefs.
3791 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3793 2017-10-18 Martin Liska <mliska@suse.cz>
3795 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3797 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3799 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3800 (s390_sched_init): Do not reset s390_sched_state if we entered the
3801 current basic block via a fallthru edge and all others are unlikely.
3803 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3805 * config/s390/s390.c (NUM_SIDES): New variable.
3806 (LONGRUNNING_THRESHOLD): New variable.
3807 (LATENCY_FACTOR): New variable.
3808 (s390_sched_score): Decrease score for long-running instructions on
3810 (s390_sched_variable_issue): Perform bookkeeping for long-running
3813 2017-10-18 Richard Biener <rguenther@suse.de>
3815 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3816 Simplify with removal of the parameter rename map.
3817 (set_rename): Likewise.
3818 (should_copy_to_new_region): Likewise.
3819 (graphite_copy_stmts_from_block): Likewise.
3820 (copy_bb_and_scalar_dependences): Remove initialization of
3821 unused copied_bb_map.
3823 (copy_internal_parameters): Likewise.
3824 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3825 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3826 Use INTEGRAL_TYPE_P.
3827 (parameter_index_in_region_1): Rename to ...
3828 (assign_parameter_index_in_region): ... this. Assert we have
3829 a parameter we handle.
3830 (scan_tree_for_params): Adjust.
3831 * sese.h (parameter_rename_map_t): Remove.
3832 (struct sese_info_t): Remove unused parameter_rename_map and
3833 copied_bb_map members.
3834 * sese.c (new_sese_info): Adjust.
3835 (free_sese_info): Likewise.
3837 2017-10-18 Martin Liska <mliska@suse.cz>
3840 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3841 on an abnormal edge.
3843 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3845 * doc/invoke.texi (ffunction-sections and fdata-sections):
3848 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3850 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3851 the use statement can throw internally.
3853 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3855 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3856 any RTX present on the RHS of a SET.
3857 * compare-elim.c (try_eliminate_compare): Restore comment.
3859 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3861 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3862 may be also called on tcc_type nodes.
3863 * langhooks.c (lhd_tree_size): Likewise.
3865 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3867 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3868 format_warning_at_substring.
3869 (maybe_warn): Convert source_range * param to a location_t. Pass
3870 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3871 (format_directive): Remove code to extract source_ranges and
3872 source_range * in favor of just a location_t.
3873 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3875 * substring-locations.c (format_warning_va): Convert
3876 source_range * param to a location_t.
3877 (format_warning_at_substring): Likewise.
3878 * substring-locations.h (format_warning_va): Likewise.
3879 (format_warning_at_substring): Likewise.
3881 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3883 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3885 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3886 and scatter/gather ops.
3888 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3889 vec_gather_load and vec_scatter_store.
3890 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3891 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3892 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3893 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3894 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3895 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3897 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3899 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3900 Detect FTST instruction and handle its register pops. Only pop
3901 second operand if can_pop_second_op is true.
3902 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3903 set can_pop_second_op to false in the compare_for_stack_reg call.
3905 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3906 output_fp_compare for stack register operands.
3907 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3908 instructions here. Do not emit stack register pops here. Assert
3909 that FCOMPP pops next to top stack register. Rewrite function.
3911 2017-10-17 Nathan Sidwell <nathan@acm.org>
3914 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3915 use DECL_ASSEMBLER_NAME_RAW.
3918 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3921 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3922 Wilco Dijkstra <wilco.dijkstra@arm.com>
3924 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3925 converstion to Pmode from the buf_addr.
3927 2017-10-17 Richard Biener <rguenther@suse.de>
3929 * graphite-dependences.c (scop_get_reads_and_writes): Change
3930 output parameters to references.
3932 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3934 PR 71026/tree-optimization
3935 * fold-const.c (distribute_real_division): Removed.
3936 (fold_binary_loc): Remove calls to distribute_real_divison.
3938 2017-10-17 Richard Biener <rguenther@suse.de>
3940 * graphite-scop-detection.c
3941 (scop_detection::stmt_has_simple_data_refs_p): Always use
3942 the full nest as region.
3943 (try_generate_gimple_bb): Likewise.
3944 * sese.c (scalar_evolution_in_region): Simplify now that
3945 SCEV can handle instantiation in regions.
3946 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3947 in the non-loop part of a function if requested.
3949 2017-10-17 Richard Biener <rguenther@suse.de>
3951 PR tree-optimization/82563
3952 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3954 (graphite_regenerate_ast_isl): Call it.
3955 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3957 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3959 PR tree-optimization/82549
3960 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3961 Formatting fixes. Instead of calling make_bit_field_ref with negative
3964 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3966 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3967 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3968 _mm_maskz_reduce_ss): New.
3969 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3970 __builtin_ia32_reducess_mask): Ditto..
3971 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3972 * config/i386/sse.md (reduces<mode>): Renamed to ...
3973 (reduces<mode><mask_scalar_name>): ... this.
3974 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3976 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3977 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3979 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3981 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3982 * selftest-run-tests.c (selftest::run_tests): Call
3983 selftest::unique_ptr_tests_cc_tests.
3984 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3985 * unique-ptr-tests.cc: New file.
3987 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3990 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3992 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3993 (make_hard_regno_born, make_hard_regno_dead): Update
3994 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3996 2017-10-16 Jeff Law <law@redhat.com>
3998 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4000 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4002 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4004 2017-10-16 Olivier Hainque <hainque@adacore.com>
4006 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4007 with_cpu if we were configured for an e500v2 target cpu name.
4009 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4011 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4012 * doc/invoke.texi: Document +nodsp as a valid extension for
4015 2017-10-16 Martin Liska <mliska@suse.cz>
4017 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4018 (test_set_range): Likewise.
4019 (test_range_functions): Rename to ...
4020 (test_bit_in_range): ... this.
4021 (sbitmap_c_tests): Add new test.
4023 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4025 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4027 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4028 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4030 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4032 * config/aarch64/aarch64-builtins.c
4033 (aarch64_types_quadopu_lane_qualifiers): New.
4034 (TYPES_QUADOPU_LANE): New.
4035 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4036 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4037 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4038 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4039 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4040 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4041 (Vdottype, DOTPROD): New.
4042 (sur): Add SDOT and UDOT.
4044 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4046 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4047 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4048 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4050 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4051 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4052 Enable TARGET_DOTPROD.
4053 (cortex-a75.cortex-a55): Likewise.
4054 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4056 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4058 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4059 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4061 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4063 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4064 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4065 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4066 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4067 * config/arm/types.md (neon_dot, neon_dot_q): New.
4068 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4070 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4072 * config/arm/arm.h (TARGET_DOTPROD): New.
4073 * config/arm/arm.c (arm_arch_dotprod): New.
4074 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4075 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4076 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4077 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4078 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4079 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4080 * doc/invoke.texi (armv8.2-a): Document dotprod
4082 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4084 * i386.c (ix86_vec_cost): New function.
4085 (ix86_rtx_costs): Handle vector operations better.
4086 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4087 * x86-tune-costs.h: Add new costs to all tables.
4089 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4091 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4093 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4094 divsd, sqrtss and sqrtsd
4095 * x86-tune-costs.h: Add new entries to all costs.
4096 (znver1_cost): Fix to match real instruction latencies.
4098 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4099 Michael Collison <michael.collison@arm.com>
4101 * compare-elim.c: Include emit-rtl.h.
4102 (can_merge_compare_into_arith): New function.
4103 (try_validate_parallel): Likewise.
4104 (try_merge_compare): Likewise.
4105 (try_eliminate_compare): Call the above when no previous clobber
4107 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4110 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4114 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4115 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4119 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4120 to be any operand_equal_p operands. For & (B - 1) require
4121 B to be power of 2. Recognize
4122 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4124 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4127 * optabs.c (expand_memory_blockage): Fix call of
4128 targetm.have_memory_blockage.
4130 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4133 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4134 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4135 objects to extra_objs instead of overwriting it.
4137 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4139 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4140 Use any_fp_register_operand as operand[3] predicate. Simplify
4141 equality test for operands[2] and operands[4] memory location.
4142 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4143 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4144 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4145 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4146 any_fp_register_operand as operand[1] predicate. Simplify
4147 equality test for operands[0] and operands[3] memory location.
4148 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4149 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4150 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4152 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4154 * target-insns.def: Add memory_blockage.
4155 * optabs.c (expand_memory_blockage): New function.
4156 (expand_asm_memory_barrier): Rename ...
4157 (expand_asm_memory_blockage): ... to this.
4158 (expand_mem_thread_fence): Call expand_memory_blockage
4159 instead of expand_asm_memory_barrier.
4160 (expand_mem_singnal_fence): Ditto.
4161 (expand_atomic_load): Ditto.
4162 (expand_atomic_store): Ditto.
4163 * doc/md.texi (Standard Pattern Names For Generation):
4164 Document memory_blockage instruction pattern.
4166 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4168 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4169 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4170 * config/rl78/rl78.md: New define_expand "adddi3".
4172 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4174 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4177 2017-10-13 Jeff Law <law@redhat.com>
4179 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4181 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4184 * internal-fn.c (expand_mul_overflow): If both operands have
4185 the same highpart of -1 or 0 and the topmost bit of lowpart
4186 is different, overflow is if res <= 0 rather than res < 0.
4188 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4190 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4191 TARGET_P9_VECTOR code for unaligned_load case.
4193 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4195 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4197 2017-10-13 Nathan Sidwell <nathan@acm.org>
4199 * tree-core.h (tree_contains_struct): Make bool.
4200 * tree.c (tree_contains_struct): Likewise.
4201 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4202 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4203 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4204 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4205 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4207 2017-10-13 Richard Biener <rguenther@suse.de>
4209 * graphite-isl-ast-to-gimple.c
4210 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4211 parameters and dominance check.
4212 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4213 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4214 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4215 Do not update SSA form here or do intermediate IL verification.
4216 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4217 (graphite_initialize): Remove check on the number of loops in
4218 the function and inline into graphite_transform_loops.
4219 (graphite_finalize): Inline into graphite_transform_loops.
4220 (graphite_transform_loops): Perform SSA update and IL verification
4222 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4224 2017-10-13 Richard Biener <rguenther@suse.de>
4226 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4227 graphite_expression_type_precision): Avoid global constructor
4229 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4230 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4231 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4232 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4233 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4234 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4236 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4239 * config/i386/i386.h (ix86_red_zone_size): New.
4240 * config/i386/i386.md (push peephole2s): Replace
4241 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4243 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4244 Alan Hayward <alan.hayward@arm.com>
4245 David Sherwood <david.sherwood@arm.com>
4247 * combine.c (can_change_dest_mode): Reject changes in
4248 REGMODE_NATURAL_SIZE.
4250 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4251 Alan Hayward <alan.hayward@arm.com>
4252 David Sherwood <david.sherwood@arm.com>
4254 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4255 (expand_debug_source_expr): Likewise.
4256 * combine.c (combine_simplify_rtx): Likewise.
4257 * cse.c (fold_rtx): Likewise.
4258 * fwprop.c (canonicalize_address): Likewise.
4259 * targhooks.c (default_shift_truncation_mask): Likewise.
4261 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4262 Alan Hayward <alan.hayward@arm.com>
4263 David Sherwood <david.sherwood@arm.com>
4265 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4266 (widened_mode): Likewise.
4267 (expand_unop): Likewise.
4268 * ree.c (transform_ifelse): Likewise.
4269 (merge_def_and_ext): Likewise.
4270 (combine_reaching_defs): Likewise.
4271 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4273 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4274 Alan Hayward <alan.hayward@arm.com>
4275 David Sherwood <david.sherwood@arm.com>
4277 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4278 * combine.c (gen_lowpart_for_combine): Likewise.
4279 * dwarf2out.c (rtl_for_decl_location): Likewise.
4280 * final.c (alter_subreg): Likewise.
4281 * rtlhooks.c (gen_lowpart_general): Likewise.
4282 (gen_lowpart_if_possible): Likewise.
4284 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4285 Alan Hayward <alan.hayward@arm.com>
4286 David Sherwood <david.sherwood@arm.com>
4288 * calls.c (expand_call): Use subreg_lowpart_offset.
4289 * cse.c (cse_insn): Likewise.
4290 * regcprop.c (copy_value): Likewise.
4291 (copyprop_hardreg_forward_1): Likewise.
4293 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4296 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4297 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4298 =Q constraints to +Q and into insn condition add check
4299 that operands[0] and operands[1] are equal.
4300 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4301 =Q constraints to +Q and into insn condition add check
4302 that operands[0] is equal to either operands[1] or operands[2].
4305 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4306 instead of handling MINUS_EXPR twice (once for each argument),
4307 canonicalize operand order and handle just once, use rtype where
4308 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4311 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4312 any values of __C while still being pattern recognizable as a simple
4315 2017-10-13 Richard Biener <rguenther@suse.de>
4317 PR tree-optimization/82451
4319 2017-10-02 Richard Biener <rguenther@suse.de>
4321 PR tree-optimization/82355
4322 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4323 a mapping for the enclosing loop but avoid generating one for
4325 (copy_bb_and_scalar_dependences): Remove premature codegen
4326 error on PHIs in blocks duplicated into multiple places.
4327 * graphite-scop-detection.c
4328 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4329 in the region use it as loop and nest to analyze the DR in.
4330 (try_generate_gimple_bb): Likewise.
4331 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4332 (add_loop_constraints): For blocks in a loop not in the region
4333 create a dimension with a single iteration.
4334 * sese.h (gbb_loop_at_index): Remove assert.
4336 * cfgloop.c (loop_preheader_edge): For the loop tree root
4337 return the single successor of the entry block.
4338 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4339 Reset the SCEV hashtable and niters.
4340 * graphite-scop-detection.c
4341 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4342 assert that we only have POLYNOMIAL_CHREC that vary in loops
4343 contained in the region.
4344 (scop_detection::graphite_can_represent_expr): Adjust.
4345 (scop_detection::stmt_has_simple_data_refs_p): For loops
4346 not in the region set loop to NULL. The nest is now the
4347 entry edge to the region.
4348 (try_generate_gimple_bb): Likewise.
4349 * sese.c (scalar_evolution_in_region): Adjust for
4350 instantiate_scev change.
4351 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4352 Make nest parameter the edge into the region.
4353 (create_data_ref): Likewise.
4354 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4355 entry edge into a region and adjust instantiate_scev calls.
4356 (create_data_ref): Likewise.
4357 (graphite_find_data_references_in_stmt): Likewise.
4358 (find_data_references_in_stmt): Pass the loop preheader edge
4359 from the nest argument.
4360 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4361 parameter the edge into the region.
4362 (instantiate_parameters): Use the loop preheader edge as entry.
4363 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4365 (get_instantiated_value_entry): Make instantiate_below parameter
4366 the edge into the region.
4367 (instantiate_scev_name): Likewise. Adjust dominance checks,
4368 when we cannot use loop-based instantiation instantiate by
4369 walking use-def chains.
4370 (instantiate_scev_poly): Adjust.
4371 (instantiate_scev_binary): Likewise.
4372 (instantiate_scev_convert): Likewise.
4373 (instantiate_scev_not): Likewise.
4374 (instantiate_array_ref): Remove.
4375 (instantiate_scev_3): Likewise.
4376 (instantiate_scev_2): Likewise.
4377 (instantiate_scev_1): Likewise.
4378 (instantiate_scev_r): Do not blindly handle N-operand trees.
4379 Do not instantiate array-refs. Handle all constants and invariants.
4380 (instantiate_scev): Make instantiate_below parameter
4381 the edge into the region.
4382 (resolve_mixers): Use the loop preheader edge for the region
4383 parameter to instantiate_scev_r.
4384 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4386 2017-10-13 Richard Biener <rguenther@suse.de>
4388 PR tree-optimization/82525
4389 * graphite-isl-ast-to-gimple.c
4390 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4392 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4393 Fail code generation when we cannot represent the isl integer.
4394 (binary_op_to_tree): Elide modulo operations that are no-ops
4395 in the type we code generate. Remove now superfluous code
4398 2017-10-13 Richard Biener <rguenther@suse.de>
4400 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4401 (scop_detection::harmful_loop_in_region): Remove premature
4402 IV type restriction.
4403 (scop_detection::graphite_can_represent_scev): We can handle
4404 pointer IVs just fine.
4406 2017-10-13 Alan Modra <amodra@gmail.com>
4408 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4409 "Clobbers and Scratch Registers". Add paragraph on
4410 alternative to clobbers for scratch registers and OpenBLAS
4413 2017-10-13 Alan Modra <amodra@gmail.com>
4415 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4416 example of a memory input for a string of known length. Move
4417 commentary out of table. Add a number of new examples
4418 covering array memory inputs.
4420 2017-10-12 Martin Liska <mliska@suse.cz>
4422 PR tree-optimization/82493
4423 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4424 (test_range_functions): New function.
4425 (sbitmap_c_tests): Likewise.
4426 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4427 * selftest.h (sbitmap_c_tests): New function.
4429 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4431 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4433 * config/rs6000/amo.h: Fix spacing issue.
4435 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4438 * config/i386/i386.md (*ashl<mode>3_mask_1,
4439 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4440 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4443 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4445 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4446 (profile_probability): Set max_probability
4447 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4449 * profile-count.c (profile_probability::differs_from_p): Do not
4450 rely on max_probaiblity == 10000
4452 2017-10-12 Jeff Law <law@redhat.com>
4454 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4457 2017-10-12 Martin Sebor <msebor@redhat.com>
4461 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4462 (handle_alias_pairs): Call it.
4463 * common.opt (-Wattribute-alias): New option.
4464 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4465 * doc/invoke.texi (-Wattribute-alias): Document.
4467 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4470 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4472 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4474 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4475 (make_hard_regno_born, make_hard_regno_dead): Update
4476 bb_killed_pseudos and bb_gen_pseudos.
4478 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4480 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4482 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4484 * config/alpha/alpha.c (alpha_split_conditional_move):
4485 Use std::swap instead of manually swapping.
4486 (alpha_stdarg_optimize_hook): Ditto.
4487 (alpha_canonicalize_comparison): Ditto.
4489 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4491 * tree-loop-distribution.c (struct builtin_info): New struct.
4492 (struct partition): Refactor fields into struct builtin_info.
4493 (partition_free): Free struct builtin_info.
4494 (build_size_arg_loc, build_addr_arg_loc): Delete.
4495 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4496 information from struct builtin_info.
4497 (find_single_drs): New function refactored from classify_partition.
4498 Also moved builtin validity checks to this function.
4499 (compute_access_range, alloc_builtin): New functions.
4500 (classify_builtin_st, classify_builtin_ldst): New functions.
4501 (classify_partition): Refactor code into functions find_single_drs,
4502 classify_builtin_st and classify_builtin_ldst.
4503 (distribute_loop): Don't do runtime alias check when distributing
4505 (find_seed_stmts_for_distribution): New function.
4506 (pass_loop_distribution::execute): Refactor code finding seed
4507 stmts into above function. Support distribution for the innermost
4508 two-level loop nest. Adjust dump information.
4510 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4512 * tree-loop-distribution.c: Adjust the general comment.
4513 (NUM_PARTITION_THRESHOLD): New macro.
4514 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4515 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4516 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4517 in call to build_partition_graph.
4518 (finalize_partitions): New parameter. Make loop distribution more
4519 conservative by fusing more partitions.
4520 (distribute_loop): Don't do runtime alias check in case of loop nest
4522 (find_seed_stmts_for_distribution): New function.
4523 (prepare_perfect_loop_nest): New function.
4524 (pass_loop_distribution::execute): Refactor code finding seed stmts
4525 and loop nest into above functions. Support loop nest distribution.
4526 Adjust dump information accordingly.
4528 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4530 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4531 and set PTYPE_SEQUENTIAL for merged partition.
4533 2017-10-12 Richard Biener <rguenther@suse.de>
4535 PR tree-optimization/69728
4537 2017-09-19 Richard Biener <rguenther@suse.de>
4539 PR tree-optimization/69728
4540 * graphite-sese-to-poly.c (schedule_error): New global.
4541 (add_loop_schedule): Handle empty domain by failing the
4543 (build_original_schedule): Handle schedule_error.
4545 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4546 domain by returning an unchanged schedule.
4548 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4550 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4551 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4553 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4555 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4558 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4561 * expr.c (store_field): Don't optimize away bitsize == 0 store
4562 from CALL_EXPR with addressable return type.
4564 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4566 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4567 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4568 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4569 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4570 TARGET_ISEL instead of TARGET_ISEL<sel>.
4572 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4574 * config/rs6000/rs6000.c
4575 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4577 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4579 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4581 (reg_or_cint_operand): Fix comment.
4582 (reg_or_zero_operand): New predicate.
4583 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4584 * config/rs6000/rs6000.c (output_isel): Delete.
4585 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4586 instead of reg_or_cint_operand. Output instruction directly (not via
4588 (isel_unsigned_<mode>): Ditto.
4589 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4590 gpc_reg_operand. Add an instruction alternative for this. Output
4591 instruction directly.
4592 (*isel_reversed_unsigned_<mode>): Ditto.
4594 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4596 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4597 (TARGET_CANONICALIZE_COMPARISON): Define.
4599 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4602 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4603 Check whether the dest is REG before adding REG_EQUIV note.
4605 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4608 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4610 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4611 (make_hard_regno_born, make_hard_regno_dead): Update
4612 bb_killed_pseudos and bb_gen_pseudos.
4614 2017-10-11 Nathan Sidwell <nathan@acm.org>
4616 * incpath.h (enum incpath_kind): Name enum, prefix values.
4617 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4618 * incpath.c (heads, tails): Use INC_MAX.
4619 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4620 (merge_include_chains, split_quote_chain,
4621 register_include_chains): Update incpath_kind names.
4622 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4623 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4625 (add_framework_path, darwin_register_objc_includes): Likewise.
4626 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4628 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4630 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4631 Do not use float_operator operator predicate.
4632 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4633 * config/i386/predicates.md (float_operator): Remove predicate.
4635 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4637 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4638 (*jccxf_i387): Ditto.
4639 (*jcc<mode>_i387): Ditto.
4640 (*jccu<mode>_i387): Ditto.
4641 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4642 (*jcc_*_i387 splitters): Remove.
4643 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4644 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4645 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4648 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4650 * profile-count.h (slow_safe_scale_64bit): New function.
4651 (safe_scale_64bit): New inline.
4652 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4653 * profile-count.c: Include wide-int.h
4654 (slow_safe_scale_64bit): New.
4656 2017-10-11 Nathan Sidwell <nathan@acm.org>
4658 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4659 HAS_DECL_ASSEMBLER_NAME_P.
4660 * gimple-expr.c (gimple_decl_printable_name: Check
4661 HAS_DECL_ASSEMBLER_NAME_P too.
4662 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4663 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4664 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4665 * passes.c (rest_of_decl_compilation): Check
4666 HAS_DECL_ASSEMBLER_NAME_P too.
4667 * recog.c (verify_changes): Likewise.
4668 * tree-pretty-print.c (dump_decl_name): Likewise.
4669 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4671 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4672 (SET_DECL_ASSEMBLER_NAME): Use it.
4673 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4674 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4675 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4677 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4679 * config.gcc (i386, x86_64): Add extra objects.
4680 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4681 (ix86_min_insn_size): Declare.
4682 (ix86_issue_rate): Declare.
4683 (ix86_adjust_cost): Declare.
4684 (ia32_multipass_dfa_lookahead): Declare.
4685 (ix86_macro_fusion_p): Declare.
4686 (ix86_macro_fusion_pair_p): Declare.
4687 (ix86_bd_has_dispatch): Declare.
4688 (ix86_bd_do_dispatch): Declare.
4689 (ix86_core2i7_init_hooks): Declare.
4690 (ix86_atom_sched_reorder): Declare.
4691 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4692 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4693 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4694 (rip_relative_addr_p): Rename to ...
4695 (ix86_rip_relative_addr_p): ... this one; export.
4696 (memory_address_length): Update.
4697 (ix86_issue_rate): Move to x86-tune-sched.c.
4698 (ix86_flags_dependent): Move to x86-tune-sched.c.
4699 (ix86_agi_dependent): Move to x86-tune-sched.c.
4700 (exact_dependency_1): Move to x86-tune-sched.c.
4701 (exact_store_load_dependency): Move to x86-tune-sched.c.
4702 (ix86_adjust_cost): Move to x86-tune-sched.c.
4703 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4704 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4705 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4706 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4707 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4708 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4709 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4710 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4711 (min_insn_size): Rename to ...
4712 (ix86_min_insn_size): ... this one; export.
4713 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4714 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4715 (core2i7_first_cycle_multipass_backtrack): Move to
4716 x86-tune-sched-core.c.
4717 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4718 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4719 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4720 (ix86_avoid_jump_mispredicts): Update.
4721 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4722 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4723 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4724 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4725 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4726 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4727 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4728 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4729 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4730 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4731 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4732 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4733 (BIG): Move to ix86-tune-sched-bd.c.
4734 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4735 (enum insn_path): Move to ix86-tune-sched-bd.c.
4736 (get_mem_group): Move to ix86-tune-sched-bd.c.
4737 (is_cmp): Move to ix86-tune-sched-bd.c.
4738 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4739 (is_branch): Move to ix86-tune-sched-bd.c.
4740 (is_prefetch): Move to ix86-tune-sched-bd.c.
4741 (init_window): Move to ix86-tune-sched-bd.c.
4742 (allocate_window): Move to ix86-tune-sched-bd.c.
4743 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4744 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4745 (process_end_window): Move to ix86-tune-sched-bd.c.
4746 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4747 (find_constant): Move to ix86-tune-sched-bd.c.
4748 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4749 (has_immediate): Move to ix86-tune-sched-bd.c.
4750 (get_insn_path): Move to ix86-tune-sched-bd.c.
4751 (get_insn_group): Move to ix86-tune-sched-bd.c.
4752 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4753 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4754 (add_insn_window): Move to ix86-tune-sched-bd.c.
4755 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4756 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4757 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4758 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4759 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4760 (do_dispatch): Move to ix86-tune-sched-bd.c.
4761 (has_dispatch): Move to ix86-tune-sched-bd.c.
4762 * config/i386/t-i386: Add new object files.
4763 * config/i386/x86-tune-costs.h: New file.
4764 * config/i386/x86-tune-sched-atom.c: New file.
4765 * config/i386/x86-tune-sched-bd.c: New file.
4766 * config/i386/x86-tune-sched-core.c: New file.
4767 * config/i386/x86-tune-sched.c: New file.
4769 2017-10-11 Liu Hao <lh_mouse@126.com>
4771 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4772 the generic version below instead.
4773 (should_colorize): Recognize Windows consoles as terminals
4775 * pretty-print.c [__MINGW32__] (write_all): New function.
4776 [__MINGW32__] (find_esc_head): Likewise.
4777 [__MINGW32__] (find_esc_terminator): Likewise.
4778 [__MINGW32__] (eat_esc_sequence): Likewise.
4779 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4781 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4784 2017-10-11 Richard Biener <rguenther@suse.de>
4786 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4787 Properly call analyze_scalar_evolution with the loop of the stmt.
4789 2017-10-11 Richard Biener <rguenther@suse.de>
4791 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4792 * tree-core.h (tree_base): Add chrec_var union member.
4793 * tree.h (CHREC_VAR): Remove.
4794 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4795 * tree-chrec.h (build_polynomial_chrec): Adjust.
4796 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4797 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4799 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4801 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4802 * match.pd: ... here.
4803 ((T) X == (T) Y): Relax condition.
4805 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4807 PR tree-optimization/82472
4808 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4810 (break_alias_scc_partitions): Update postorder number.
4812 2017-10-11 Martin Liska <mliska@suse.cz>
4815 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4817 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4819 2017-10-11 Martin Liska <mliska@suse.cz>
4821 * print-rtl.c (print_insn): Move declaration of idbuf
4822 to same scope as name.
4824 2017-10-11 Martin Liska <mliska@suse.cz>
4829 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4832 2017-10-11 Martin Liska <mliska@suse.cz>
4835 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4838 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4840 * config.gcc (default_gnu_indirect_function): Default to yes for
4841 arm*-*-linux* with glibc.
4843 2017-10-11 Richard Biener <rguenther@suse.de>
4845 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4846 default-defs and types we do not want to analyze.
4847 (interpret_loop_phi): Replace unreachable code with an assert.
4848 (compute_scalar_evolution_in_loop): Remove and inline ...
4849 (analyze_scalar_evolution_1): ... here, replacing condition with
4850 what makes the intent clearer. Remove handling of cases
4851 get_scalar_evolution now handles.
4853 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4855 PR rtl-optimization/81434
4856 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4857 comment for main loop. In sched_group_found if, also add checks for
4858 pass and min_cost_group.
4860 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4862 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4863 (rs6000_insn_cost): New function.
4864 * config/rs6000/rs6000.md (cost): New attribute.
4866 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4867 H.J. Lu <hongjiu.lu@intel.com>
4871 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4872 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4873 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4874 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4875 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4876 OPTION_MASK_ISA_MMX and some other bit set require both
4877 mmx and the ISAs without the mmx bit.
4878 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4879 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4880 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4881 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4882 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4883 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4884 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4885 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4886 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4887 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4888 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4889 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4890 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4891 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4892 Add OPTION_MASK_ISA_MMX.
4894 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4896 * config.gcc (armv7*-*-freebsd*): New target.
4897 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4899 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4901 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4902 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4905 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4907 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4908 a template parameter.
4909 (WIDE_INT_REF_FOR): Update accordingly.
4910 * tree.h (wi::int_traits <const_tree>): Delete.
4911 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4912 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4913 (wi::tree_to_wide_ref): New typedef.
4914 (wi::to_wide): New function.
4915 * calls.c (get_size_range): Use wi::to_wide when operating on
4917 * cgraph.c (cgraph_node::create_thunk): Likewise.
4918 * config/i386/i386.c (ix86_data_alignment): Likewise.
4919 (ix86_local_alignment): Likewise.
4920 * dbxout.c (stabstr_O): Likewise.
4921 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4922 * expr.c (const_vector_from_tree): Likewise.
4923 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4924 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4925 (fold_negate_expr_1, int_const_binop_1, const_binop)
4926 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4927 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4928 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4929 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4930 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4931 (fold_not_const, round_up_loc): Likewise.
4932 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4933 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4934 (alloca_call_type): Likewise.
4935 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4936 * godump.c (go_output_typedef): Likewise.
4937 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4938 * internal-fn.c (get_min_precision): Likewise.
4939 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4940 * ipa-polymorphic-call.c
4941 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4942 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4943 (ipa_modify_call_arguments): Likewise.
4944 * match.pd: Likewise.
4945 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4946 * print-tree.c (print_node_brief, print_node): Likewise.
4947 * stmt.c (expand_case): Likewise.
4948 * stor-layout.c (layout_type): Likewise.
4949 * tree-affine.c (tree_to_aff_combination): Likewise.
4950 * tree-cfg.c (group_case_labels_stmt): Likewise.
4951 * tree-data-ref.c (dr_analyze_indices): Likewise.
4952 (prune_runtime_alias_test_list): Likewise.
4953 * tree-dump.c (dequeue_and_dump): Likewise.
4954 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4955 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4956 * tree-pretty-print.c (dump_generic_node): Likewise.
4957 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4958 (simple_iv_with_niters): Likewise.
4959 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4960 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4961 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4962 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4963 (refine_value_range_using_guard, number_of_iterations_ne_max)
4964 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4965 (get_cst_init_from_scev, record_nonwrapping_iv)
4966 (scev_var_range_cant_overflow): Likewise.
4967 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4968 * tree-ssa-pre.c (compute_avail): Likewise.
4969 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4970 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4971 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4972 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4973 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4974 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4975 (dump_case_nodes, try_switch_expansion): Likewise.
4976 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4977 (vect_do_peeling): Likewise.
4978 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4979 * tree-vect-stmts.c (vectorizable_load): Likewise.
4980 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4981 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4982 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4983 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4984 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4985 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4986 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4987 (evrp_dom_walker::before_dom_children): Likewise.
4988 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4989 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4990 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4991 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4992 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4993 (get_range_pos_neg): Likewise.
4994 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4995 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4996 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4997 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4998 * config/avr/avr.c (avr_fold_builtin): Likewise.
4999 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5000 * config/msp430/msp430.c (msp430_attr): Likewise.
5001 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5002 * config/powerpcspe/powerpcspe-c.c
5003 (altivec_resolve_overloaded_builtin): Likewise.
5004 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5005 (rs6000_expand_ternop_builtin): Likewise.
5006 * config/rs6000/rs6000-c.c
5007 (altivec_resolve_overloaded_builtin): Likewise.
5008 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5009 (rs6000_expand_ternop_builtin): Likewise.
5010 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5012 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5014 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5015 when copying loop nest with only one inner loop.
5017 2017-10-10 Richard Biener <rguenther@suse.de>
5019 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5020 blocks if SCEV is active.
5021 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5023 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5024 (scev_initialize): Assert we are not yet initialized.
5026 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5028 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5029 inner loop's exit stmt by making it always exit the loop, otherwise
5030 we would generate an infinite empty loop.
5032 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5034 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5035 renaming variables in new preheader if it's deleted.
5037 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5039 * tree-loop-distribution.c (struct partition): Remove unused field
5040 loops of the structure.
5041 (partition_alloc, partition_free): Ditto.
5042 (build_rdg_partition_for_vertex): Ditto.
5044 2017-10-09 Jeff Law <law@redhat.com>
5046 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5047 return type to match prototype and documentation.
5049 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5051 * config/rs6000/rs6000.c (processor_costs): Move to ...
5052 * config/rs6000/rs6000.h: ... here.
5053 (rs6000_cost): Declare.
5055 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5057 * except.c (setjmp_fn): New global variable.
5058 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5059 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5060 if DONT_USE_BUILTIN_SETJMP is defined.
5062 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5064 * target.def (insn_cost): New hook.
5065 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5066 * doc/tm.texi: Regenerate.
5067 * rtlanal.c (insn_cost): Use the new hook.
5069 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5071 * combine.c (combine_validate_cost): Compute the new insn_cost,
5072 not just pattern_cost.
5073 (try_combine): Adjust comment.
5075 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5077 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5079 * combine.c (uid_insn_cost): Adjust comment.
5080 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5082 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5083 * dse.c (find_shift_sequence): Ditto.
5084 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5085 (bb_valid_for_noce_process_p): Use pattern_cost.
5086 * rtl.h (insn_rtx_cost): Delete.
5087 (pattern_cost): New prototype.
5088 (insn_cost): New prototype.
5089 * rtlanal.c (insn_rtx_cost): Rename to...
5090 (pattern_cost): ... this.
5093 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5095 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5096 (*jcc<mode>_0_r_i387): Ditto.
5097 (*jccxf_r_i387): Ditto.
5098 (*jcc<mode>_r_i387): Ditto.
5099 (*jccu<mode>_r_i387): Ditto.
5100 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5101 (*jcc): Rename from *jcc_1.
5103 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5105 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5106 deferred rescans after the lvx/stvx recombination pre-pass.
5108 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5110 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5111 memory operation instruction support.
5112 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5113 (rs6000-ibm-aix[789]*): Likewise.
5114 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5115 Document new functions.
5117 2017-10-09 Richard Biener <rguenther@suse.de>
5119 PR tree-optimization/82397
5120 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5121 equality only for semantically equal trees.
5123 2017-10-09 Richard Biener <rguenther@suse.de>
5125 PR tree-optimization/82449
5126 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5127 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5128 allow constant addresses.
5129 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5132 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5134 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5137 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5140 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5143 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5146 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5148 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5151 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5152 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5154 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5156 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5157 (WI_BINARY_PREDICATE_RESULT): Likewise.
5158 (wi::binary_traits::operator_result): New type.
5159 (wi::binary_traits::predicate_result): Likewise.
5160 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5161 (generic_wide_int::operator==, generic_wide_int::operator!=)
5162 (generic_wide_int::operator&, generic_wide_int::and_not)
5163 (generic_wide_int::operator|, generic_wide_int::or_not)
5164 (generic_wide_int::operator^, generic_wide_int::operator+
5165 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5167 (operator~, unary operator-, operator==, operator!=, operator&)
5168 (operator|, operator^, operator+, binary operator-, operator*): New
5170 * expr.c (get_inner_reference): Use wi::bit_and_not.
5171 * fold-const.c (fold_binary_loc): Likewise.
5172 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5173 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5174 (bit_value_binop): Likewise.
5175 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5176 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5177 (extract_range_from_binary_expr_1): Likewise.
5178 (masked_increment): Likewise.
5179 (simplify_bit_ops_using_ranges): Likewise.
5181 2017-10-09 Martin Jambor <mjambor@suse.cz>
5184 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5185 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5186 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5188 (hsa_fixup_mov_insn_type): New function.
5189 (hsa_op_with_type::get_in_type): Use it.
5190 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5191 immediates in an assert.
5192 (hsa_op_with_type::extend_int_to_32bit): New method.
5193 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5194 types. Convert to dest type if necessary.
5195 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5196 (reg_for_gimple_ssa): Pass false as min32int to
5197 hsa_type_for_scalar_tree_type.
5198 (gen_hsa_addr): Fixup type when creating addresable temporary.
5199 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5200 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5201 necessary. Call hsa_fixup_mov_insn_type.
5202 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5203 extend operands and convert to dest type if necessary.
5204 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5205 to dest type if necessary.
5206 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5207 if conversion nt necessary and size matches.
5208 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5209 to dest type if necessary.
5210 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5211 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5213 (gen_hsa_clrsb): Likewise.
5214 (gen_hsa_ffs): Likewise.
5215 (gen_hsa_divmod): Extend operands and convert to dest type if
5217 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5219 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5221 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5222 Remove empty default arguments. Use a brace block as output
5224 (conditional return): Ditto.
5226 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5227 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5229 (group_ending_nop): Ditto.
5230 (doloop_end): Ditto.
5231 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5232 (splitters for those): Ditto.
5234 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5236 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5237 a conditional jump (and the compare for it) so that pc_rtx is the
5239 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5240 for the deleted and renamed ctr<mode>_internal[234] patterns.
5241 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5242 Delete second conditional return pattern.
5243 (ctr<mode>_internal2): Delete this second bdnz pattern.
5244 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5245 (ctr<mode>_internal4): Delete this second bdz pattern.
5247 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5249 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5250 (always_initialized_rtx_for_ssa_name_p): New predicate.
5251 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5252 (finish_out_of_ssa): Free new field of SA.
5253 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5254 * tree-ssa-coalesce.c: Include tree-ssa.h.
5255 (get_parm_default_def_partitions): Remove extern keyword.
5256 (get_undefined_value_partitions): New function.
5257 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5258 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5260 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5262 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5264 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5266 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5268 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5269 for targets that preffer 128bit.
5271 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5273 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5275 2017-10-08 Olivier Hainque <hainque@adacore.com>
5277 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5278 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5280 (thumb_set_return_address): Likewise.
5282 2017-10-08 Olivier Hainque <hainque@adacore.com>
5284 * common/config/arm/arm-common.c (arm_except_unwind_info):
5285 Handle DWARF2_UNWIND_INFO.
5287 2017-10-07 Michael Collison <michael.collison@arm.com>
5289 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5292 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5294 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5295 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5296 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5297 defined, force the creation of a new block for a dispatch label.
5299 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5301 * invoke.texi (Wsuggest-attribute=cold): Document.
5302 * common.opt (Wsuggest-attribute=cold): New
5303 * ipa-pure-const.c (warn_function_cold): New function.
5304 * predict.c (compute_function_frequency): Use it.
5305 * predict.h (warn_function_cold): Declare.
5307 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5309 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5312 2017-10-06 Martin Liska <mliska@suse.cz>
5314 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5315 keyword for member functions.
5316 (struct sanopt_tree_couple): New struct.
5317 (struct sanopt_tree_couple_hash): New function.
5318 (struct sanopt_ctx): Add new hash_map.
5319 (has_dominating_ubsan_ptr_check): New function.
5320 (record_ubsan_ptr_check_stmt): Likewise.
5321 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5322 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5323 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5325 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5328 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5329 aarch64_simd_valid_immediate on CONST_VECTORs.
5330 (aarch64_reg_or_bic_imm): Likewise.
5332 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5334 PR rtl-optimization/82396
5335 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5337 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5339 * graphite-dependences.c (scop_get_reads): Move code to...
5340 (scop_get_must_writes): Move code to...
5341 (scop_get_may_writes): Move code to...
5342 (scop_get_reads_and_writes): ... here.
5343 (scop_get_dependences): Call scop_get_reads_and_writes.
5345 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5347 PR tree-optimization/82434
5348 * fold-const.h (can_native_encode_type_p,
5349 can_native_encode_string_p): Remove.
5350 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5351 don't encode anything, just return what would be otherwise returned.
5352 (native_encode_fixed, native_encode_complex, native_encode_vector):
5354 (native_encode_string): Likewise. Inline by hand
5355 can_native_encode_string_p.
5356 (can_native_encode_type_p): Remove.
5357 (can_native_encode_string_p): Remove.
5358 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5359 STRING_CSTs using can_native_encode_string_p, test all
5360 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5361 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5362 argument from native_encode_expr.
5363 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5364 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5367 2017-10-06 Richard Biener <rguenther@suse.de>
5369 PR tree-optimization/82397
5370 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5371 operand_equal_p but rely on data_ref_compare_tree for detecting
5373 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5374 to match up with dr_group_sort_cmp.
5376 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5379 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5381 * config/s390/s390-builtin-types.def: Regenerate.
5383 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5386 * config/s390/s390-builtin-types.def: Regenerate.
5387 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5388 Change flag from B_VXE to B_VX.
5389 (s390_vec_min_dbl): Remove B_VXE flag.
5391 2017-10-06 Richard Biener <rguenther@suse.de>
5393 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5394 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5395 translate_isl_ast_to_gimple::is_valid_rename,
5396 translate_isl_ast_to_gimple::get_rename,
5397 translate_isl_ast_to_gimple::get_def_bb_for_const,
5398 translate_isl_ast_to_gimple::get_new_name,
5399 translate_isl_ast_to_gimple::collect_all_ssa_names,
5400 translate_isl_ast_to_gimple::copy_loop_phi_args,
5401 translate_isl_ast_to_gimple::collect_all_ssa_names,
5402 translate_isl_ast_to_gimple::copy_loop_phi_args,
5403 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5404 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5405 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5406 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5407 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5408 translate_isl_ast_to_gimple::copy_cond_phi_args,
5409 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5410 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5411 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5412 translate_isl_ast_to_gimple::rename_uses,
5413 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5414 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5415 (set_rename_for_each_def): Likewise.
5416 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5417 here. Handle rewriting SCEV analyzable uses here.
5418 (copy_bb_and_scalar_dependences): Generate code for PHI
5420 (graphite_regenerate_ast_isl): Adjust.
5421 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5422 (add_write, add_read): New functions.
5423 (build_cross_bb_scalars_def): Use it and simplify.
5424 (build_cross_bb_scalars_use): Likewise.
5425 (graphite_find_cross_bb_scalar_vars): Inline into...
5426 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5427 simulating out-of-SSA. Compute liveout and add dependencies.
5428 (build_scops): Force an empty entry block.
5429 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5431 (sese_build_liveouts): Declare.
5432 (sese_trivially_empty_bb_p): Likewise.
5433 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5434 compute liveout and debug_liveout.
5435 (sese_bad_liveouts_use): Remove.
5436 (sese_reset_debug_liveouts_bb): Likewise.
5437 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5438 (sese_build_liveouts): Build liveout and debug_liveout and store
5440 (new_sese_info): Adjust.
5441 (free_sese_info): Likewise.
5442 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5443 do not build liveout here.
5444 (move_sese_in_condition): Adjust region entry.
5445 (scev_analyzable_p): Match up with chrec_apply requirements.
5446 (sese_trivially_empty_bb_p): New.
5447 * tree-into-ssa.c (get_reaching_def): Properly support generating
5448 default-defs for incremental rewrite of anonymous names.
5450 2017-10-06 Richard Biener <rguenther@suse.de>
5452 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5453 precision do not perform modulo reduction.
5455 2017-10-06 Richard Biener <rguenther@suse.de>
5457 PR tree-optimization/82436
5458 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5459 conservatively choose the vectorization factor when checking
5460 whether we can perform the required load permutation.
5461 (vect_transform_slp_perm_load): Assert when we may not fail.
5463 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5465 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5466 message for incompatible -msdata=* and -mcall-* options.
5468 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5470 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5471 rate for post-reload scheduling.
5473 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5475 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5477 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5479 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5480 to improve monte carlo in scimark.
5482 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5484 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5485 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5486 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5487 pentium4_cost, nocona_cost): Set reassociation width to 1.
5488 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5489 width to 2 for fp operations and 1 otherwise.
5490 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5492 (atom_cost): Set reassociation width to 2.
5493 (slm_cost, generic_cost): Set fp reassociation width
5494 to 2 and 1 otherwise.
5495 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5496 (core_cost): Set fp reassociation width to 4 and vector to 2.
5497 (ix86_reassociation_width): Rewrite using cost table; special case
5498 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5499 and TARGET_AVX128_OPTIMAL.
5500 * config/i386/i386.h (processor_costs): Add
5501 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5502 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5503 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5504 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5505 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5506 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5508 2017-10-05 Nathan Sidwell <nathan@acm.org>
5510 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5512 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5514 * config/arm/arm.c (arm_test_fpu_data): New.
5515 (arm_run_selftests): Call arm_test_fpu_data.
5517 2017-10-04 Nathan Sidwell <nathan@acm.org>
5519 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5521 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5523 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5525 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5526 check type for aarch64_simd_valid_immediate.
5527 (aarch64_output_simd_mov_immediate): Update prototype.
5528 (aarch64_simd_valid_immediate): Update prototype.
5529 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5530 support for ORR-immediate.
5531 (and<mode>3): modified pattern to add support for BIC-immediate.
5532 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5533 now checks for valid immediate for BIC and ORR based on new enum
5535 (aarch64_output_simd_mov_immediate): Function now used to output
5536 BIC/ORR imm as well based on new enum argument.
5537 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5539 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5540 (aarch64_reg_or_bic_imm): Likewise.
5542 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5544 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5545 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5547 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5551 PR rtl-optimization/82396
5552 * haifa-sched.c (autopref_multipass_init): Simplify
5554 (autopref_rank_data): Simplify sort order.
5555 * sched-int.h (autopref_multipass_data_): Remove
5556 multi_mem_insn_p, min_offset and max_offset.
5558 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5560 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5562 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5564 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5567 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5569 * doc/sourcebuild.texi: Document vect_long_mult.
5571 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5573 PR tree-optimization/82413
5574 * fold-const.c (build_range_check): Use widest_int when comparing
5575 the maximum ETYPE value with HIGH.
5577 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5579 PR rtl-optimization/82396
5580 * haifa-sched.c (autopref_multipass_init): Simplify
5582 (autopref_rank_data): Simplify sort order.
5583 * sched-int.h (autopref_multipass_data_): Remove
5584 multi_mem_insn_p, min_offset and max_offset.
5586 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5588 PR tree-optimization/82381
5589 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5590 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5593 PR tree-optimization/82374
5594 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5595 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5596 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5597 current_function_decl to the new decl.
5599 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5601 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5602 helper macro for IEEE float128 hardware built-in functions.
5603 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5605 (TRUNCF128_ODD): Likewise.
5606 (ADDF128_ODD): Likewise.
5607 (SUBF128_ODD): Likewise.
5608 (MULF128_ODD): Likewise.
5609 (DIVF128_ODD): Likewise.
5610 (FMAF128_ODD): Likewise.
5611 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5612 UNSPEC_TRUNC_ROUND_TO_ODD.
5613 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5614 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5615 floating point round to odd instructions.
5616 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5617 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5618 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5619 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5620 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5621 (trunc<mode>sf2_hw): Change the truncate with round to odd
5622 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5623 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5624 to odd hardware instructions.
5625 (sub<mode>3_odd): Likewise.
5626 (mul<mode>3_odd): Likewise.
5627 (div<mode>3_odd): Likewise.
5628 (sqrt<mode>2_odd): Likewise.
5629 (fma<mode>4_odd): Likewise.
5630 (fms<mode>4_odd): Likewise.
5631 (nfma<mode>4_odd): Likewise.
5632 (nfms<mode>4_odd): Likewise.
5633 (trunc<mode>df2_odd): Change the truncate with round to odd
5634 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5636 * doc/extend.texi (PowerPC built-in functions): Update documentation
5637 for existing IEEE float128-bit built-in functions. Add built-in
5638 functions that generate the IEEE 128-bit floating point round to
5641 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5643 PR rtl-optimization/77729
5644 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5645 to (X&(C1&~C2))|C2 transformations.
5647 2017-10-03 Martin Jambor <mjambor@suse.cz>
5649 PR tree-optimization/82363
5650 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5651 mismatch, mark lacc written regardless of racc.
5653 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5655 PR tree-optimization/82381
5656 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5657 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5658 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5659 fallthrough into reassoc_stmt_dominates_stmt_p.
5662 * combine.c (combine_instructions): Don't combine in unreachable
5665 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5668 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5669 function to not use the have_cpu variable. Do not set cpu_index,
5670 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5672 (rs6000_valid_attribute_p): Remove duplicate initializations of
5673 old_optimize and func_optimize.
5674 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5675 (rs6000_activate_target_options): Make global.
5676 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5679 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5681 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5682 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5683 Set *poffset to 0 and *psize and *pmax_size to -1 if
5684 *poffset + *psize overflows in HOST_WIDE_INT.
5686 PR tree-optimization/82387
5687 PR tree-optimization/82388
5688 PR tree-optimization/82389
5689 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5690 instead of live_bytes non-NULL.
5692 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5695 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5698 2017-10-02 Richard Biener <rguenther@suse.de>
5700 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5701 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5702 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5704 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5706 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5707 requested precision matches the type's.
5708 * calls.c (alloc_max_size): Calculate the new candidate size as
5709 a widest_int and use wi::to_widest when comparing it with the
5710 current candidate size.
5711 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5712 zero rather than integer_zero_node.
5713 * match.pd: Check for a no-op conversion before using wi::add
5714 rather than after. Use tree_to_uhwi when summing small shift
5715 counts into an unsigned int.
5717 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5718 Alan Hayward <alan.hayward@arm.com>
5719 David Sherwood <david.sherwood@arm.com>
5722 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5723 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5724 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5725 POINTER_AND_FP_REGS.
5727 2017-10-02 Richard Biener <rguenther@suse.de>
5729 PR tree-optimization/82355
5730 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5731 a mapping for the enclosing loop but avoid generating one for
5733 (copy_bb_and_scalar_dependences): Remove premature codegen
5734 error on PHIs in blocks duplicated into multiple places.
5735 * graphite-scop-detection.c
5736 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5737 in the region use it as loop and nest to analyze the DR in.
5738 (try_generate_gimple_bb): Likewise.
5739 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5740 (add_loop_constraints): For blocks in a loop not in the region
5741 create a dimension with a single iteration.
5742 * sese.h (gbb_loop_at_index): Remove assert.
5744 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5746 * omp-expand.c (adjust_context_scope): New function.
5747 (expand_parallel_call): Call adjust_context_scope.
5749 2017-10-01 Jeff Law <law@redhat.com>
5751 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5752 dom_opt_dom_walker class with direct access to private members.
5753 Add comments. Call test_for_singularity.
5754 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5755 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5756 m_dummy_cond anymore.
5757 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5759 (pass_dominator:execute): Build the dummy_cond here and pass it
5760 to the dom_opt_dom_walker ctor.
5761 (test_for_singularity): New function.
5763 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5764 Maya Rashish <coypu@sdf.org>
5766 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5767 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5768 (alpha*-*-netbsd*) Use nbsd_tm_file.
5769 (arm*-*-netbsdelf*) Likewise.
5770 (i[34567]86-*-netbsdelf*) Likewise.
5771 (x86_64-*-netbsd*) Likewise.
5772 (mips*-*-netbsd*) Likewise.
5773 (powerpc-*-netbsd*) Likewise.
5774 (sh*-*-netbsd*) Likewise.
5775 (sparc-*-netbsdelf*) Likewise.
5776 (sparc64-*-netbsd*) Likewise.
5777 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5779 (vax-*-netbsdelf*) Likewise.
5780 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5781 (UINT_FAST8_TYPE) Likewise.
5782 (INT_FAST16_TYPE) Check CHAR_FAST16.
5783 (UINT_FAST16_TYPE) Likewise.
5785 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5788 * config/i386/i386.md
5789 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5790 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5791 *divmodsi4_zext_2): New define_insn_and_split.
5792 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5793 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5794 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5795 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5796 New define_insn_and_split.
5797 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5798 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5799 operands[1] having DImode when mode is SImode.
5801 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5802 always SImode for DIV and MOD in REG_EQUAL notes.
5804 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5807 * match.pd: Fix handling of NaNs in pattern.
5809 2017-09-29 Jeff Law <law@redhat.com>
5811 * sbitmap.c (bitmap_bit_in_range_p): New function.
5812 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5813 * tree-ssa-dse.c (live_bytes_read): New function.
5814 (dse_classify_store): Ignore reads of dead bytes.
5816 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5817 typos and whitespace errors.
5818 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5819 * config/s390/s390.c (s390_emit_prologue): Likewise.
5821 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5824 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5825 with a symbol for LRA.
5827 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5829 PR rtl-optimization/82338
5830 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5832 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5834 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5835 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5836 (qsort_chk): Declare.
5837 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5838 (qsort_chk): New function.
5840 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5842 PR tree-optimization/82337
5843 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5844 phi definition if the PHI result appears in an abnormal PHI.
5845 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5846 basis appears in an abnormal PHI.
5848 2017-09-29 Richard Biener <rguenther@suse.de>
5850 * graphite-isl-ast-to-gimple.c
5851 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5852 (binary_op_to_tree): Use it.
5853 (get_rename_from_scev): Likewise.
5854 (copy_loop_phi_nodes): Likewise.
5855 (copy_bb_and_scalar_dependences): Likewise.
5856 (translate_pending_phi_nodes): Likewise.
5858 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5861 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5862 for movabsq $(i32 << shift), r64.
5864 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5866 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5867 index when encoding %esp as %rsp to avoid 0x67 prefix.
5869 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5871 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5872 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5874 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5876 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5877 Extensions with more than 16 double VFP registers.
5878 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5879 to_clear_mask and all code related to it. Replace the remaining
5880 entry by a sbitmap and adapt code accordingly.
5882 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5884 * brig-builtins.def: Change pure attributes to const.
5886 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5888 * config.gcc (default_gnu_indirect_function): Default to yes for
5889 sparc*-*-linux* with glibc.
5891 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5893 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5894 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5895 when creating .init_array and .fini_array sections with priority
5898 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5901 * config/aarch64/aarch64.c
5902 (aarch64_builtin_support_vector_misalignment): Always return false
5903 when misalignment is unknown.
5905 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5907 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5908 this function to return false if the definition used by the swap
5909 instruction is artificial, or if the memory address from which the
5910 constant value is loaded is not represented by a base address held
5911 in a register or if the base address register is a frame or stack
5912 pointer. Additionally, return false if the base address of the
5913 loaded constant is a SYMBOL_REF but is not considered to be a
5915 (replace_swapped_load_constant): New function.
5916 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5917 loaded constant vector with a load of a swapped constant vector.
5919 2017-09-27 Carl Love <cel@us.ibm.com>
5921 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5922 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5923 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5926 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5928 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5929 first, always call autopref_rank_data otherwise.
5931 2017-09-27 Richard Biener <rguenther@suse.de>
5933 * graphite-scop-detection.c (find_scop_parameters): Move
5934 loop bound handling ...
5935 (gather_bbs::before_dom_children): ... here, avoiding the need
5936 to build scop_info->loop_nest.
5937 (record_loop_in_sese): Remove.
5938 * sese.h (sese_info_t::loop_nest): Remove.
5939 * sese.c (new_sese_info): Do not allocate loop_nest.
5940 (free_sese_info): Do not free loop_nest.
5942 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5945 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5946 lhs from calls if the lhs has addressable type.
5948 2017-09-27 Richard Biener <rguenther@suse.de>
5950 * graphite.h (scop::max_alias_set): New member.
5951 * graphite-scop-detection.c: Remove references to non-existing
5952 --param in comments.
5953 (build_alias_sets): Record the maximum alias set used for drs.
5954 (build_scops): Support zero as unlimited for
5955 --param graphite-max-arrays-per-scop.
5956 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5958 (build_poly_sr_1): ... here. Compute alias set based on the
5959 maximum alias set used for drs rather than
5960 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5962 2017-09-27 Richard Biener <rguenther@suse.de>
5964 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5965 --param loop-block-tile-size=0 to disable tiling.
5967 2017-09-27 Richard Biener <rguenther@suse.de>
5969 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5970 (graphite-max-nb-scop-params): Document special value zero.
5971 * domwalk.h (dom_walker::STOP): New symbolical constant.
5972 (dom_walker::dom_walker): Add optional parameter for bb to
5974 (dom_walker::~dom_walker): Declare.
5975 (dom_walker::before_dom_children): Document STOP return value.
5976 (dom_walker::m_user_bb_to_rpo): New member.
5977 (dom_walker::m_bb_to_rpo): Likewise.
5978 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5979 mapping here if not provided by the user.
5980 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5981 provided by the user.
5982 (dom_walker::STOP): Define.
5983 (dom_walker::walk): Do not compute bb to RPO mapping here.
5984 Support STOP return value from before_dom_children to stop
5986 * graphite-optimize-isl.c (optimize_isl): If the schedule
5987 is the same still generate code if -fgraphite-identity
5988 or -floop-parallelize-all are given.
5989 * graphite-scop-detection.c: Include cfganal.h.
5990 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5992 (gather_bbs::before_dom_children): Return STOP for BBs
5994 (build_scops): Compute bb to RPO mapping and pass it to
5995 the domwalk. Treat --param graphite-max-nb-scop-params=0
5996 as not limiting the number of params.
5997 * graphite.c (graphite_initialize): Remove limit on the
5998 number of basic-blocks in a function.
5999 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6000 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6001 default value of 10.
6003 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6005 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6006 Adjust code to eliminate needing to do the shift right 32-bits
6007 operation after XSCVDPSPN.
6009 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6011 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6012 ((X / Y) != 0 -> X >= Y): Likewise.
6014 2017-09-26 Carl Love <cel@us.ibm.com>
6016 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6017 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6018 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6019 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6020 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6021 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6022 definitions and overloading.
6023 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6024 statement for P9V_BUILTIN_XST_LEN_R.
6025 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6026 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6027 define_expand and define_insn for the instructions and builtins.
6028 * doc/extend.texi: Update the built-in documentation file for the new
6030 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6031 define_insn for the instructions
6033 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6036 * gcc/config/netbsd-protos.h: New file.
6037 * gcc/config/netbsd.c: New file.
6038 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6039 * gcc/config/t-netbsd: New file.
6040 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6041 (tmake_file) Add t-netbsd.
6042 (extra_objs) Add netbsd.o.
6044 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6048 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6050 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6052 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6053 sign extension from a vector register to a GPR by doing a 32-bit
6054 direct move and then an EXTSW.
6055 (extendsi<mode>2 splitter): Likewise.
6056 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6057 right or vector extract after doing XSCVDPSPN. Use
6058 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6060 (movdi_from_sf_zero_ext): Likewise.
6061 (reload_gpr_from_vsxsf): Likewise.
6062 (p8_mfvsrd_4_disf): Delete, no longer used.
6063 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6064 then needing to move the SFmode to a GPR to use the XSCVDPSP
6065 instruction instead of FRSP and XSCVDPSPN.
6066 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6067 it is adjacent to the other XSCVSPDP insns.
6068 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6069 SFmode to be in traditional Altivec registers.
6070 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6071 (vsx_xscvspdpn): Likewise.
6072 (vsx_xscvspdpn_scalar): Likewise.
6074 2017-09-26 Martin Jambor <mjambor@suse.cz>
6076 * tree-sra.c (compare_access_positions): Put integral types first,
6077 stabilize sorting of integral types, remove conditions putting
6078 non-full-precision integers last.
6079 (sort_and_splice_var_accesses): Disable scalarization if a
6080 non-integert would be represented by a non-full-precision integer.
6082 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6084 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6085 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6086 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6087 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6088 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6089 conditionals inside the function instead of around it. Call
6090 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6091 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6093 2017-09-26 Richard Biener <rguenther@suse.de>
6095 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6097 (scop_detection::build_scop_breadth): ... this. Removed.
6098 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6099 (scop_detection::harmful_stmt_in_bb): Likewise.
6100 (scop_detection::graphite_can_represent_stmt): Likewise.
6101 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6102 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6103 (scop_detection::can_represent_loop_1): ... this. Removed.
6104 (scop_detection::harmful_loop_in_region): Simplify after inlining
6105 the above and remove more quadraticness.
6106 (build_scops): Adjust.
6107 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6110 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6113 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6114 REGNO (base) == SP_REG if base is a REG.
6117 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6118 if it is different SSA_NAME.
6119 (optimize_range_tests_cmp_bitwise): New function.
6120 (optimize_range_tests): Call it.
6122 2017-09-26 Richard Biener <rguenther@suse.de>
6124 PR tree-optimization/82321
6125 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6126 for the def being inside the loop.
6128 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6130 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6132 * config/s390/s390-builtins.def: Fix constraint on op4.
6134 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6136 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6137 independent expanders.
6138 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6139 ("vec_ordered", "vec_unordered"): New expanders.
6141 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6143 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6146 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6148 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6149 vec_unpacks_lo_v16qi.
6150 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6152 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6154 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6155 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6156 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6158 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6160 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6162 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6164 ("*vec_slb<mode>"): New insn pattern.
6165 ("vec_shr_<mode>"): New expander.
6166 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6167 and force the shift count operand to V16QImode.
6168 ("vec_srb<mode>"): Set shift count mode to V16QI.
6170 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6172 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6173 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6174 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6176 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6179 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6181 2017-09-26 Richard Biener <rguenther@suse.de>
6183 PR tree-optimization/82320
6184 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6187 2017-09-25 Jeff Law <law@redhat.com>
6189 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6190 prototype for new argument.
6191 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6192 mostly extracted from rs6000_emit_allocate_stack.
6193 (rs6000_emit_probe_stack_range_stack_clash): New function.
6194 (rs6000_emit_allocate_stack): Call
6195 rs6000_emit_probe_stack_range_stack_clash as needed.
6196 (rs6000_emit_probe_stack_range): Add additional argument
6197 to call to gen_probe_stack_range{si,di}.
6198 (output_probe_stack_range): New.
6199 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6200 (output_probe_stack_range_stack_clash): New.
6201 (rs6000_emit_prologue): Emit notes into dump file as requested.
6202 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6203 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6204 Add additional operand and pass it to output_probe_stack_range.
6206 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6208 PR tree-optimization/82163
6209 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6210 (checking_verify_loop_closed_ssa): New parameter.
6211 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6212 (check_loop_closed_ssa_stmt): Delete.
6213 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6214 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6215 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6218 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6220 * brig-builtins.def: Treat HSAIL barrier builtins as
6221 setjmp/longjump style functions.
6223 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6225 * target.def (constant_alignment): New hook.
6226 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6227 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6228 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6229 * doc/tm.texi: Regenerate.
6230 * targhooks.h (default_constant_alignment): Declare.
6231 (constant_alignment_word_strings): Likewise.
6232 * targhooks.c (default_constant_alignment): New function.
6233 (constant_alignment_word_strings): Likewise.
6234 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6235 instead of CONSTANT_ALIGNMENT.
6236 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6237 (force_const_mem): Likewise.
6238 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6239 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6240 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6241 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6242 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6244 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6245 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6246 constant_alignment_word_strings.
6247 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6248 (CONSTANT_ALIGNMENT): Likewise.
6249 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6250 (arm_constant_alignment): New function.
6251 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6252 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6253 constant_alignment_word_strings.
6254 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6255 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6256 constant_alignment_word_strings.
6257 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6258 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6259 (cris_constant_alignment): New function.
6260 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6261 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6262 (epiphany_constant_alignment): New function.
6263 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6264 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6265 constant_alignment_word_strings.
6266 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6267 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6268 constant_alignment_word_strings.
6269 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6270 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6271 constant_alignment_word_strings.
6272 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6273 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6274 * config/i386/i386.c (ix86_constant_alignment): Make static.
6275 Use the same interface as the target hook.
6276 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6277 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6278 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6279 constant_alignment_word_strings.
6280 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6281 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6282 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6283 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6284 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6285 constant_alignment_word_strings.
6286 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6287 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6288 constant_alignment_word_strings.
6289 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6290 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6291 constant_alignment_word_strings.
6292 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6293 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6295 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6296 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6297 * config/mips/mips.c (mips_constant_alignment): New function.
6298 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6299 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6300 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6301 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6302 (mmix_constant_alignment): Make static. Use the same interface
6304 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6305 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6306 constant_alignment_word_strings.
6307 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6308 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6309 constant_alignment_word_strings.
6310 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6311 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6312 constant_alignment_word_strings.
6313 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6314 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6315 (rs6000_constant_alignment): New function.
6316 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6317 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6318 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6319 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6320 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6321 (rs6000_constant_alignment): New function.
6322 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6323 * config/s390/s390.c (s390_constant_alignment): New function.
6324 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6325 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6326 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6327 constant_alignment_word_strings.
6328 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6329 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6330 (sparc_constant_alignment): New function.
6331 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6332 * config/spu/spu.c (spu_constant_alignment): New function.
6333 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6334 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6335 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6336 constant_alignment_word_strings.
6337 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6338 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6339 constant_alignment_word_strings.
6340 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6341 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6342 constant_alignment_word_strings.
6343 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6344 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6345 (visium_constant_alignment): New function.
6346 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6347 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6348 (xtensa_constant_alignment): New function.
6349 * system.h (CONSTANT_ALIGNMENT): Poison.
6351 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6353 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6354 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6355 (rs6000_builtin_valid_without_lhs): New helper function.
6356 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6357 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6359 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6361 * target.h (vec_perm_indices): Use unsigned short rather than
6363 (auto_vec_perm_indices): Likewise.
6364 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6365 Use unsigned int rather than unsigned char.
6366 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6368 2017-09-25 Richard Biener <rguenther@suse.de>
6370 * cfgloop.h (sort_sibling_loops): Declare.
6371 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6372 (sort_sibling_loops): New function sorting the sibling loop list
6374 * graphite.c (graphite_transform_loops): Sort sibling loops.
6376 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6378 * target.def (vec_perm_const_ok): Change sel parameter to
6380 * optabs-query.c (can_vec_perm_p): Update accordingly.
6381 * doc/tm.texi: Regenerate.
6382 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6383 auto_vec_perm_indices and remove separate nelt field.
6384 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6385 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6386 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6387 (aarch64_expand_vec_perm_const): Update accordingly.
6388 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6389 to vec_perm_indices.
6390 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6391 auto_vec_perm_indices and remove separate nelt field.
6392 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6393 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6394 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6396 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6397 to vec_perm_indices.
6398 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6399 sel to vec_perm_indices.
6400 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6401 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6402 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6404 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6407 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6410 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6411 on the FUNCTION_DECL function context if it has a DIE that is a
6414 2017-09-25 Richard Biener <rguenther@suse.de>
6416 PR tree-optimization/82285
6417 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6420 2017-09-25 Tom de Vries <tom@codesourcery.com>
6424 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6427 2017-09-25 Richard Biener <rguenther@suse.de>
6429 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6430 ISL errors other than isl_error_quota happen. Dump if the
6431 schedule is the same.
6432 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6433 errors instead of aborting inside ISL.
6435 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6438 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6439 of libgcc_eh for m64.
6440 * config/i386/darwin64.h: Likewise.
6442 2017-09-25 Richard Biener <rguenther@suse.de>
6445 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6446 attribute for incomplete types nor twice for complete ones.
6448 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6451 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6452 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6455 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6458 * config/i386/i386.opt (mprefer-avx256): Use
6459 ix86_target_flags variable.
6460 * config/i386/i386.c (ix86_target_string): Move
6461 -mprefer-avx256 to flag2_opts.
6463 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6466 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6467 and x != -1 | y != -1 into (x & y) != -1.
6469 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6471 * config.gcc: Add new case statement to set
6472 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6473 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6474 s390x-*-linux* case statements. Added aarch64 to the list of
6475 supported architectures.
6477 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6479 PR tree-optimization/82289
6480 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6481 STMT_VINFO_RELEVANT_P.
6483 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6484 Alan Hayward <alan.hayward@arm.com>
6485 David Sherwood <david.sherwood@arm.com>
6487 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6488 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6489 (extract_range_from_binary_expr_1): Don't call
6490 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6492 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6493 Alan Hayward <alan.hayward@arm.com>
6494 David Sherwood <david.sherwood@arm.com>
6496 * target.def (preferred_vector_alignment): New hook.
6497 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6499 * doc/tm.texi: Regenerate.
6500 * targhooks.h (default_preferred_vector_alignment): Declare.
6501 * targhooks.c (default_preferred_vector_alignment): New function.
6502 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6504 (DR_TARGET_ALIGNMENT): New macro.
6505 (aligned_access_p): Update commentary.
6506 (vect_known_alignment_in_bytes): New function.
6507 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6509 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6510 Calculate the misalignment based on the target alignment rather than
6512 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6513 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6514 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6515 the target alignment, rather than masking the element misalignment
6516 with the number of elements in a vector. Also use the target
6517 alignment when calculating the maximum number of peels.
6518 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6519 instead of TYPE_ALIGN_UNIT.
6520 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6521 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6522 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6523 (vect_create_data_ref_ptr): Likewise.
6524 (vect_setup_realignment): Realign by ANDing with
6525 -DR_TARGET_MISALIGNMENT.
6526 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6527 the number of peels based on DR_TARGET_ALIGNMENT.
6528 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6529 with the guaranteed alignment boundary when deciding whether
6531 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6532 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6533 (ensure_base_align): Remove stmt_info parameter. Get the
6534 target base alignment from DR_TARGET_ALIGNMENT.
6535 (vectorizable_store): Update call accordingly. Interpret
6536 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6538 (vectorizable_load): Likewise.
6540 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6541 Alan Hayward <alan.hayward@arm.com>
6542 David Sherwood <david.sherwood@arm.com>
6544 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6545 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6546 (vect_enhance_data_refs_alignment): Likewise.
6548 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6550 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6551 error. Only quit immediately if parsing is complete.
6552 (BEGIN): Initialize fatal_err and parse_done.
6553 (begin fpu, end fpu): Check number of arguments.
6554 (begin arch, end arch): Likewise.
6555 (begin cpu, end cpu): Likewise.
6556 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6557 (optalias): Likewise.
6559 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6561 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6562 * config/arm/arm-isa.h: Delete. Move definitions to ...
6563 * arm-cpus.in: ... here. Use new feature and fgroup values.
6564 * config/arm/arm.c (arm_option_override): Use lower case for feature
6566 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6567 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6568 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6570 (print_isa_bits_for): New function.
6571 (gen_isa): New function.
6572 (gen_comm_data): Use print_isa_bits_for.
6573 (define feature): New keyword.
6574 (define fgroup): New keyword.
6575 * config/arm/t-arm (TM_H): Remove.
6576 (GTM_H): Add arm-isa.h.
6577 (arm-isa.h): Add rule to generate file.
6578 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6579 case for feature bit names.
6581 2017-09-22 Richard Biener <rguenther@suse.de>
6583 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6585 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6586 print of no dependency loops ...
6587 * graphite.c (graphite_transform_loops): ... here.
6588 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6590 (same_close_phi_node, remove_duplicate_close_phi,
6591 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6592 (canonicalize_loop_closed_ssa): ... here and simplify.
6593 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6594 (optimize_isl): Use dump_printf_loc to tell when we stopped
6595 optimizing because of an ISL timeout.
6597 2017-09-22 Richard Biener <rguenther@suse.de>
6599 PR tree-optimization/82291
6600 * tree-if-conv.c (predicate_mem_writes): Make sure to
6601 remove writes in blocks predicated with false.
6603 2017-09-22 Richard Biener <rguenther@suse.de>
6605 * sese.c: Include cfganal.h.
6606 (if_region_set_false_region): Remove.
6607 (create_if_region_on_edge): Likewise.
6608 (move_sese_in_condition): Re-implement without destroying
6611 2017-09-22 Richard Biener <rguenther@suse.de>
6613 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6614 Verify both BBs contain loop PHI nodes before dispatching to
6616 (graphite_regenerate_ast_isl): Do not recompute dominators,
6617 do not verify three times. Restructure for clarity.
6618 * graphite-scop-detection.c (same_close_phi_node,
6619 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6620 defined_in_loop_p, canonicalize_loop_closed_ssa,
6621 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6622 checking and SSA rewrite, move to ...
6623 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6624 (graphite_initialize): Do not pass in ctx, do not reset the
6625 SCEV cache, compute only dominators.
6626 (graphite_transform_loops): Allocate ISL ctx after
6627 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6628 Maintain post-dominators only around build_scops.
6629 * sese.c (if_region_set_false_region): Make static. Free
6630 and recompute dominators.
6631 (move_sese_in_condition): Assert we don't get called with
6632 post-dominators computed.
6633 * sese.h (if_region_set_false_region): Remove.
6635 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6637 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6638 mode attribute for TARGET_AVX512VL.
6640 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6642 * config/i386/i386.opt (mprefer-avx256): New option.
6643 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6645 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6646 for TARGET_PREFER_AVX256.
6647 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6649 2017-09-21 Jeff Law <law@redhat.com>
6651 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6652 Fix dump output if the only stack space is for pushed registers.
6654 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6656 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6659 2017-09-21 Martin Sebor <msebor@redhat.com>
6662 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6663 code (in C++) or code that triggers warnings.
6665 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6667 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6669 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6671 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6672 * sched-rgn.c: Ditto.
6673 * sel-sched-ir.c: Ditto.
6675 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6677 * toplev.h (set_random_seed): Adjust return type.
6678 * toplev.c (init_local_tick): Move eager initialization of random_seed
6679 to get_random_seed. Adjust comment.
6680 (init_random_seed): Inline to get_random_seed, delete.
6681 (get_random_seed): Initialize random_seed lazily.
6682 (set_random_seed): Do not return previous value.
6683 (print_switch_value): Do not call get_random_seed.
6685 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6687 * cgraph.c (delete_function_version): New, broken out from...
6688 (cgraph_node::delete_function_version): ...here. Rename to
6689 cgraph_node::delete_function_version_by_decl. Update all uses.
6690 (cgraph_node::remove): Call delete_function_version.
6692 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6695 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6696 VAR_DECLs to which addressable non-volatile parameters are mapped
6697 and for id->retvar after the return value assignment. Clear
6698 id->retval and id->retbnd after inlining.
6700 2017-09-21 Richard Biener <rguenther@suse.de>
6702 PR tree-optimization/82276
6703 PR tree-optimization/82244
6704 * tree-vrp.c (build_assert_expr_for): Set
6705 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6707 (remove_range_assertions): Revert earlier change.
6709 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6712 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6714 2017-09-21 Richard Biener <rguenther@suse.de>
6716 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6717 Restore valid IL after code generation errors.
6718 * graphite.c (graphite_transform_loops): Diagnose code
6719 generation issues as MSG_MISSED_OPTIMIZATION and continue
6720 with processing SCOPs.
6722 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6723 Alan Hayward <alan.hayward@arm.com>
6724 David Sherwood <david.sherwood@arm.com>
6726 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6727 rather than choosing between plus_constant and gen_rtx_<CODE>.
6728 * expr.c (emit_push_insn): Likewise.
6729 (expand_expr_real_2): Likewise.
6731 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6732 Alan Hayward <alan.hayward@arm.com>
6733 David Sherwood <david.sherwood@arm.com>
6735 * loop-unroll.c (split_iv): Call copy_rtx on the step.
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 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6742 calling tree_to_uhwi.
6744 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6745 Alan Hayward <alan.hayward@arm.com>
6746 David Sherwood <david.sherwood@arm.com>
6748 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6749 INTEGER_CST rather than a negative test for ADDR_EXPR.
6751 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6752 Alan Hayward <alan.hayward@arm.com>
6753 David Sherwood <david.sherwood@arm.com>
6755 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6756 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6758 2017-09-21 Richard Biener <rguenther@suse.de>
6760 PR tree-optimization/71351
6761 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6762 graphite_create_new_loop_guard): Remove, fold remaining parts
6764 (translate_isl_ast_node_for): ... here and simplify.
6766 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6769 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6770 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6771 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6772 alternative always use QI mode, for -Os imov (=R,R) alternative
6773 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6776 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6777 Jeff Law <law@redhat.com>
6779 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6780 (allocate_stack_space): New function, partially extracted from
6782 (s390_emit_prologue): Track offset to most recent stack probe.
6783 Code to allocate space moved into allocate_stack_space.
6784 Dump actions when no stack is allocated.
6785 (s390_prologue_plus_offset): New function.
6786 (s390_emit_stack_probe): Likewise.
6788 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6790 * common.opt (Wa, Wl, Wp, g, gz=): Add
6792 (gno-column-info): Remove.
6793 (gcolumn-info): Drop RejectNegative.
6795 (gno-record-gcc-switches): Remove.
6796 (grecord-gcc-switches): Drop RejectNegative.
6797 (gno-split-dwarf): Remove.
6798 (gsplit-dwarf): Drop RejectNegative.
6799 (gno-strict-dwarf): Remove.
6800 (gstrict-dwarf): Drop RejectNegative.
6801 * config/darwin.opt (gfull, gused): Add RejectNegative.
6802 * dwarf2out.c (gen_producer_string): Drop
6803 gno-record-gcc-switches handler.
6804 * optc-gen.awk: Add g to prefixes with negative forms.
6805 * opts-common.c (remapping_prefix_p): New.
6806 (find_opt): Check it.
6807 (generate_canonical_option): Test g prefix.
6808 (option_map): Add -gno- mapping.
6809 (add_misspelling_candidates): Check remapping_prefix_p.
6811 2017-09-20 Jeff Law <law@redhat.com>
6813 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6814 thinko in stack clash protection support.
6816 * explow.c (compute_stack_clash_protection_loop_data): Use
6817 CONST_INT_P instead of explicit test. Verify object is a
6818 CONST_INT_P before looking at INTVAL.
6819 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6820 instead of explicit test.
6822 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6825 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6826 address instead of to r1 and r11.
6828 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6830 * config.gcc: Support "knm".
6831 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6832 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6834 * config/i386/i386.c (m_KNM): Define.
6835 (processor_target_table): Add "knm".
6837 (ix86_option_override_internal): Add "knm".
6838 (ix86_issue_rate): Add PROCESSOR_KNM.
6839 (ix86_adjust_cost): Ditto.
6840 (ia32_multipass_dfa_lookahead): Ditto.
6841 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6842 (fold_builtin_cpu): Add M_INTEL_KNM.
6843 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6844 (processor_type): Add PROCESSOR_KNM.
6845 * config/i386/x86-tune.def: Add m_KNM.
6846 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6848 2017-09-20 Richard Biener <rguenther@suse.de>
6850 PR tree-optimization/80213
6851 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6852 are allowed in empty BBs as well.
6853 (canonicalize_loop_closed_ssa): Also look for other complex
6855 (scop_detection::get_sese): Include the loop-closed PHI block
6857 (scop_detection::merge_sese): Remove code adding extra blocks.
6858 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6859 (build_scops): Assert the final returned scop is invalid.
6861 2017-09-20 Richard Biener <rguenther@suse.de>
6863 PR tree-optimization/82264
6864 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6866 (vn_phi_lookup): Likewise.
6867 (vn_phi_insert): Likewise.
6869 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6871 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6872 that fits into uhwi or shwi, add DW_AT_const_value regardless
6873 of early_dwarf without going through RTL, using add_AT_unsigned
6876 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6877 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6878 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6879 (DEBUG_MACRO_SECTION): Likewise.
6880 (DEBUG_LTO_MACRO_SECTION): Likewise.
6881 (DEBUG_STR_DWO_SECTION): Likewise.
6882 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6883 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6884 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6885 (DEBUG_LTO_LINE_STR_SECTION): Define.
6886 (init_sections_and_labels): Initialize debug_line_str_section
6887 variable. Initialize debug_loc_section for -gdwarf-5 to
6888 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6890 2017-09-20 Richard Biener <rguenther@suse.de>
6892 * graphite-sese-to-poly.c (extract_affine): Properly handle
6893 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6895 2017-09-20 Richard Biener <rguenther@suse.de>
6897 PR tree-optimization/81373
6898 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6899 Force SESE live-out defs to be handled even if they are
6902 2017-09-19 Jeff Law <law@redhat.com>
6904 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6905 nothing for stack adjustments with REG_STACK_CHECK.
6906 * sched-deps.c (parse_add_or_inc): Reject insns with
6907 REG_STACK_CHECK from dependency breaking.
6908 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6909 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6910 * reg-notes.def (STACK_CHECK): New note.
6912 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6913 (ix86_expand_prologue): Dump stack clash info as needed.
6914 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6916 * function.c (dump_stack_clash_frame_info): New function.
6917 * function.h (dump_stack_clash_frame_info): Prototype.
6918 (enum stack_clash_probes): New enum.
6920 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6921 flag_stack_clash_protection.
6922 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6923 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6924 (arm_frame_pointer_required): Likewise.
6925 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6926 (ia64_expand_prologue): Likewise.
6927 * config/mips/mips.c (mips_expand_prologue): Likewise.
6928 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6929 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6930 (sparc_flat_expand_prologue): Likewise.
6931 * config/spu/spu.c (spu_expand_prologue): Likewise.
6933 * explow.c: Include "params.h".
6934 (anti_adjust_stack_and_probe_stack_clash): New function.
6935 (get_stack_check_protect): Likewise.
6936 (compute_stack_clash_protection_loop_data): Likewise.
6937 (emit_stack_clash_protection_loop_start): Likewise.
6938 (emit_stack_clash_protection_loop_end): Likewise.
6939 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6940 Use anti_adjust_stack_and_probe_stack_clash.
6941 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6942 (emit_stack_clash_protection_loop_start): Likewise.
6943 (emit_stack_clash_protection_loop_end): Likewise.
6944 * rtl.h (get_stack_check_protect): Prototype.
6945 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6946 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6947 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6949 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6951 * doc/tm.texi: Rebuilt.
6952 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6953 get_stack_check_protect.
6954 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6955 * config/arm/arm.c (arm_expand_prologue): Likewise.
6956 (arm_frame_pointer_required): Likewise.
6957 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6958 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6959 * config/mips/mips.c (mips_expand_prologue): Likewise.
6960 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6961 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6962 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6963 (sparc_flat_expand_prologue): Likewise.
6965 * common.opt (-fstack-clash-protection): New option.
6966 * flag-types.h (enum stack_check_type): Note difference between
6967 -fstack-check= and -fstack-clash-protection.
6968 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6969 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6970 * toplev.c (process_options): Issue warnings/errors for cases
6971 not handled with -fstack-clash-protection.
6972 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6973 (-fstack-check): Note additional problem with -fstack-check=generic.
6974 Note that -fstack-check is primarily for Ada and refer users
6975 to -fstack-clash-protection for stack-clash-protection.
6976 Document new params for stack clash protection.
6978 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6980 * config/i386/i386.c (ix86_split_long_move): Do not handle
6981 address used for LEA in a special way.
6983 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6985 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6987 2017-09-19 Martin Sebor <msebor@redhat.com>
6990 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6991 of incompatible types.
6993 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6995 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6996 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6997 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6998 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7000 2017-09-19 Richard Biener <rguenther@suse.de>
7002 PR tree-optimization/82244
7003 * tree-vrp.c (remove_range_assertions): Do not propagate
7004 a constant to abnormals but replace the assert with a copy.
7006 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7008 PR rtl-optimization/57878
7009 PR rtl-optimization/68988
7010 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7011 avoidance test involving non_reload_pseudos. Move frequency test
7012 below the general fragmentation avoidance test.
7014 2017-09-19 Richard Biener <rguenther@suse.de>
7016 PR tree-optimization/69728
7017 * graphite-sese-to-poly.c (schedule_error): New global.
7018 (add_loop_schedule): Handle empty domain by failing the
7020 (build_original_schedule): Handle schedule_error.
7022 2017-09-19 Richard Biener <rguenther@suse.de>
7024 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7025 Do not iterate to sibling loops but only to siblings of inner
7028 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7031 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7032 that the registers are different.
7034 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7036 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7037 to processor_model and "amdfam17h" to arch_names_table.
7038 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7040 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7043 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7045 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7046 Alan Hayward <alan.hayward@arm.com>
7047 David Sherwood <david.sherwood@arm.com>
7049 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7051 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7053 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7054 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7055 vect_schedule_slp_instance.
7056 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7057 Update call to vect_slp_analyze_node_operations. Simplify return
7059 (vect_slp_analyze_bb_1): Update call accordingly.
7060 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7061 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7062 (vect_schedule_slp): Update call accordingly.
7064 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7065 Alan Hayward <alan.hayward@arm.com>
7066 David Sherwood <david.sherwood@arm.com>
7068 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7069 with types that aren't in fact scalar.
7071 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7073 * tree-vect-slp.c (vect_record_max_nunits): New function,
7075 (vect_build_slp_tree_1): ...here.
7076 (vect_build_slp_tree_2): Call it for phis too.
7078 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7080 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7081 to vect_get_vec_def_for_operand when getting the mask operand.
7083 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7084 Alan Hayward <alan.hayward@arm.com>
7085 David Sherwood <david.sherwood@arm.com>
7087 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7090 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7091 Alan Hayward <alan.hayward@arm.com>
7092 David Sherwood <david.sherwood@arm.com>
7094 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7095 calculation for vector booleans.
7097 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7098 Alan Hayward <alan.hayward@arm.com>
7099 David Sherwood <david.sherwood@arm.com>
7101 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7103 (vect_transform_stmt): ...here.
7104 (vect_analyze_stmt): Use it instead of calling
7105 vectorizable_live_operation directly.
7107 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7109 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7110 non-SIMT targets in acc vector loops.
7112 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7114 * configure.ac: Add arc and check if assembler supports gdwarf2.
7115 * configure: Regenerate.
7117 2017-09-18 Richard Biener <rguenther@suse.de>
7119 PR tree-optimization/82220
7120 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7121 epilogue niters from the min_profitable_iters compute.
7123 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7126 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7127 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7128 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7130 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7133 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7134 switching to a new text section.
7136 2017-09-18 Richard Biener <rguenther@suse.de>
7138 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7140 (build_alias_set): Reject aliases with no access function.
7142 2017-09-18 Richard Biener <rguenther@suse.de>
7144 PR tree-optimization/79622
7145 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7147 (build_cross_bb_scalars_use): Likewise.
7149 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7151 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7153 2017-09-18 Alan Modra <amodra@gmail.com>
7156 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7157 stack_pointer_rtx for count 0. Update comments. Break up
7158 large rtl expression.
7160 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7162 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7163 Increase to 20 bytes.
7164 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7165 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7166 or avx version of the stub.
7168 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7171 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7172 compute the minimum stack alignment. Also update preferred stack
7173 boundary for leaf functions.
7175 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7177 PR tree-optimization/82228
7178 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7181 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7183 * common/config/nds32/nds32-common.c
7184 (nds32_option_optimization_table): Refine formatting.
7185 (nds32_option_optimization_table): Use -fsched-pressure and
7186 -fomit-frame-pointer for specific optimization level.
7188 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7190 * config/nds32/nds32.c: Refine formatting and comments.
7191 * config/nds32/nds32.h: Likewise.
7192 * config/nds32/nds32.md: Likewise.
7193 * config/nds32/nds32-cost.c: Likewise.
7194 * config/nds32/nds32-isr.c: Likewise.
7195 * config/nds32/nds32-md-auxiliary.c: Likewise.
7196 * config/nds32/nds32-multiple.md: Likewise.
7197 * config/nds32/nds32-predicates.c: Likewise.
7199 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7200 Jakub Jelinek <jakub@redhat.com>
7202 Add support for -std=c++2a.
7203 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7205 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7207 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7210 * doc/extend.texi (Common Function Attributes): Add
7211 references to ARM, AArch64, and S/390 specific attributes.
7212 (Function Specific Option Pragmas): Add AArch64 and S/390
7213 to list of back ends that support the target pragma.
7215 2017-09-15 Nathan Sidwell <nathan@acm.org>
7217 * doc/standards.texi: Fix C++17 description. Update URLs for
7220 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7222 * common.opt (Wcast-align=strict): New warning option.
7223 * doc/invoke.texi: Document -Wcast-align=strict.
7225 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7227 * cgraph.h (cgraph_thunk_info): Add comments.
7228 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7229 assert for VIRTUAL_* arguments stricter.
7231 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7233 PR tree-optimization/71026
7234 * match.pd: Move RDIV patterns from fold-const.c
7235 * fold-const.c (distribute_real_division): Removed.
7236 (fold_binary_loc): Remove calls to distribute_real_divison.
7238 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7240 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7241 c++1z and gnu++1z as deprecated. Change other references to
7242 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7243 Change -Wc++1z-compat to -Wc++17-compat.
7244 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7245 * dwarf2out.c (highest_c_language): Handle C++17.
7246 (gen_compile_unit_die): Likewise.
7248 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7250 PR rtl-optimization/82192
7251 * combine.c (make_extraction): Don't look through non-paradoxical
7252 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7255 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7256 Alan Hayward <alan.hayward@arm.com>
7257 David Sherwood <david.sherwood@arm.com>
7259 * target.def (function_arg_offset): New hook.
7260 * targhooks.h (default_function_arg_offset): Declare.
7261 * targhooks.c (default_function_arg_offset): New function.
7262 * function.c (locate_and_pad_parm): Use
7263 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7264 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7265 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7266 * doc/tm.texi: Regenerate.
7267 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7268 * config/spu/spu.c (spu_function_arg_offset): New function.
7269 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7270 * system.h (FUNCTION_ARG_OFFSET): Poison.
7272 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7273 Alan Hayard <alan.hayward@arm.com>
7274 David Sherwood <david.sherwood@arm.com>
7276 * target.def (truly_noop_truncation): New hook.
7277 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7278 than TRULY_NOOP_TRUNCATION.
7279 * hooks.h (hook_bool_uint_uint_true): Declare.
7280 * hooks.c (hook_bool_uint_uint_true): New function.
7281 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7282 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7283 * doc/tm.texi: Regenerate.
7284 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7285 rather than TRULY_NOOP_TRUNCATION in comments.
7286 (simplify_comparison): Likewise.
7287 (record_truncated_value): Likewise.
7288 * expmed.c (extract_bit_field_1): Likewise.
7289 (extract_split_bit_field): Likewise.
7290 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7291 instead of TRULY_NOOP_TRUNCATION.
7292 * function.c (assign_parm_setup_block): Likewise.
7293 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7294 * rtlhooks.c: Include target.h.
7295 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7296 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7297 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7298 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7299 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7300 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7301 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7302 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7303 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7304 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7305 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7306 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7307 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7308 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7309 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7310 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7311 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7312 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7313 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7314 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7315 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7316 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7317 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7318 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7319 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7320 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7321 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7322 rather than TRULY_NOOP_TRUNCATION in comments.
7323 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7324 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7325 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7326 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7327 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7328 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7329 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7330 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7331 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7332 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7333 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7334 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7335 rather than TRULY_NOOP_TRUNCATION in comments.
7336 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7337 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7338 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7339 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7340 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7341 TRULY_NOOP_TRUNCATION condition.
7342 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7343 (TRULY_NOOP_TRUNCATION): Delete.
7344 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7345 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7346 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7347 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7348 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7349 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7350 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7351 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7352 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7353 rather than TRULY_NOOP_TRUNCATION in comments.
7354 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7355 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7356 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7357 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7358 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7359 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7361 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7364 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7365 (*cmp_ior): Likewise.
7366 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7367 (*ior_scc_scc_cmp): Likewise.
7368 (*and_scc_scc): Likewise.
7369 (*and_scc_scc_cmp): Likewise.
7371 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7372 Alan Hayard <alan.hayward@arm.com>
7373 David Sherwood <david.sherwood@arm.com>
7375 * target.def (can_change_mode_class): New hook.
7376 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7377 (hard_regno_nregs): Likewise.
7378 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7379 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7380 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7381 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7382 (LOAD_EXTEND_OP): Update accordingly.
7383 * doc/tm.texi: Regenerate.
7384 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7385 CANNOT_CHANGE_MODE_CLASS.
7386 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7387 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7388 * combine.c (simplify_set): Update accordingly.
7389 * emit-rtl.c (validate_subreg): Likewise.
7390 * recog.c (general_operand): Likewise.
7391 * regcprop.c (mode_change_ok): Likewise.
7392 * reload1.c (choose_reload_regs): Likewise.
7393 (inherit_piecemeal_p): Likewise.
7394 * rtlanal.c (simplify_subreg_regno): Likewise.
7395 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7396 instead of CANNOT_CHANGE_MODE_CLASS.
7397 (reload_cse_simplify_operands): Likewise.
7398 * reload.c (push_reload): Use targetm.can_change_mode_class
7399 instead of CANNOT_CHANGE_MODE_CLASS.
7400 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7401 REG_CANNOT_CHANGE_MODE_P.
7402 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7403 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7404 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7405 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7406 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7407 (arm_can_change_mode_class): New function.
7408 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7409 than CANNOT_CHANGE_MODE_CLASS in comments.
7410 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7411 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7412 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7413 (ix86_can_change_mode_class): ...this new function, inverting the
7414 sense of the return value.
7415 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7416 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7417 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7418 (ia64_can_change_mode_class): New function.
7419 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7420 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7421 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7422 (m32c_can_change_mode_class): ...this new function, inverting the
7423 sense of the return value.
7424 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7425 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7426 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7427 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7428 (mips_can_change_mode_class): ...this new function, inverting the
7429 sense of the return value.
7430 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7431 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7432 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7433 (msp430_can_change_mode_class): New function.
7434 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7435 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7436 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7437 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7438 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7439 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7440 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7441 (pa_cannot_change_mode_class): Replace with...
7442 (pa_can_change_mode_class): ...this new function, inverting the
7443 sense of the return value.
7444 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7445 than CANNOT_CHANGE_MODE_CLASS in comments.
7446 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7447 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7448 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7449 (pdp11_cannot_change_mode_class): Replace with...
7450 (pdp11_can_change_mode_class): ...this new function, inverting the
7451 sense of the return value.
7452 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7453 * config/powerpcspe/powerpcspe-protos.h
7454 (rs6000_cannot_change_mode_class_ptr): Delete.
7455 * config/powerpcspe/powerpcspe.c
7456 (rs6000_cannot_change_mode_class_ptr): Delete.
7457 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7458 (rs6000_option_override_internal): Assign to
7459 targetm.can_change_mode_class instead of
7460 rs6000_cannot_change_mode_class_ptr.
7461 (rs6000_cannot_change_mode_class): Replace with...
7462 (rs6000_can_change_mode_class): ...this new function, inverting the
7463 sense of the return value.
7464 (rs6000_debug_cannot_change_mode_class): Replace with...
7465 (rs6000_debug_can_change_mode_class): ...this new function.
7466 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7467 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7468 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7469 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7470 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7472 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7473 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7474 (rs6000_option_override_internal): Assign to
7475 targetm.can_change_mode_class instead of
7476 rs6000_cannot_change_mode_class_ptr.
7477 (rs6000_cannot_change_mode_class): Replace with...
7478 (rs6000_can_change_mode_class): ...this new function, inverting the
7479 sense of the return value.
7480 (rs6000_debug_cannot_change_mode_class): Replace with...
7481 (rs6000_debug_can_change_mode_class): ...this new function.
7482 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7483 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7484 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7485 (s390_can_change_mode_class): ...this new function, inverting the
7486 sense of the return value.
7487 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7488 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7489 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7490 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7491 (sh_cannot_change_mode_class): Replace with...
7492 (sh_can_change_mode_class): ...this new function, inverting the
7493 sense of the return value.
7494 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7495 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7496 (sparc_can_change_mode_class): New function.
7497 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7498 * config/spu/spu.c (spu_can_change_mode_class): New function.
7499 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7500 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7501 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7502 (visium_can_change_mode_class): New function.
7503 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7505 2017-09-15 Richard Biener <rguenther@suse.de>
7507 PR tree-optimization/82217
7508 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7509 but not undefined case.
7511 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7514 * postreload.c (reload_cse_simplify_operands): Skip
7515 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7517 2017-09-15 Richard Biener <rguenther@suse.de>
7519 PR tree-optimization/68823
7520 * graphite-scop-detection.c (build_alias_set): If we have a
7521 possible dependence check whether we can handle them by just
7522 looking at the DRs DR_ACCESS_FNs.
7523 (build_scops): If build_alias_set fails, fail the SCOP.
7525 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7527 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7528 to support float128 built-in functions that require the ISA 3.0
7530 (BU_FLOAT128_3_HW): Likewise.
7531 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7533 (FMAF128): Likewise.
7535 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7536 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7537 floating point instructions.
7538 (rs6000_invalid_builtin): Likewise.
7539 (rs6000_builtin_mask_names): Likewise.
7540 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7541 (RS6000_BTM_FLOAT128_HW): Likewise.
7542 (RS6000_BTM_COMMON): Likewise.
7543 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7545 * doc/extend.texi (RS/6000 built-in functions): Document the
7546 IEEE 128-bit floating point square root and fused multiply-add
7549 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7551 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7552 reg (r2) isn't in the set of registers defined in the prologue.
7554 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7555 Alan Hayward <alan.hayward@arm.com>
7556 David Sherwood <david.sherwood@arm.com>
7558 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7559 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7560 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7561 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7562 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7564 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7565 max_vectorization_factor.
7566 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7568 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7569 Alan Hayward <alan.hayward@arm.com>
7570 David Sherwood <david.sherwood@arm.com>
7572 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7573 (vect_worthwhile_without_simd_p): Declare.
7574 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7575 (vectorizable_reduction): Use it.
7576 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7577 (vectorizable_operation): Likewise.
7579 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7580 Alan Hayward <alan.hayward@arm.com>
7581 David Sherwood <david.sherwood@arm.com>
7583 * tree-vectorizer.h (vect_get_num_copies): New function.
7584 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7585 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7586 (vectorizable_induction): Likewise.
7587 (vectorizable_live_operation): Likewise.
7588 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7589 (vectorizable_bswap): Likewise.
7590 (vectorizable_call): Likewise.
7591 (vectorizable_conversion): Likewise.
7592 (vectorizable_assignment): Likewise.
7593 (vectorizable_shift): Likewise.
7594 (vectorizable_operation): Likewise.
7595 (vectorizable_store): Likewise.
7596 (vectorizable_load): Likewise.
7597 (vectorizable_condition): Likewise.
7598 (vectorizable_comparison): Likewise.
7599 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7601 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7602 Alan Hayward <alan.hayward@arm.com>
7603 David Sherwood <david.sherwood@arm.com>
7605 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7606 of vect_init_vector.
7608 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7609 Alan Hayward <alan.hayward@arm.com>
7610 David Sherwood <david.sherwood@arm.com>
7612 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7613 an inline wrapper that provides a location.
7614 (gimple_build_vector): Likewise.
7615 * gimple-fold.c (gimple_build_vector_from_val): New function.
7616 (gimple_build_vector): Likewise.
7617 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7618 functions to build the initial value. Always return a gimple value.
7619 (get_initial_defs_for_reduction): Likewise. Only compute
7621 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7622 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7623 (vectorizable_induction): Use gimple_build_vector rather than
7626 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7627 Alan Hayward <alan.hayward@arm.com>
7628 David Sherwood <david.sherwood@arm.com>
7630 * target.h (vec_perm_indices): New typedef.
7631 (auto_vec_perm_indices): Likewise.
7632 * optabs-query.h: Include target.h
7633 (can_vec_perm_p): Take a vec_perm_indices *.
7634 * optabs-query.c (can_vec_perm_p): Likewise.
7635 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7636 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7637 * tree-vect-generic.c (lower_vec_perm): Likewise.
7638 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7639 (vect_grouped_load_supported): Likewise.
7640 (vect_shift_permute_load_chain): Likewise.
7641 (vect_permute_store_chain): Use auto_vec_perm_indices.
7642 (vect_permute_load_chain): Likewise.
7643 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7644 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7645 Update uses of can_vec_perm_p.
7646 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7647 mode with a number of elements. Take a vec_perm_indices *.
7648 (vect_create_epilog_for_reduction): Update accordingly.
7649 Use auto_vec_perm_indices.
7650 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7651 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7652 (vect_transform_slp_perm_load): Likewise.
7653 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7654 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7655 (vect_gen_perm_mask_checked): Likewise.
7656 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7657 (vect_gen_perm_mask_checked): Likewise.
7658 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7659 (vectorizable_store): Likewise.
7660 (vectorizable_load): Likewise.
7661 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7662 (vectorizable_bswap): Likewise.
7664 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7665 Alan Hayward <alan.hayward@arm.com>
7666 David Sherwood <david.sherwood@arm.com>
7668 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7669 * tree.c (build_vector): Likewise.
7670 (build_vector_from_ctor): Update accordingly.
7671 (build_vector_from_val): Likewise.
7672 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7673 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7674 * tree-vect-generic.c (add_rshift): Likewise.
7675 (expand_vector_divmod): Likewise.
7676 (optimize_vector_constructor): Likewise.
7677 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7678 (vect_transform_slp_perm_load): Likewise.
7679 (vect_schedule_slp_instance): Likewise.
7680 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7681 (vectorizable_call): Likewise.
7682 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7683 * expmed.c (make_tree): Likewise.
7684 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7685 a vector passed to build_vector.
7686 (fold_convert_const): Likewise.
7687 (exact_inverse): Likewise.
7688 (fold_ternary_loc): Likewise.
7689 (fold_relational_const): Likewise.
7690 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7691 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7692 (const_unop): Likewise. Store the reduction accumulator in a
7693 variable rather than an array.
7694 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7695 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7696 the new vector, rather than constructing it after the input arrays.
7697 (native_interpret_vector): Use auto_vec<tree> when building
7698 a vector passed to build_vector. Add elements in order.
7699 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7700 auto_vec<tree> when building a vector passed to build_vector.
7701 (vect_create_epilog_for_reduction): Likewise.
7702 (vectorizable_induction): Likewise.
7703 (get_initial_def_for_reduction): Likewise. Fix indentation of
7705 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7707 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7708 passed to build_vector.
7710 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7711 Alan Hayward <alan.hayward@arm.com>
7712 David Sherwood <david.sherwood@arm.com>
7714 * tree-core.h (tree_base::u): Add an "nelts" field.
7715 (tree_vector): Use VECTOR_CST_NELTS as the length.
7716 * tree.c (tree_size): Likewise.
7717 (make_vector): Initialize VECTOR_CST_NELTS.
7718 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7719 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7720 TYPE_VECTOR_SUBPARTS.
7721 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7722 as the number of units, setting it from VECTOR_CST_NELTS.
7723 (const_vector_from_tree): Likewise.
7724 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7725 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7726 (fold_negate_expr_1): Likewise.
7727 (fold_convert_const): Likewise.
7728 (const_binop): Likewise. Differentiate the number of output and
7730 (const_unop): Likewise.
7731 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7732 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7733 in cases that did the opposite.
7735 2017-09-14 Richard Biener <rguenther@suse.de>
7737 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7740 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7742 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7744 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7747 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7748 if and where to split a bb, except for splitting before debug insn
7749 sequences followed by non-label real insn. Delete debug insns
7750 in between basic blocks.
7752 * combine.c (make_compound_operation_int): Formatting fixes.
7754 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7755 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7756 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7757 * config/sol2.h (LINK_EH_SPEC): Likewise.
7758 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7759 * config/s390/linux.h (LINK_SPEC): Likewise.
7760 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7761 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7762 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7763 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7764 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7765 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7766 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7767 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7769 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7771 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7773 (ENDFILE_LINUX_SPEC): Likewise.
7774 (LINK_EH_SPEC): Likewise.
7775 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7776 (LINK_OS_LINUX_SPEC32): Likewise.
7777 (LINK_OS_LINUX_SPEC64): Likewise.
7778 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7779 (LINK_OS_LINUX_SPEC): Likewise.
7781 2017-09-13 Martin Liska <mliska@suse.cz>
7784 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7785 CASE_HIGH is NULL_TREE.
7787 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7788 Alan Hayward <alan.hayward@arm.com>
7789 David Sherwood <david.sherwood@arm.com>
7791 * target.def (secondary_memory_needed): New hook.
7792 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7793 instead of SECONDARY_MEMORY_NEEDED.
7794 (secondary_memory_needed_mode): Likewise.
7795 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7796 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7797 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7798 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7799 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7800 * doc/tm.texi: Regenerate.
7801 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7802 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7803 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7804 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7805 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7806 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7807 mode argument first and change the reg_class arguments to reg_class_t.
7808 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7809 Make static. Update the call to inline_secondary_memory_needed.
7810 (ix86_register_move_cost): Update the call to
7811 inline_secondary_memory_needed.
7812 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7813 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7815 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7816 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7818 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7819 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7820 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7821 and match hook interface. Add comment from mips.h.
7822 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7823 * config/mmix/mmix.md (truncdfsf2): Refer to
7824 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7826 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7827 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7828 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7829 (pa_secondary_memory_needed): New function.
7830 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7831 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7832 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7833 (pdp11_secondary_memory_needed): Make static and match hook interface.
7834 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7835 * config/powerpcspe/powerpcspe-protos.h
7836 (rs6000_secondary_memory_needed_ptr): Delete.
7837 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7839 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7840 (rs6000_option_override_internal): Assign to
7841 targetm.secondary_memory_needed rather than
7842 rs6000_secondary_memory_needed_ptr.
7843 (rs6000_secondary_memory_needed): Match hook interface.
7844 (rs6000_debug_secondary_memory_needed): Likewise.
7845 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7846 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7847 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7848 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7849 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7850 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7852 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7853 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7854 (rs6000_option_override_internal): Assign to
7855 targetm.secondary_memory_needed rather than
7856 rs6000_secondary_memory_needed_ptr.
7857 (rs6000_secondary_memory_needed): Match hook interface.
7858 (rs6000_debug_secondary_memory_needed): Likewise.
7859 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7860 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7861 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7862 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7863 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7864 (sparc_secondary_memory_needed): New function.
7865 * lra-constraints.c (check_and_process_move): Refer to
7866 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7868 (curr_insn_transform): Likewise.
7869 (process_alt_operands): Use targetm.secondary_memory_needed
7870 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7871 (check_secondary_memory_needed_p): Likewise.
7872 (choose_split_class): Likewise.
7873 * reload.c: Unconditionally include code that was previously
7874 conditional on SECONDARY_MEMORY_NEEDED.
7875 (push_secondary_reload): Use targetm.secondary_memory_needed
7876 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7877 (push_reload): Likewise.
7878 * reload1.c: Unconditionally include code that was previously
7879 conditional on SECONDARY_MEMORY_NEEDED.
7880 (choose_reload_regs): Use targetm.secondary_memory_needed
7881 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7882 (gen_reload): Likewise.
7883 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7885 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7886 Alan Hayward <alan.hayward@arm.com>
7887 David Sherwood <david.sherwood@arm.com>
7889 * target.def (secondary_memory_needed_mode): New hook:
7890 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7891 * targhooks.h (default_secondary_memory_needed_mode): New function.
7892 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7893 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7894 * doc/tm.texi: Regenerate.
7895 * lra-constraints.c (check_and_process_move): Use
7896 targetm.secondary_memory_needed_mode instead of
7897 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7898 (curr_insn_transform): Likewise.
7899 * reload.c (get_secondary_mem): Likewise.
7900 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7901 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7903 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7904 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7905 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7906 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7907 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7909 * config/powerpcspe/powerpcspe-protos.h
7910 (rs6000_secondary_memory_needed_mode): Delete.
7911 * config/powerpcspe/powerpcspe.c
7912 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7913 (rs6000_secondary_memory_needed_mode): Make static.
7914 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7915 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7917 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7919 (rs6000_secondary_memory_needed_mode): Make static.
7920 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7921 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7922 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7923 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7924 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7926 (sparc_secondary_memory_needed_mode): New function.
7927 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7929 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7931 * config/aarch64/constraints.md (Umq): New constraint.
7932 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7934 (mov<mode>): Update condition.
7936 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7938 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7939 when bitposition is the same.
7941 2017-09-13 Richard Biener <rguenther@suse.de>
7943 * dwarf2out.c (output_die_symbol): Remove.
7944 (output_die): Do not output a DIEs symbol.
7946 2017-09-13 Richard Biener <rguenther@suse.de>
7949 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7950 default-def to avoid breaking iterator update with the weird
7951 interaction with cgraph_update_edges_for_call_stmt_node.
7953 2017-09-13 Richard Biener <rguenther@suse.de>
7955 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7956 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7957 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7958 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7960 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7962 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7963 Disable pc relative literal load irrespective of
7964 TARGET_FIX_ERR_A53_84341 for default.
7966 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7968 * config/sparc/sparc.c (output_return): Output the source location of
7969 the insn in the delay slot, if any.
7970 (output_sibcall): Likewise.
7972 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7975 * common.opt (-static-pie): New alias.
7976 (shared): Negate static-pie.
7977 (-no-pie): Update help text.
7979 (static-pie): New option.
7980 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7981 -static-pie support.
7982 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7983 (LINK_EH_SPEC): Likewise.
7984 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7985 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7986 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7987 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7988 (init_gcc_specs): Likewise.
7989 (init_spec): Likewise.
7990 (display_help): Update help message for -pie.
7991 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7994 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7996 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7997 (movdi_aarch64): Likewise.
7998 (movti_aarch64): Likewise.
8000 2017-09-12 Simon Wright <simon@pushface.org>
8003 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8004 calculation of the minor version, always output as 0.
8006 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8009 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8010 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8011 on it early, rather than manual conversion late. For
8012 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8013 instead of performing manual conversion.
8015 2017-09-12 Carl Love <cel@us.ibm.com>
8017 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8018 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8020 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8021 VMULOSW): Add definitions.
8022 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8023 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8024 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8025 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8026 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8028 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8030 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8032 (movti_aarch64): Likewise.
8033 (movdf_aarch64): Likewise.
8034 (movtf_aarch64): Likewise.
8035 (load_pairdi): Likewise.
8036 (store_pairdi): Likewise.
8037 (load_pairdf): Likewise.
8038 (store_pairdf): Likewise.
8039 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8040 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8041 (ldr_got_small_<mode>): Likewise.
8042 (ldr_got_small_28k_<mode>): Likewise.
8043 (ldr_got_tiny): Likewise.
8044 * config/aarch64/iterators.md (ldst_sz): New.
8045 (ldpstp_sz): Likewise.
8046 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8048 (thunderx_load): Split load_8 to load_16.
8049 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8051 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8052 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8053 (xgene1_store_pair): Split store_8 to store_16.
8054 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8055 (falkor_st_0_st_sd): Split store_8 to store_16.
8057 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8059 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8060 and store1/2/3/4 to store_4/8/12/16.
8061 * config/aarch64/aarch64.md: Update for rename.
8062 * config/arm/arm.md: Likewise.: Likewise.
8063 * config/arm/arm.c: Likewise.
8064 * config/arm/thumb1.md: Likewise.
8065 * config/arm/thumb2.md: Likewise.
8066 * config/arm/vfp.md: Likewise.
8067 * config/arm/arm-generic.md: Likewise.
8068 * config/arm/arm1020e.md: Likewise.
8069 * config/arm/arm1026ejs.md: Likewise.
8070 * config/arm/arm1136jfs.md: Likewise.
8071 * config/arm/arm926ejs.md: Likewise.
8072 * config/arm/cortex-a15.md: Likewise.
8073 * config/arm/cortex-a17.md: Likewise.
8074 * config/arm/cortex-a5.md: Likewise.
8075 * config/arm/cortex-a53.md: Likewise.
8076 * config/arm/cortex-a57.md: Likewise.
8077 * config/arm/cortex-a7.md: Likewise.
8078 * config/arm/cortex-a8.md: Likewise.
8079 * config/arm/cortex-a9.md: Likewise.
8080 * config/arm/cortex-m4.md: Likewise.
8081 * config/arm/cortex-m7.md: Likewise.
8082 * config/arm/cortex-r4.md: Likewise.
8083 * config/arm/exynos-m1.md: Likewise.
8084 * config/arm/fa526.md: Likewise.
8085 * config/arm/fa606te.md: Likewise.
8086 * config/arm/fa626te.md: Likewise.
8087 * config/arm/fa726te.md: Likewise.
8088 * config/arm/fmp626.md: Likewise.
8089 * config/arm/iwmmxt.md: Likewise.
8090 * config/arm/ldmstm.md: Likewise.
8091 * config/arm/marvell-pj4.md: Likewise.
8092 * config/arm/xgene1.md: Likewise.
8093 * config/aarch64/thunderx.md: Likewise.
8094 * config/aarch64/thunderx2t99.md: Likewise.
8095 * config/aarch64/falkor.md: Likewise.
8097 2017-09-12 Martin Liska <mliska@suse.cz>
8099 * attribs.c (private_lookup_attribute): New function.
8100 * attribs.h (private_lookup_attribute): Declared here.
8101 (lookup_attribute): Called from this place.
8103 2017-09-12 Richard Biener <rguenther@suse.de>
8105 PR tree-optimization/82157
8106 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8107 stmts with side-effects.
8109 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8110 Alan Hayward <alan.hayward@arm.com>
8111 David Sherwood <david.sherwood@arm.com>
8113 * target.def (hard_regno_nregs): New hook.
8114 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8115 * targhooks.h (default_hard_regno_nregs): Declare.
8116 * targhooks.c (default_hard_regno_nregs): New function.
8117 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8118 (TARGET_HARD_REGNO_NREGS): ...this hook.
8119 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8120 (CLASS_MAX_NREGS): Likewise.
8121 * doc/tm.texi: Regenerate.
8122 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8123 instead of HARD_REGNO_NREGS.
8124 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8125 HARD_REGNO_NREGS in the comment.
8126 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8127 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8128 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8129 Return an unsigned int.
8130 (TARGET_HARD_REGNO_NREGS): Redefine.
8131 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8132 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8133 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8134 (arc_hard_regno_nregs): New function.
8135 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8136 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8137 (arm_hard_regno_nregs): New function.
8138 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8139 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8140 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8141 (TARGET_HARD_REGNO_NREGS): Redefine.
8142 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8143 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8144 (HARD_REGNO_NREGS): Delete.
8145 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8146 (cr16_hard_regno_nregs): New function.
8147 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8148 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8149 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8150 (cris_hard_regno_nregs): New function.
8151 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8152 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8153 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8154 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8155 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8156 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8157 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8158 (frv_hard_regno_nregs): Make static. Take and return an
8160 (frv_class_max_nregs): Remove outdated copy of documentation.
8161 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8162 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8163 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8164 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8165 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8166 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8167 (TARGET_HARD_REGNO_NREGS): Redefine.
8168 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8169 (CLASS_MAX_NREGS): Update comment.
8170 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8171 (ia64_hard_regno_nregs): New function.
8172 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8173 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8174 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8175 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8176 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8178 (m32c_hard_regno_nregs): Likewise. Make static.
8179 (TARGET_HARD_REGNO_NREGS): Redefine.
8180 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8181 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8182 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8183 (m68k_hard_regno_nregs): New function.
8184 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8185 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8186 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8187 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8188 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8189 Take and return an unsigned int.
8190 (TARGET_HARD_REGNO_NREGS): Redefine.
8191 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8192 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8193 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8194 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8195 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8196 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8197 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8198 (msp430_hard_regno_nregs): Make static. Take and return an
8200 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8201 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8202 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8203 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8204 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8205 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8206 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8207 (TARGET_HARD_REGNO_NREGS): Redefine.
8208 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8209 (PA_HARD_REGNO_NREGS): ...this.
8210 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8211 (PA_HARD_REGNO_NREGS): ...this.
8212 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8213 (pa_hard_regno_nregs): New function.
8214 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8215 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8216 (pdp11_hard_regno_nregs): New function.
8217 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8218 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8219 (rs6000_hard_regno_nregs_hook): New function.
8220 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8221 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8222 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8223 Take and return an unsigned int. Move earlier in file.
8224 (TARGET_HARD_REGNO_NREGS): Redefine.
8225 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8226 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8227 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8228 (rl78_hard_regno_nregs): Make static. Take and return an
8230 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8231 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8232 (rs6000_hard_regno_nregs_hook): New function.
8233 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8234 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8235 (TARGET_HARD_REGNO_NREGS): Redefine.
8236 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8237 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8238 instead of HARD_REGNO_NREGS.
8239 (s390_hard_regno_nregs): New function.
8240 (s390_hard_regno_mode_ok): Add comment from s390.h.
8241 (TARGET_HARD_REGNO_NREGS): Redefine.
8242 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8243 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8244 (sh_hard_regno_nregs): New function.
8245 (sh_pass_in_reg_p): Use it.
8246 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8247 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8248 (sparc_hard_regno_nregs): New function.
8249 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8250 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8251 (spu_function_arg_advance): Use it, supplying a valid register number.
8252 (TARGET_HARD_REGNO_NREGS): Redefine.
8253 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8254 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8255 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8256 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8257 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8258 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8259 (CLASS_MAX_NREGS): Remove copy of old documentation.
8260 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8261 (visium_hard_regno_nregs): New function.
8262 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8263 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8264 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8265 xtensa_hard_regno_nregs): New function.
8266 * system.h (HARD_REGNO_NREGS): Poison.
8268 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8270 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8271 hard_regno_nregs instead of HARD_REGNO_NREGS.
8272 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8273 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8274 (c6x_expand_epilogue): Likewise.
8275 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8276 (frv_read_iacc_argument): Likewise.
8277 * config/sh/sh.c: Include regs.h.
8278 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8279 (regs_used): Likewise.
8280 (output_stack_adjust): Likewise.
8281 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8282 * expmed.c: Include regs.h.
8283 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8284 * ree.c: Include regs.h.
8285 (combine_reaching_defs): Use hard_regno_nregs instead of
8287 (add_removable_extension): Likewise.
8289 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8291 * regs.h (hard_regno_nregs): Turn into a function.
8292 (end_hard_regno): Update accordingly.
8293 * caller-save.c (setup_save_areas): Likewise.
8294 (save_call_clobbered_regs): Likewise.
8295 (replace_reg_with_saved_mem): Likewise.
8296 (insert_restore): Likewise.
8297 (insert_save): Likewise.
8298 * combine.c (can_change_dest_mode): Likewise.
8299 (move_deaths): Likewise.
8300 (distribute_notes): Likewise.
8301 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8302 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8303 (rs6000_split_multireg_move): Likewise.
8304 (rs6000_register_move_cost): Likewise.
8305 (rs6000_memory_move_cost): Likewise.
8306 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8307 (rs6000_split_multireg_move): Likewise.
8308 (rs6000_register_move_cost): Likewise.
8309 (rs6000_memory_move_cost): Likewise.
8310 * cselib.c (cselib_reset_table): Likewise.
8311 (cselib_lookup_1): Likewise.
8312 * emit-rtl.c (set_mode_and_regno): Likewise.
8313 * function.c (aggregate_value_p): Likewise.
8314 * ira-color.c (setup_profitable_hard_regs): Likewise.
8315 (check_hard_reg_p): Likewise.
8316 (calculate_saved_nregs): Likewise.
8317 (assign_hard_reg): Likewise.
8318 (improve_allocation): Likewise.
8319 (calculate_spill_cost): Likewise.
8320 * ira-emit.c (modify_move_list): Likewise.
8321 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8322 (ira_hard_reg_in_set_p): Likewise.
8323 * ira.c (setup_reg_mode_hard_regset): Likewise.
8324 (clarify_prohibited_class_mode_regs): Likewise.
8325 (check_allocation): Likewise.
8326 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8327 (lra_setup_reg_renumber): Likewise.
8328 (setup_try_hard_regno_pseudos): Likewise.
8329 (spill_for): Likewise.
8330 (assign_hard_regno): Likewise.
8331 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8332 * lra-constraints.c (in_class_p): Likewise.
8333 (lra_constraint_offset): Likewise.
8334 (simplify_operand_subreg): Likewise.
8335 (lra_constraints): Likewise.
8336 (split_reg): Likewise.
8337 (split_if_necessary): Likewise.
8338 (invariant_p): Likewise.
8339 (inherit_in_ebb): Likewise.
8340 * lra-lives.c (process_bb_lives): Likewise.
8341 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8342 (get_hard_regs): Likewise.
8343 (do_remat): Likewise.
8344 * lra-spills.c (assign_spill_hard_regs): Likewise.
8345 * mode-switching.c (create_pre_exit): Likewise.
8346 * postreload.c (reload_combine_recognize_pattern): Likewise.
8347 * recog.c (peep2_find_free_register): Likewise.
8348 * regcprop.c (kill_value_regno): Likewise.
8349 (set_value_regno): Likewise.
8350 (copy_value): Likewise.
8351 (maybe_mode_change): Likewise.
8352 (find_oldest_value_reg): Likewise.
8353 (copyprop_hardreg_forward_1): Likewise.
8354 * regrename.c (check_new_reg_p): Likewise.
8355 (regrename_do_replace): Likewise.
8356 * reload.c (push_reload): Likewise.
8357 (combine_reloads): Likewise.
8358 (find_dummy_reload): Likewise.
8359 (operands_match_p): Likewise.
8360 (find_reloads): Likewise.
8361 (find_equiv_reg): Likewise.
8362 (reload_adjust_reg_for_mode): Likewise.
8363 * reload1.c (count_pseudo): Likewise.
8364 (count_spilled_pseudo): Likewise.
8365 (find_reg): Likewise.
8366 (clear_reload_reg_in_use): Likewise.
8367 (free_for_value_p): Likewise.
8368 (allocate_reload_reg): Likewise.
8369 (choose_reload_regs): Likewise.
8370 (reload_adjust_reg_for_temp): Likewise.
8371 (emit_reload_insns): Likewise.
8372 (delete_output_reload): Likewise.
8373 * rtlanal.c (subreg_get_info): Likewise.
8374 * sched-deps.c (sched_analyze_reg): Likewise.
8375 * sel-sched.c (init_regs_for_mode): Likewise.
8376 (mark_unavailable_hard_regs): Likewise.
8377 (choose_best_reg_1): Likewise.
8378 (verify_target_availability): Likewise.
8379 * valtrack.c (dead_debug_insert_temp): Likewise.
8380 * var-tracking.c (track_loc_p): Likewise.
8381 (emit_note_insn_var_location): Likewise.
8382 * varasm.c (make_decl_rtl): Likewise.
8383 * reginfo.c (choose_hard_reg_mode): Likewise.
8384 (init_reg_modes_target): Refer directly to
8385 this_target_regs->x_hard_regno_nregs.
8387 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8389 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8390 instead of hard_regno_nregs.
8392 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8394 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8395 end_hard_regno instead of hard_regno_nregs.
8396 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8397 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8398 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8399 * ira-color.c (improve_allocation): Likewise.
8400 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8401 * lra-lives.c (mark_regno_live): Likewise.
8402 (mark_regno_dead): Likewise.
8403 * lra-remat.c (operand_to_remat): Likewise.
8404 * lra.c (collect_non_operand_hard_regs): Likewise.
8405 * postreload.c (reload_combine_note_store): Likewise.
8406 (move2add_valid_value_p): Likewise.
8407 * reload.c (regno_clobbered_p): Likewise.
8409 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8411 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8413 * config/v850/v850.c (v850_reorg): Likewise.
8414 * reload.c (refers_to_regno_for_reload_p): Likewise.
8415 (find_equiv_reg): Likewise.
8416 * reload1.c (reload_reg_reaches_end_p): Likewise.
8418 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8420 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8422 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8423 * config/arm/arm.c (output_move_neon): Likewise.
8424 (arm_attr_length_move_neon): Likewise.
8425 (neon_split_vcombine): Likewise.
8426 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8427 (c6x_mark_reg_written): Likewise.
8428 (c6x_dwarf_register_span): Likewise.
8429 * config/i386/i386.c (ix86_save_reg): Likewise.
8430 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8431 (rws_access_reg): Likewise.
8432 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8433 * mode-switching.c (create_pre_exit): Likewise.
8434 * ree.c (combine_reaching_defs): Likewise.
8435 (add_removable_extension): Likewise.
8436 * regcprop.c (find_oldest_value_reg): Likewise.
8437 (copyprop_hardreg_forward_1): Likewise.
8438 * reload.c (reload_inner_reg_of_subreg): Likewise.
8439 (push_reload): Likewise.
8440 (combine_reloads): Likewise.
8441 (find_dummy_reload): Likewise.
8442 (reload_adjust_reg_for_mode): Likewise.
8443 * reload1.c (find_reload_regs): Likewise.
8444 (forget_old_reloads_1): Likewise.
8445 (reload_reg_free_for_value_p): Likewise.
8446 (reload_adjust_reg_for_temp): Likewise.
8447 (emit_reload_insns): Likewise.
8448 (delete_output_reload): Likewise.
8449 * sel-sched.c (choose_best_reg_1): Likewise.
8450 (choose_best_pseudo_reg): Likewise.
8452 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8453 Alan Hayward <alan.hayward@arm.com>
8454 David Sherwood <david.sherwood@arm.com>
8456 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8457 * target.def (slow_unaligned_access): New hook.
8458 * targhooks.h (default_slow_unaligned_access): Declare.
8459 * targhooks.c (default_slow_unaligned_access): New function.
8460 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8461 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8462 * doc/tm.texi: Regenerate.
8463 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8464 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8465 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8467 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8468 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8470 (rs6000_slow_unaligned_access): New function.
8471 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8472 (expand_block_compare): Likewise.
8473 (expand_strn_compare): Likewise.
8474 (rs6000_rtx_costs): Likewise.
8475 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8476 (riscv_slow_unaligned_access): Likewise.
8477 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8478 (riscv_slow_unaligned_access_p): ...this and make static.
8479 (riscv_option_override): Update accordingly.
8480 (riscv_slow_unaligned_access): New function.
8481 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8482 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8483 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8484 (rs6000_slow_unaligned_access): New function.
8485 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8486 (rs6000_rtx_costs): Likewise.
8487 * config/rs6000/rs6000-string.c (expand_block_compare)
8488 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8489 of SLOW_UNALIGNED_ACCESS.
8490 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8491 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8492 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8493 of SLOW_UNALIGNED_ACCESS.
8494 * expmed.c (simple_mem_bitfield_p): Likewise.
8495 * expr.c (alignment_for_piecewise_move): Likewise.
8496 (emit_group_load_1): Likewise.
8497 (emit_group_store): Likewise.
8498 (copy_blkmode_from_reg): Likewise.
8499 (emit_push_insn): Likewise.
8500 (expand_assignment): Likewise.
8501 (store_field): Likewise.
8502 (expand_expr_real_1): Likewise.
8503 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8504 * lra-constraints.c (simplify_operand_subreg): Likewise.
8505 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8506 * gimple-ssa-store-merging.c: Likewise in block comment at start
8508 * tree-ssa-strlen.c: Include target.h.
8509 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8510 of SLOW_UNALIGNED_ACCESS.
8511 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8513 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8515 PR rtl-optimization/82185
8516 * expmed.c (emit_store_flag_int): Only test tem if it has been
8519 2017-09-12 Richard Biener <rguenther@suse.de>
8522 * match.pd ((FTYPE) N CMP CST): Fix typo.
8524 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8526 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8528 (mips_near_type_p): Add 'short_call' attribute as a synonym
8530 * doc/extend.texi (short_call): Document new function attribute.
8532 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8535 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8536 assertion check that in the condition.
8537 (get_atomic_generic_size): Likewise. Before testing if parameter
8538 has pointer type, if it has array type, call for C++
8539 default_conversion to perform array-to-pointer conversion.
8541 2017-09-12 Richard Biener <rguenther@suse.de>
8543 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8544 for operations we cannot scalarize.
8546 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8548 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8549 vectors heap vectors. Clean up comments.
8550 Make visited_bbs a reference.
8551 (profitable_jump_thread_path): Make GC
8552 vectors heap vectors. Clean up comments.
8554 (convert_and_register_jump_thread_path): Make GC vectors heap
8556 (check_subpath_and_update_thread_path): Same. Clean up comments.
8557 Make visited_bbs a reference.
8558 (handle_phi): Abstract common code to to
8559 register_jump_thread_path_if_profitable.
8560 Rename VAR_BB to DEF_BB.
8562 Make GC vectors heap vectors.
8563 Make visited_bbs a reference.
8564 (handle_assignment): Same.
8565 (register_jump_thread_path_if_profitable): New.
8566 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8568 Make GC vectors heap vectors. Clean up comments.
8569 Make visited_bbs a reference.
8570 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8571 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8574 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8577 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8578 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8580 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8582 Revert r251800 and r251799.
8584 2017-09-11 Martin Jambor <mjambor@suse.cz>
8587 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8588 arguments in advance.
8589 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8590 use it to find predecessor edges.
8591 (naive_outof_ssa): Collect vector of predecessors.
8593 2017-09-08 Jason Merrill <jason@redhat.com>
8595 PR c++/70029 - ICE with ref-qualifier and -flto
8596 * langhooks.h (struct lang_hooks_for_types): Add
8597 copy_lang_qualifiers.
8598 * attribs.c (build_type_attribute_qual_variant): Use it.
8599 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8601 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8602 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8604 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8607 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8608 (*mulsi3_sp64): New instruction.
8609 (mulsi3): New expander.
8611 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8613 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8615 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8617 * sancov.c: Include memmodel.h.
8619 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8622 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8625 2017-09-07 Carl Love <cel@us.ibm.com>
8627 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8628 the sldi instruction.
8630 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8632 * sancov.c: Include tm_p.h.
8634 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8637 * output.h (switch_to_other_text_partition): New declaration.
8638 * varasm.c (switch_to_other_text_partition): New function.
8639 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8640 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8641 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8642 to the other text partition before emitting LCL label and switch back
8643 after emitting the word after it.
8645 2017-09-07 Richard Biener <rguenther@suse.de>
8647 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8648 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8649 critical edges here, after loop init.
8650 (pass_data_pre): Remove PROP_no_crit_edges flags.
8651 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8652 for valueization of call args to avoid leaking VN_TOP.
8653 (visit_use): Assert we do not visit default defs.
8654 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8655 Use error_mark_node to more easily detect leaking VN_TOP.
8656 All default-defs are varying, not VN_TOP. Mark them visited.
8657 (run_scc_vn): Make code match comment.
8659 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8661 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8662 OPTION_MASK_FLOAT128_KEYWORD.
8663 (POWERPC_MASKS): Likewise.
8664 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8665 support for the -mfloat128-type option, and make -mfloat128
8666 default on PowerPC Linux systems. Define or undefine
8667 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8668 Define __float128 to be __ieee128 if IEEE 128-bit support is
8669 enabled, or undefine it.
8670 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8671 Delete defining __FLOAT128_TYPE__.
8672 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8673 -mfloat128-type option and make -mfloat128 default on PowerPC
8675 (TARGET_FLOAT128_TYPE): Likewise.
8676 (-mfloat128-type): Likewise.
8677 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8678 Delete the -mfloat128-type option and make -mfloat128 default on
8679 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8680 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8681 128-bit floating point is enabled. Change tests from using
8682 -mfloat128-type to -mfloat128.
8683 (rs6000_mangle_type): Use the correct mangling for the __float128
8684 type even if normal long double is restricted to 64-bits.
8685 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8687 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8688 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8689 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8691 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8692 documentation for -mfloat128.
8694 2017-09-06 Olivier Hainque <hainque@adacore.com>
8696 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8698 2017-09-06 Wish Wu <wishwu007@gmail.com>
8699 Jakub Jelinek <jakub@redhat.com>
8701 * asan.c (initialize_sanitizer_builtins): Add
8702 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8703 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8704 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8705 BT_FN_VOID_UINT64_PTR variables.
8706 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8707 (BT_FN_VOID_UINT16_UINT16): Likewise.
8708 (BT_FN_VOID_UINT32_UINT32): Likewise.
8709 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8710 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8711 (BT_FN_VOID_UINT64_PTR): Likewise.
8712 * common.opt (flag_sanitize_coverage): New variable.
8713 (fsanitize-coverage=trace-pc): Remove.
8714 (fsanitize-coverage=): Add.
8715 * flag-types.h (enum sanitize_coverage_code): New enum.
8716 * fold-const.c (fold_range_test): Disable non-short-circuit
8717 optimization if flag_sanitize_coverage.
8718 (fold_truth_andor): Likewise.
8719 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8720 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8721 (coverage_sanitizer_opts): New array.
8722 (get_closest_sanitizer_option): Add OPTS argument, handle also
8723 OPT_fsanitize_coverage_.
8724 (parse_sanitizer_options): Adjusted to also handle
8725 OPT_fsanitize_coverage_.
8726 (common_handle_option): Add OPT_fsanitize_coverage_.
8727 * sancov.c (instrument_comparison, instrument_switch): New function.
8728 (sancov_pass): Add trace-cmp support.
8729 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8730 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8731 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8732 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8733 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8734 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8735 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8736 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8737 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8738 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8740 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8742 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8743 error. Only quit immediately if parsing is complete.
8744 (BEGIN): Initialize fatal_err and parse_done.
8745 (begin fpu, end fpu): Check number of arguments.
8746 (begin arch, end arch): Likewise.
8747 (begin cpu, end cpu): Likewise.
8748 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8749 (optalias): Likewise.
8751 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8753 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8754 * config/arm/arm-isa.h: Delete. Move definitions to ...
8755 * arm-cpus.in: ... here. Use new feature and fgroup values.
8756 * config/arm/arm.c (arm_option_override): Use lower case for feature
8758 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8759 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8760 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8762 (print_isa_bits_for): New function.
8763 (gen_isa): New function.
8764 (gen_comm_data): Use print_isa_bits_for.
8765 (define feature): New keyword.
8766 (define fgroup): New keyword.
8767 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8768 (arm-isa.h): Add rule to generate file.
8769 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8770 case for feature bit names.
8772 2017-09-06 Richard Biener <rguenther@suse.de>
8774 * tree-ssa-pre.c (NECESSARY): Remove.
8775 (create_expression_by_pieces): Do not touch pass-local flags.
8776 (insert_into_preds_of_block): Likewise.
8777 (do_pre_regular_insertion): Likewise.
8778 (eliminate_insert): Likewise.
8779 (eliminate_dom_walker::before_dom_children): Likewise.
8780 (fini_eliminate): Do not look at inserted_exprs.
8781 (mark_operand_necessary): Remove.
8782 (remove_dead_inserted_code): Replace with simple work-list
8783 algorithm based on inserted_exprs and SSA uses.
8784 (pass_pre::execute): Re-order fini_eliminate and
8785 remove_dead_inserted_code.
8787 2017-09-06 Olivier Hainque <hainque@adacore.com>
8789 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8790 for VxWorks 7. Adjust surrounding comments.
8792 2017-09-06 Richard Biener <rguenther@suse.de>
8794 * gimple-ssa-strength-reduction.c
8795 (find_candidates_dom_walker::before_dom_children): Also allow
8798 2017-09-06 Richard Biener <rguenther@suse.de>
8800 PR tree-optimization/82108
8801 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8802 for gap in the non-permutation SLP case.
8804 2017-09-06 Martin Jambor <mjambor@suse.cz>
8806 PR tree-optimization/82078
8807 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8808 add_access_to_work_queue...
8809 (build_accesses_from_assign): ...here.
8810 (propagate_all_subaccesses): Make sure racc is the group
8811 representative, if there is one.
8813 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8816 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8819 2017-09-06 Richard Biener <rguenther@suse.de>
8821 * gimple-ssa-strength-reduction.c
8822 (find_candidates_dom_walker::before_doom_children): Use a
8823 type and not a mode check.
8825 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8828 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8830 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8832 2017-09-05 Jeff Law <law@redhat.com>
8834 PR tree-optimization/64910
8835 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8836 cases where we have 3 or more operands.
8838 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8841 * omp-low.c (lower_omp_for): Recompute tree invariant if
8842 gimple_omp_for_initial/final is ADDR_EXPR.
8845 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8846 into gimple val before gimplification fo the COND_EXPR.
8848 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8850 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8851 REGION_COPY argument.
8852 (thread_through_all_blocks): Remove unused argument to
8853 duplicate_thread_path.
8855 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8856 Alan Hayward <alan.hayward@arm.com>
8857 David Sherwood <david.sherwood@arm.com>
8859 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8860 Take a scalar_mode rather than a machine_mode.
8861 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8862 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8863 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8864 (aarch64_gen_adjusted_ldpstp): Likewise.
8865 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8867 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8868 Alan Hayward <alan.hayward@arm.com>
8869 David Sherwood <david.sherwood@arm.com>
8871 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8872 Take a scalar_int_mode instead of a machine_mode.
8873 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8874 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8875 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8876 (aarch64_simd_attr_length_rglist): Delete.
8877 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8878 a scalar_int_mode instead of a machine_mode.
8879 (aarch64_add_offset): Likewise.
8880 (aarch64_internal_mov_immediate): Likewise
8881 (aarch64_add_constant_internal): Likewise.
8882 (aarch64_add_constant): Likewise.
8883 (aarch64_movw_imm): Likewise.
8884 (aarch64_rtx_arith_op_extract_p): Likewise.
8885 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8886 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8887 Remove assert that the mode isn't a vector.
8888 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8889 (aarch64_expand_mov_immediate): Update calls after above changes.
8890 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8891 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8892 (aarch64_move_imm): Likewise.
8893 (aarch64_can_const_movi_rtx_p): Likewise.
8894 (aarch64_strip_extend): Likewise.
8895 (aarch64_extr_rtx_p): Likewise.
8896 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8897 a CONST_INT when the mode parameter is VOIDmode.
8898 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8900 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8902 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8903 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8904 (bitwise_type_for_mode): Update accordingly.
8906 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8908 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8909 * stor-layout.c (mode_for_size_tree): Likewise.
8910 (mode_for_array): Update accordingly.
8911 (layout_decl): Likewise.
8912 (compute_record_mode): Likewise. Only set the mode once.
8914 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8916 * target.def (get_mask_mode): Change return type to opt_mode.
8918 * doc/tm.texi: Regenerate.
8919 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8920 * targhooks.c (default_get_mask_mode): Likewise.
8921 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8922 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8923 targetm.get_mask_mode.
8924 * tree.c (build_truth_vector_type): Likewise.
8926 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8928 * machmode.h (mode_for_vector): Return an opt_mode.
8929 * stor-layout.c (mode_for_vector): Likewise.
8930 (mode_for_int_vector): Update accordingly.
8931 (layout_type): Likewise.
8932 * config/i386/i386.c (emit_memmov): Likewise.
8933 (ix86_expand_set_or_movmem): Likewise.
8934 (ix86_expand_vector_init): Likewise.
8935 (ix86_get_mask_mode): Likewise.
8936 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8938 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8939 * expmed.c (extract_bit_field_1): Likewise.
8940 * expr.c (expand_expr_real_2): Likewise.
8941 * optabs-query.c (can_vec_perm_p): Likewise.
8942 (can_vec_mask_load_store_p): Likewise.
8943 * optabs.c (expand_vec_perm): Likewise.
8944 * targhooks.c (default_get_mask_mode): Likewise.
8945 * tree-vect-stmts.c (vectorizable_store): Likewise.
8946 (vectorizable_load): Likewise.
8947 (get_vectype_for_scalar_type_and_size): Likewise.
8949 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8951 * machmode.h (mode_for_int_vector): New function.
8952 * stor-layout.c (mode_for_int_vector): Likewise.
8953 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8954 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8955 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8956 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8957 (s390_expand_vcond): Likewise.
8959 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8961 * machmode.h (opt_machine_mode): New type.
8962 (opt_mode<T>): Allow construction from anything that can be
8964 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8965 (mode_for_size): Return an opt_machine_mode.
8966 * stor-layout.c (mode_for_size): Likewise.
8967 (mode_for_size_tree): Update call accordingly.
8968 (bitwise_mode_for_mode): Likewise.
8969 (make_fract_type): Likewise.
8970 (make_accum_type): Likewise.
8971 * caller-save.c (replace_reg_with_saved_mem): Update call
8973 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8974 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8975 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8976 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8977 * expmed.c (extract_bit_field_1): Likewise.
8978 * reload.c (get_secondary_mem): Likewise.
8979 * varasm.c (assemble_integer): Likewise.
8980 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8983 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8985 * machmode.h (decimal_float_mode_for_size): New function.
8986 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8987 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8988 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8989 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8990 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8991 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8993 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8995 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8996 (get_builtin_sync_mode): Likewise.
8997 (expand_ifn_atomic_compare_exchange): Likewise.
8998 (expand_builtin_atomic_clear): Likewise.
8999 (expand_builtin_atomic_test_and_set): Likewise.
9000 (fold_builtin_atomic_always_lock_free): Likewise.
9001 * calls.c (compute_argument_addresses): Likewise.
9002 (emit_library_call_value_1): Likewise.
9003 (store_one_arg): Likewise.
9004 * combine.c (combine_instructions): Likewise.
9005 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9006 * config/arm/arm.c (arm_function_value): Likewise.
9007 (aapcs_allocate_return_reg): Likewise.
9008 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9009 * config/i386/i386.c (construct_container): Likewise.
9010 (ix86_gimplify_va_arg): Likewise.
9011 (ix86_expand_sse_cmp): Likewise.
9012 (emit_memmov): Likewise.
9013 (emit_memset): Likewise.
9014 (expand_small_movmem_or_setmem): Likewise.
9015 (ix86_expand_pextr): Likewise.
9016 (ix86_expand_pinsr): Likewise.
9017 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9018 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9020 * config/mips/mips.c (mips_function_value_1) Likewise.
9021 (mips_block_move_straight): Likewise.
9022 (mips_expand_ins_as_unaligned_store): Likewise.
9023 * config/powerpcspe/powerpcspe.c
9024 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9025 (rs6000_darwin64_record_arg_flush): Likewise.
9026 * config/rs6000/rs6000.c
9027 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9028 (rs6000_darwin64_record_arg_flush): Likewise.
9029 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9030 (sparc_function_value_1): Likewise.
9031 * config/spu/spu.c (adjust_operand): Likewise.
9032 (spu_emit_branch_or_set): Likewise.
9033 (arith_immediate_p): Likewise.
9034 * emit-rtl.c (gen_lowpart_common): Likewise.
9035 * expr.c (expand_expr_real_1): Likewise.
9036 * function.c (assign_parm_setup_block): Likewise.
9037 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9038 * reload1.c (alter_reg): Likewise.
9039 * stor-layout.c (mode_for_vector): Likewise.
9040 (layout_type): Likewise.
9042 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9044 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9045 (spu_convert_move): Likewise.
9046 * lower-subreg.c (resolve_simple_move): Likewise.
9048 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9051 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9052 define_insn to a define_expand.
9053 (altivec_vsum2sws_direct): New define_insn.
9054 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9056 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9058 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9061 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9066 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9067 realign stack if stack alignment needed is less than incoming
9070 2017-09-05 Marek Polacek <polacek@redhat.com>
9073 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9076 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9078 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9080 2017-09-05 Richard Biener <rguenther@suse.de>
9082 PR tree-optimization/82084
9083 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9085 2017-09-05 Richard Biener <rguenther@suse.de>
9087 PR tree-optimization/82102
9088 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9090 2017-09-05 Martin Liska <mliska@suse.cz>
9092 PR tree-optimization/82032
9093 * tree-cfg.c (generate_range_test): New function.
9094 * tree-cfg.h (generate_range_test): Declared here.
9095 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9096 (cleanup_control_expr_graph): Use it.
9097 * tree-switch-conversion.c (try_switch_expansion): Remove
9099 (emit_case_nodes): Use generate_range_test.
9101 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9104 * config/i386/i386.md (*<btsc><mode>_mask): Add
9105 TARGET_USE_BT to insn constraint.
9106 (*btr<mode>_mask): Ditto.
9108 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9110 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9111 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9113 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9116 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9117 TARGET_NEON and TARGET_IWMMXT.
9118 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9119 TARGET_NEON and TARGET_IWMMXT.
9120 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9122 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9124 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9125 (ix86_rewrite_tls_address): Ditto.
9126 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9127 (ix86_rewrite_tls_address_1): Ditto.
9128 (ix86_rewrite_tls_address): Ditto.
9129 * config/i386/predicates.md (tls_address_pattern): New predicate.
9130 * config/i386/i386.md (TLS address splitter): New splitter.
9132 2017-09-04 Richard Biener <rguenther@suse.de>
9134 PR tree-optimization/82084
9135 * fold-const.h (can_native_encode_string_p): Declare.
9136 * fold-const.c (can_native_encode_string_p): Factor out from ...
9137 (native_encode_string): ... here.
9138 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9139 vectorizing stores from constants we later cannot handle.
9141 2017-09-04 Marek Polacek <polacek@redhat.com>
9144 * doc/invoke.texi: Update -Wtautological-compare documentation.
9146 2017-09-04 Jeff Law <law@redhat.com>
9148 PR tree-optimization/64910
9149 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9150 swap the first and last operand if the last is a constant.
9152 2017-09-04 Marek Polacek <polacek@redhat.com>
9155 * convert.c (do_narrow): When sanitizing signed integer overflows,
9156 bail out for signed types.
9157 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9159 2017-09-04 Richard Biener <rguenther@suse.de>
9161 PR tree-optimization/82060
9162 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9163 Move devirtualization after stmt folding and before EH/AB/noreturn
9164 cleanup to get the stmt refs canonicalized. Use a bool instead
9165 of gimple_modified_p since that doesn't work for NOPs. Schedule
9166 NOPs generated by folding for removal.
9168 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9169 Alan Hayward <alan.hayward@arm.com>
9170 David Sherwood <david.sherwood@arm.com>
9172 * coretypes.h (pad_direction): New enum.
9173 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9174 (FUNCTION_ARG_PADDING): Likewise.
9175 * target.def (function_arg_padding): New hook.
9176 * targhooks.h (default_function_arg_padding): Declare.
9177 * targhooks.c (default_function_arg_padding): New function.
9178 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9179 (TARGET_FUNCTION_ARG_PADDING): ...this.
9180 * doc/tm.texi: Regenerate.
9181 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9182 instead of direction.
9183 (compute_argument_addresses): Likewise.
9184 (load_register_parameters): Likewise.
9185 (emit_library_call_value_1): Likewise.
9186 (store_one_arg): Use targetm.calls.function_arg_padding instead
9187 of FUNCTION_ARG_PADDING.
9188 (must_pass_in_stack_var_size_or_pad): Likewise.
9189 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9190 (emit_group_store): Likewise.
9191 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9192 instead of FUNCTION_ARG_PADDING.
9193 (emit_push_insn): Likewise, and propagate enum change throughout
9195 * function.h (direction): Delete.
9196 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9198 * function.c (assign_parm_find_stack_rtl): Likewise.
9199 (assign_parm_setup_block_p): Likewise.
9200 (assign_parm_setup_block): Likewise.
9201 (gimplify_parameters): Likewise.
9202 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9203 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9205 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9206 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9207 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9208 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9209 (aarch64_function_arg_padding): ...this new function.
9210 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9211 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9212 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9213 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9214 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9215 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9216 (arm_pad_arg_upward): Replace with...
9217 (arm_function_arg_padding): ...this new function.
9218 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9220 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9221 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9222 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9223 (ia64_hpux_function_arg_padding): Replace with...
9224 (ia64_function_arg_padding): ...this new function. Use pad_direction
9225 instead of direction. Check for TARGET_HPUX.
9226 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9227 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9228 (iq2000_function_arg_padding): New function.
9229 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9230 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9231 (mips_function_arg_padding): ...this new function.
9232 (mips_pad_reg_upward): Update accordingly.
9233 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9234 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9235 targetm.calls.function_arg_padding.
9236 (FUNCTION_ARG_PADDING): Delete.
9237 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9238 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9239 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9240 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9241 (nios2_block_reg_padding): Return pad_direction instead of direction.
9242 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9243 instead of direction.
9244 (nios2_function_arg_padding): Likewise. Make static.
9245 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9246 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9247 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9248 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9249 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9250 (pa_function_arg_padding): Make static. Return pad_direction instead
9252 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9253 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9254 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9255 instead of direction. Use targetm.calls.function_arg_padding.
9256 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9257 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9258 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9259 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9260 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9262 (function_arg_padding): Rename to...
9263 (rs6000_function_arg_padding): ...this. Make static. Return
9264 pad_direction instead of direction.
9265 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9266 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9267 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9268 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9269 instead of direction. Use targetm.calls.function_arg_padding.
9270 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9271 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9272 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9273 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9274 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9275 (function_arg_padding): Rename to...
9276 (rs6000_function_arg_padding): ...this. Make static. Return
9277 pad_direction instead of direction.
9278 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9279 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9280 * config/s390/s390.c (s390_function_arg_padding): New function.
9281 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9282 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9283 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9284 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9285 (function_arg_padding): Rename to...
9286 (sparc_function_arg_padding): ...this. Make static. Return
9287 pad_direction instead of direction.
9288 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9289 * config/spu/spu.c (spu_function_arg_padding): New function.
9290 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9291 * system.h (FUNCTION_ARG_PADDING): Poison.
9293 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9294 Alan Hayward <alan.hayward@arm.com>
9295 David Sherwood <david.sherwood@arm.com>
9297 * target.def (modes_tieable_p): New hook.
9298 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9299 (TARGET_MODES_TIEABLE_P): ...this.
9300 * doc/tm.texi.in: Regenerate.
9301 * hooks.h (hook_bool_mode_mode_true): Declare.
9302 * hooks.c (hook_bool_mode_mode_true): New function.
9303 * combine.c (subst): Use targetm.modes_tieable_p instead of
9305 * dse.c (find_shift_sequence): Likewise.
9306 * expmed.c (extract_low_bits): Likewise.
9307 * lower-subreg.c: Include target.h.
9308 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9310 * rtlanal.c (rtx_cost): Likewise.
9311 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9312 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9313 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9314 (TARGET_MODES_TIEABLE_P): Redefine.
9315 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9316 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9317 (TARGET_MODES_TIEABLE_P): Redefine.
9318 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9319 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9320 (arc_modes_tieable_p): New function.
9321 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9322 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9323 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9324 (arm_modes_tieable_p): Make static.
9325 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9326 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9327 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9328 (TARGET_MODES_TIEABLE_P): Redefine.
9329 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9330 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9331 (TARGET_MODES_TIEABLE_P): Redefine.
9332 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9333 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9334 (cr16_modes_tieable_p): New function.
9335 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9336 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9337 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9338 (TRULY_NOOP_TRUNCATION): Update comment.
9339 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9340 (TRULY_NOOP_TRUNCATION): Update comment.
9341 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9342 (frv_modes_tieable_p): New function.
9343 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9344 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9345 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9346 (TARGET_MODES_TIEABLE_P): Redefine.
9347 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9348 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9349 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9350 (TARGET_MODES_TIEABLE_P): Redefine.
9351 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9352 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9353 (ia64_modes_tieable_p): New function.
9354 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9355 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9356 (iq2000_modes_tieable_p): New function.
9357 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9358 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9359 (lm32_modes_tieable_p): New function.
9360 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9361 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9362 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9363 (TARGET_MODES_TIEABLE_P): Redefine.
9364 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9365 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9366 (m32r_modes_tieable_p): New function.
9367 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9368 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9369 (m68k_modes_tieable_p): New function.
9370 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9371 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9372 (mcore_modes_tieable_p): New function.
9373 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9374 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9376 (TARGET_MODES_TIEABLE_P): Redefine.
9377 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9378 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9379 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9380 (TARGET_MODES_TIEABLE_P): Redefine.
9381 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9382 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9383 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9384 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9385 (mn10300_modes_tieable_p): ...this and make static.
9386 (TARGET_MODES_TIEABLE_P): Redefine.
9387 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9388 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9389 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9390 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9391 (msp430_modes_tieable_p): Make static.
9392 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9393 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9394 (TARGET_MODES_TIEABLE_P): Redefine.
9395 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9396 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9397 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9398 (TARGET_MODES_TIEABLE_P): Redefine.
9399 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9400 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9401 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9402 (TARGET_MODES_TIEABLE_P): Redefine.
9403 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9404 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9405 (pdp11_modes_tieable_p): New function.
9406 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9407 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9408 (rs6000_modes_tieable_p): New function.
9409 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9410 * config/powerpcspe/powerpcspe.md: Update comment.
9411 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9412 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9413 (TARGET_MODES_TIEABLE_P): Redefine.
9414 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9415 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9416 (rl78_modes_tieable_p): New function.
9417 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9418 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9419 (rs6000_modes_tieable_p): New function.
9420 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9421 * config/rs6000/rs6000.md: Update comment.
9422 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9423 * config/rx/rx.c (rx_modes_tieable_p): New function.
9424 (TARGET_MODES_TIEABLE_P): Redefine.
9425 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9426 * config/s390/s390.c (s390_modes_tieable_p): New function.
9427 (TARGET_MODES_TIEABLE_P): Redefine.
9428 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9429 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9430 (sh_modes_tieable_p): New function.
9431 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9432 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9433 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9434 (sparc_modes_tieable_p): Make static.
9435 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9436 * config/spu/spu.c (spu_modes_tieable_p): New function.
9437 (TARGET_MODES_TIEABLE_P): Redefine.
9438 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9439 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9440 (TARGET_MODES_TIEABLE_P): Redefine.
9441 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9442 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9443 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9444 * config/v850/v850.c (v850_modes_tieable_p): New function.
9445 (TARGET_MODES_TIEABLE_P): Redefine.
9446 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9447 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9448 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9449 (visium_modes_tieable_p): New function.
9450 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9451 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9452 (xtensa_modes_tieable_p): New function.
9453 * system.h (MODES_TIEABLE_P): Poison.
9455 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9456 Alan Hayward <alan.hayward@arm.com>
9457 David Sherwood <david.sherwood@arm.com>
9459 * target.def (hard_regno_mode_ok): New hook.
9460 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9461 (TARGET_HARD_REGNO_MODE_OK): ...this.
9462 * doc/tm.texi.in: Regenerate.
9463 * hooks.h (hook_bool_uint_mode_true): Declare.
9464 * hooks.c (hook_bool_uint_mode_true): New function.
9465 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9467 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9468 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9469 instead of HARD_REGNO_MODE_OK.
9470 * caller-save.c: Include target.h.
9471 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9473 * combine.c (can_combine_p): Likewise.
9474 (combinable_i3pat): Likewise.
9475 (can_change_dest_mode): Likewise.
9476 * expr.c (init_expr_target): Likewise.
9477 (convert_move): Likewise.
9478 (convert_modes): Likewise.
9479 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9480 (setup_prohibited_mode_move_regs): Likewise.
9481 * ira.h (target_ira): Likewise.
9482 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9483 * lra-constraints.c (process_alt_operands): Likewise.
9484 (split_reg): Likewise.
9485 * recog.c (peep2_find_free_register): Likewise.
9486 * ree.c (combine_reaching_defs): Likewise.
9487 * regcprop.c (maybe_mode_change): Likewise.
9488 * reginfo.c (init_reg_sets_1): Likewise.
9489 (choose_hard_reg_mode): Likewise.
9490 (simplifiable_subregs): Likewise.
9491 * regrename.c (check_new_reg_p): Likewise.
9492 * reload.c (find_valid_class): Likewise.
9493 (find_valid_class_1): Likewise.
9494 (reload_inner_reg_of_subreg): Likewise.
9495 (push_reload): Likewise.
9496 (combine_reloads): Likewise.
9497 (find_dummy_reload): Likewise.
9498 (find_reloads): Likewise.
9499 * reload1.c (find_reg): Likewise.
9500 (set_reload_reg): Likewise.
9501 (allocate_reload_reg): Likewise.
9502 (choose_reload_regs): Likewise.
9503 (reload_adjust_reg_for_temp): Likewise.
9504 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9505 (simplify_subreg_regno): Likewise.
9506 * sel-sched.c (init_regs_for_mode): Likewise.
9507 * varasm.c (make_decl_rtl): Likewise.
9508 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9509 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9511 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9512 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9513 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9514 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9515 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9516 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9517 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9518 (arc_mode_class): Delete.
9519 (HARD_REGNO_MODE_OK): Delete.
9520 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9521 (arc_hard_regno_mode_ok): Rename old array to...
9522 (arc_hard_regno_mode_ok_modes): ...this.
9523 (arc_conditional_register_usage): Update accordingly.
9524 (arc_mode_class): Make static.
9525 (arc_hard_regno_mode_ok): New function.
9526 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9527 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9528 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9529 (arm_hard_regno_mode_ok): Make static.
9530 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9532 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9533 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9534 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9536 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9537 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9538 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9539 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9540 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9541 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9542 * config/bfin/predicates.md (valid_reg_operand): Use
9543 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9544 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9545 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9546 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9547 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9548 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9549 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9550 (cr16_hard_regno_mode_ok): Make static and return a bool.
9551 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9552 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9553 (cris_hard_regno_mode_ok): New function.
9554 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9555 (epiphany_mode_class): Delete.
9556 (HARD_REGNO_MODE_OK): Delete.
9557 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9558 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9559 (hard_regno_mode_ok): Rename to...
9560 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9561 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9562 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9564 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9565 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9566 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9567 (frv_hard_regno_mode_ok): Make static and return a bool.
9568 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9570 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9571 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9572 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9573 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9575 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9576 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9577 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9578 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9580 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9581 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9582 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9583 (ia64_hard_regno_mode_ok): New function.
9584 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9585 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9586 (iq2000_hard_regno_mode_ok): New function.
9587 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9588 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9589 (lm32_hard_regno_mode_ok): New function.
9590 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9591 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9592 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9593 instead of HARD_REGNO_MODE_OK.
9594 (m32c_hard_regno_ok): Rename to...
9595 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9596 (m32c_cannot_change_mode_class): Update accordingly.
9597 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9598 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9599 (m32r_mode_class): Delete.
9600 (HARD_REGNO_MODE_OK): Delete.
9601 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9602 (m32r_hard_regno_mode_ok): Rename to...
9603 (m32r_hard_regno_modes): ...this.
9604 (m32r_mode_class): Make static.
9605 (m32r_hard_regno_mode_ok): New function.
9606 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9607 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9608 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9609 (m68k_hard_regno_mode_ok): Make static.
9610 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9611 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9612 (mcore_hard_regno_mode_ok): New function.
9613 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9614 (HARD_REGNO_MODE_OK): Delete.
9615 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9617 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9618 (microblaze_hard_regno_mode_ok): New function.
9619 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9620 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9621 (mips_hard_regno_mode_ok): Delete.
9622 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9623 (mips_hard_regno_mode_ok_p): ...this and make static.
9624 (mips_hard_regno_mode_ok_p): Rename to...
9625 (mips_hard_regno_mode_ok_uncached): ...this.
9626 (mips_hard_regno_mode_ok): New function.
9627 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9628 of HARD_REGNO_MODE_OK.
9629 (mips_option_override): Update after above name changes.
9630 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9631 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9632 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9633 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9634 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9635 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9636 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9637 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9638 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9639 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9640 (msp430_hard_regno_mode_ok): Make static and return a bool.
9641 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9642 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9643 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9645 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9646 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9647 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9648 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9649 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9650 (PA_HARD_REGNO_MODE_OK): ...this
9651 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9652 (PA_HARD_REGNO_MODE_OK): ...this.
9653 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9654 (pa_hard_regno_mode_ok): New function.
9655 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9656 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9657 (pdp11_hard_regno_mode_ok): New function.
9658 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9659 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9661 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9663 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9664 (rs6000_hard_regno_mode_ok): Rename to...
9665 (rs6000_hard_regno_mode_ok_uncached): ...this.
9666 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9667 (rs6000_hard_regno_mode_ok): New function.
9668 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9669 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9670 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9671 (riscv_hard_regno_mode_ok): ...this and make static.
9672 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9673 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9674 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9675 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9676 (rl78_hard_regno_mode_ok): Make static and return bool.
9677 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9678 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9680 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9681 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9682 (rs6000_hard_regno_mode_ok): Rename to...
9683 (rs6000_hard_regno_mode_ok_uncached): ...this.
9684 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9685 (rs6000_hard_regno_mode_ok): New function.
9686 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9687 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9688 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9689 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9690 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9691 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9692 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9693 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9694 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9695 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9696 (sh_hard_regno_mode_ok): Make static.
9697 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9698 instead of HARD_REGNO_MODE_OK.
9699 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9700 (sparc_mode_class): Delete.
9701 (HARD_REGNO_MODE_OK): Delete.
9702 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9703 (hard_regno_mode_classes): Make static.
9704 (sparc_mode_class): Likewise.
9705 (sparc_hard_regno_mode_ok): New function.
9706 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9707 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9708 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9710 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9711 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9712 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9713 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9714 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9715 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9716 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9717 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9718 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9719 (visium_hard_regno_mode_ok): New function.
9720 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9721 instead of HARD_REGNO_MODE_OK.
9722 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9723 (HARD_REGNO_MODE_OK): Delete.
9724 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9725 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9726 (xtensa_option_override): Update accordingly.
9727 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9728 (xtensa_hard_regno_mode_ok): New function.
9729 * system.h (HARD_REGNO_MODE_OK): Poison.
9731 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9732 Alan Hayward <alan.hayward@arm.com>
9733 David Sherwood <david.sherwood@arm.com>
9735 * target.def (hard_regno_call_part_clobbered): New hook.
9736 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9737 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9738 * doc/tm.texi: Regenerate.
9739 * hooks.h (hook_bool_uint_mode_false): Declare.
9740 * hooks.c (hook_bool_uint_mode_false): New function.
9741 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9742 * cselib.c (cselib_process_insn): Use
9743 targetm.hard_regno_call_part_clobbered instead of
9744 HARD_REGNO_CALL_PART_CLOBBERED.
9745 * ira-conflicts.c (ira_build_conflicts): Likewise.
9746 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9747 * lra-constraints.c (need_for_call_save_p): Likewise.
9748 * lra-lives.c: Include target.h.
9749 (check_pseudos_live_through_calls): Use
9750 targetm.hard_regno_call_part_clobbered instead of
9751 HARD_REGNO_CALL_PART_CLOBBERED.
9752 * regcprop.c: Include target.h.
9753 (copyprop_hardreg_forward_1): Use
9754 targetm.hard_regno_call_part_clobbered instead of
9755 HARD_REGNO_CALL_PART_CLOBBERED.
9756 * reginfo.c (choose_hard_reg_mode): Likewise.
9757 * regrename.c (check_new_reg_p): Likewise.
9758 * reload.c (find_equiv_reg): Likewise.
9759 * reload1.c (emit_reload_insns): Likewise.
9760 * sched-deps.c (deps_analyze_insn): Likewise.
9761 * sel-sched.c (init_regs_for_mode): Likewise.
9762 (mark_unavailable_hard_regs): Likewise.
9763 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9764 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9765 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9767 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9768 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9769 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9771 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9773 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9774 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9775 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9777 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9778 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9779 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9781 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9782 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9784 * config/powerpcspe/powerpcspe.c
9785 (rs6000_hard_regno_call_part_clobbered): New function.
9786 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9787 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9788 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9790 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9791 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9792 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9794 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9795 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9796 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9798 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9799 Alan Hayward <alan.hayward@arm.com>
9800 David Sherwood <david.sherwood@arm.com>
9802 * rtl.h (subreg_memory_offset): Declare.
9803 * emit-rtl.c (subreg_memory_offset): New function.
9804 * expmed.c (store_bit_field_1): Use it.
9805 * expr.c (undefined_operand_subword_p): Likewise.
9806 * simplify-rtx.c (simplify_subreg): Likewise.
9808 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9810 PR rtl-optimization/57448
9813 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9814 using atomic_load pattern.
9815 (expand_atomic_store): Likewise.
9817 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9820 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9821 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9822 handling. Use replace_call_with_value with NULL instead of
9823 gsi_replace, unlink_stmt_vdef and release_defs.
9825 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9828 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9830 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9833 * rtl.h (emit_library_call_value_1): Declare.
9834 (emit_library_call): Replace declaration with a series of overloads.
9835 Remove the parameter count argument.
9836 (emit_library_call_value): Likewise.
9837 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9838 with an "rtx_mode_t *".
9839 (emit_library_call_value): Delete.
9840 (emit_library_call): Likewise.
9841 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9842 (asan_emit_allocas_unpoison): Likewise.
9843 * builtins.c (expand_builtin_powi): Likewise.
9844 (expand_asan_emit_allocas_unpoison): Likewise.
9845 * cfgexpand.c (expand_main_function): Likewise.
9846 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9847 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9848 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9849 * config/arm/arm.c (arm_trampoline_init): Likewise.
9850 (arm_call_tls_get_addr): Likewise.
9851 (arm_expand_divmod_libfunc): Likewise.
9852 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9853 (smulsi3_highpart): Likewise.
9854 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9855 (c6x_expand_compare): Likewise.
9856 (c6x_expand_movmem): Likewise.
9857 * config/frv/frv.c (frv_trampoline_init): Likewise.
9858 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9859 (ix86_expand_divmod_libfunc): Likewise.
9860 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9861 (ia64_expand_compare): Likewise.
9862 (ia64_profile_hook): Likewise.
9863 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9864 (nonlocal_goto): Likewise.
9865 (restore_stack_nonlocal): Likewise.
9866 * config/m32r/m32r.c (block_move_call): Likewise.
9867 (m32r_trampoline_init): Likewise.
9868 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9869 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9870 (m68k_call_m68k_read_tp): Likewise.
9871 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9872 (microblaze_expand_divide): Likewise.
9873 * config/mips/mips.h (mips_args): Likewise.
9874 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9875 (MIPS_ICACHE_SYNC): Likewise.
9876 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9877 (nios2_trampoline_init): Likewise.
9878 * config/pa/pa.c (hppa_tls_call): Likewise.
9879 (pa_trampoline_init): Likewise.
9880 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9881 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9882 (expand_strn_compare): Likewise.
9883 (rs6000_generate_compare): Likewise.
9884 (rs6000_expand_float128_convert): Likewise.
9885 (output_profile_hook): Likewise.
9886 (rs6000_trampoline_init): Likewise.
9887 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9888 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9889 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9890 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9891 (rs6000_generate_compare): Likewise.
9892 (rs6000_expand_float128_convert): Likewise.
9893 (output_profile_hook): Likewise.
9894 (rs6000_trampoline_init): Likewise.
9895 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9896 * config/sh/sh.c (sh_trampoline_init): Likewise.
9897 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9898 (sparc_emit_float_lib_cmp): Likewise.
9899 (sparc32_initialize_trampoline): Likewise.
9900 (sparc64_initialize_trampoline): Likewise.
9901 (sparc_profile_hook): Likewise.
9902 * config/spu/spu.c (ea_load_store): Likewise.
9903 * config/spu/spu.md (floatunssidf2): Likewise.
9904 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9905 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9906 * config/visium/visium.c (expand_block_move_4): Likewise.
9907 (expand_block_move_2): Likewise.
9908 (expand_block_move_1): Likewise.
9909 (expand_block_set_4): Likewise.
9910 (expand_block_set_2): Likewise.
9911 (expand_block_set_1): Likewise.
9912 (visium_trampoline_init): Likewise.
9913 (visium_profile_hook): Likewise.
9914 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9915 (xtensa_setup_frame_addresses): Likewise.
9916 (xtensa_trampoline_init): Likewise.
9917 * except.c (sjlj_emit_function_enter): Likewise.
9918 (sjlj_emit_function_exit): Likewise.
9919 * explow.c (allocate_dynamic_stack_space): Likewise.
9920 (probe_stack_range): Likewise.
9921 * expr.c (convert_mode_scalar): Likewise.
9922 * optabs.c (expand_binop): Likewise.
9923 (expand_twoval_binop_libfunc): Likewise.
9924 (expand_unop): Likewise.
9925 (prepare_cmp_insn): Likewise.
9926 (prepare_float_lib_cmp): Likewise.
9927 (expand_float): Likewise.
9928 (expand_fix): Likewise.
9929 (expand_fixed_convert): Likewise.
9930 (maybe_emit_sync_lock_test_and_set): Likewise.
9931 (expand_atomic_compare_and_swap): Likewise.
9932 (expand_mem_thread_fence): Likewise.
9933 (expand_atomic_fetch_op): Likewise.
9935 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9937 * doc/generic.texi (OpenACC): Adjust URL.
9938 * doc/invoke.texi (C Dialect Options): Ditto.
9940 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9942 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9943 predicate for operand 1. Add (m,<S>) constraint.
9944 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9945 Prevent memory operand 1 with register operand 2.
9947 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9949 PR rtl-optimization/82024
9950 * combine.c (try_combine): If the combination result is a PARALLEL,
9951 and we only need to retain the SET in there that would be placed
9952 at I2, check that we can place that at I3 instead, before doing so.
9954 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9957 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9959 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9960 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9963 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9964 Jeff Law <law@redhat.com>
9966 * varasm.c (bss_initializer_p): Do not put constants into .bss
9967 (categorize_decl_for_section): Handle bss_initializer_p returning
9968 false when DECL_INITIAL is NULL.
9970 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9973 * config/s390/s390.c (s390_can_inline_p): New function.
9975 2017-09-01 Jeff Law <law@redhat.com>
9977 PR tree-optimization/82052
9978 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9979 Always initialize the returned slot after a hash table miss
9980 when INSERT is true.
9982 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9984 * config/s390/s390.md (mem_signal_fence): Remove.
9985 * doc/md.texi (mem_signal_fence): Remove.
9986 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9988 * target-insns.def (mem_signal_fence): Remove.
9990 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9993 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9996 * asan.c (create_odr_indicator): Strip name encoding from assembler
9997 name before appending it after __odr_asan_.
9999 2017-09-01 Martin Liska <mliska@suse.cz>
10001 PR tree-optimization/82059
10002 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10003 frequency only when an edge is redirected.
10005 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10007 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10008 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10009 (arc_conditional_register_usage): Remove ARC600 lp_count
10011 (arc_file_start): Emit Tag_ARC_CPU_variation.
10012 (arc_can_use_doloop_p): New conditions to use ZOLs.
10013 (hwloop_fail): New function.
10014 (hwloop_optimize): Likewise.
10015 (hwloop_pattern_reg): Likewise.
10016 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10017 (arc_reorg_loops): New function, calls reorg_loops.
10018 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10019 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10021 (arc_loop_hazard): Remove function, functionality moved into
10023 (arc_hazard): Remove arc_loop_hazard call.
10024 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10025 into hwloop_optimize.
10026 (arc_label_align): Remove ZOL handling.
10027 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10028 * config/arc/arc.md (doloop_begin): Remove pattern.
10029 (doloop_begin_i): Likewise.
10030 (doloop_end_i): Likewise.
10031 (doloop_fallback): Likewise.
10032 (doloop_fallback_m): Likewise.
10033 (doloop_end): Reimplement expand.
10034 (arc_lp): New pattern for LP instruction.
10035 (loop_end): New pattern.
10036 (loop_fail): Likewise.
10037 (decrement_and_branch_until_zero): Likewise.
10038 * config/arc/arc.opt (mlpc-width): New option.
10039 * doc/invoke.texi (mlpc-width): Document option.
10041 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10043 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10044 (arc_ccfsm_advance): Fix checking for delay slots.
10045 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10047 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10049 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10051 (movhi_insn): Update store instruction constraint which are saving
10052 6-bit short immediates.
10053 (movsi_insn): Consider also short scaled load operations.
10054 (zero_extendhisi2_i): Use Usd constraint instead of T.
10055 (extendhisi2_i): Add q constraint.
10056 (arc_clzsi2): Add type and length attributes.
10057 (arc_ctzsi2): Likewise.
10058 * config/arc/constraints.md (Usc): Update constraint, the
10059 assembler can parse two relocations for a single instruction.
10061 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10063 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10064 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10066 2017-08-31 Olivier Hainque <hainque@adacore.com>
10068 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10069 match as powerpc-wrs-vxworks*.
10071 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10073 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10074 register constraint for by-element operand.
10075 (aarch64_mls_elt_merge<mode>): Likewise.
10077 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10079 * config/arc/arc.c (arc_can_follow_jump): Check for short
10082 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10084 * config.gcc: Use g.opt for arc.
10085 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10086 functionality moved to ...
10087 (legitimate_scaled_address_p): New function, ...here.
10088 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10089 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10090 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10092 (arc_override_options): Handle G option.
10093 (arc_output_pic_addr_const): Correct function definition.
10094 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10095 (arc_decl_anon_ns_mem_p): Delete.
10096 (arc_in_small_data_p): Overhaul this function to take into
10097 consideration the value given via G option.
10098 (arc_rewrite_small_data_1): Renamed and corrected old
10099 arc_rewrite_small_data function.
10100 (arc_rewrite_small_data): New function.
10101 (small_data_pattern): Don't use pic_offset_table_rtx.
10102 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10103 * config/arc/simdext.md (movmisalignv2hi): Use
10104 prepare_move_operands function.
10106 (movmisalign*): Likewise.
10107 * doc/invoke.texi (ARC options): Document -G option.
10109 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10111 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10113 * config/arc/arc.c (arc_print_operand): Output scalled address for
10114 sdata whenever is possible.
10115 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10116 load/stores are available.
10117 (compact_sda_memory_operand): Check for the alignment required by
10118 code density instructions.
10119 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10121 * config/arc/constraints.md (Usd): Update constraint.
10122 (Us0): New constraint.
10123 (Usc): Update constraint.
10125 2017-08-31 Richard Biener <rguenther@suse.de>
10127 PR middle-end/82054
10128 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10129 function only once.
10131 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10133 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10134 Resize type_signature.
10136 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10137 Alan Hayward <alan.hayward@arm.com>
10138 David Sherwood <david.sherwood@arm.com>
10140 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10141 subregs whose inner modes can be stored in GPRs.
10142 (aarch64_classify_index): Likewise.
10144 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10145 Alan Hayward <alan.hayward@arm.com>
10146 David Sherwood <david.sherwood@arm.com>
10148 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10149 (V_INT_EQUIV): ...this.
10150 (v_cmp_result): Rename to...
10151 (v_int_equiv): ...this.
10152 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10153 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10154 (copysign<mode>3): Likewise.
10155 (aarch64_simd_bsl<mode>_internal): Likewise.
10156 (aarch64_simd_bsl<mode>): Likewise.
10157 (vec_cmp<mode><mode>): Likewise.
10158 (vcond<mode><mode>): Likewise.
10159 (vcond<v_cmp_mixed><mode>): Likewise.
10160 (vcondu<mode><v_cmp_mixed>): Likewise.
10161 (aarch64_cm<optab><mode>): Likewise.
10162 (aarch64_cmtst<mode>): Likewise.
10163 (aarch64_fac<optab><mode>): Likewise.
10164 (vec_perm_const<mode>): Likewise.
10165 (vcond_mask_<mode><v_cmp_result>): Rename to...
10166 (vcond_mask_<mode><v_int_equiv>): ...this.
10167 (vec_cmp<mode><v_cmp_result>): Rename to...
10168 (vec_cmp<mode><v_int_equiv>): ...this.
10170 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10171 Alan Hayward <alan.hayward@arm.com>
10172 David Sherwood <david.sherwood@arm.com>
10174 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10176 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10177 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10178 (UNSPEC_LD4_DREG): New unspecs.
10179 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10180 (aarch64_ld2<mode>_dreg_be): Replace with...
10181 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10183 (aarch64_ld3<mode>_dreg_le)
10184 (aarch64_ld3<mode>_dreg_be): Replace with...
10185 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10187 (aarch64_ld4<mode>_dreg_le)
10188 (aarch64_ld4<mode>_dreg_be): Replace with...
10189 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10192 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10194 PR tree-optimization/81987
10195 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10196 insert an initializer in a location not dominated by the stride
10199 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10201 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10202 on the entire header of the finally block in the fallthru case.
10204 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10206 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10208 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10210 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10211 rs6000_emit_move_from_cr and call renamed function.
10212 (rs6000_emit_prologue): Call renamed functions.
10213 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10214 movesi_from_cr, remove volatile CRs.
10216 2017-08-30 Jon Beniston <jon@beniston.com>
10217 Richard Biener <rguenther@suse.de>
10219 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10220 of VECTOR_MODE_P check.
10221 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10222 element vector types.
10224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10226 * df.h (df_read_modify_subreg_p): Remove in favor of...
10227 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10228 const_rtx instead of an rtx.
10229 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10230 * df-problems.c (df_word_lr_mark_ref): Likewise.
10231 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10232 (mark_pseudo_reg_dead): Likewise.
10233 (mark_ref_dead): Likewise.
10234 * reginfo.c (init_subregs_of_mode): Likewise.
10235 * sched-deps.c (sched_analyze_1): Likewise.
10236 * df-scan.c (df_def_record_1): Likewise.
10237 (df_uses_record): Likewise.
10238 (df_read_modify_subreg_p): Remove in favor of...
10239 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10240 const_rtx instead of an rtx.
10242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10243 Alan Hayward <alan.hayward@arm.com>
10244 David Sherwood <david.sherwood@arm.com>
10246 * rtl.h (partial_subreg_p): New function.
10247 * caller-save.c (save_call_clobbered_regs): Use it.
10248 * calls.c (expand_call): Likewise.
10249 * combine.c (combinable_i3pat): Likewise.
10250 (simplify_set): Likewise.
10251 (make_extraction): Likewise.
10252 (make_compound_operation_int): Likewise.
10253 (gen_lowpart_or_truncate): Likewise.
10254 (force_to_mode): Likewise.
10255 (make_field_assignment): Likewise.
10256 (reg_truncated_to_mode): Likewise.
10257 (record_truncated_value): Likewise.
10258 (move_deaths): Likewise.
10259 * cse.c (record_jump_cond): Likewise.
10260 (cse_insn): Likewise.
10261 * cselib.c (cselib_lookup_1): Likewise.
10262 * expmed.c (extract_bit_field_using_extv): Likewise.
10263 * function.c (assign_parm_setup_reg): Likewise.
10264 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10265 * ira-build.c (create_insn_allocnos): Likewise.
10266 * lra-coalesce.c (merge_pseudos): Likewise.
10267 * lra-constraints.c (match_reload): Likewise.
10268 (simplify_operand_subreg): Likewise.
10269 (curr_insn_transform): Likewise.
10270 * lra-lives.c (process_bb_lives): Likewise.
10271 * lra.c (new_insn_reg): Likewise.
10272 (lra_substitute_pseudo): Likewise.
10273 * regcprop.c (mode_change_ok): Likewise.
10274 (maybe_mode_change): Likewise.
10275 (copyprop_hardreg_forward_1): Likewise.
10276 * reload.c (push_reload): Likewise.
10277 (find_reloads): Likewise.
10278 (find_reloads_subreg_address): Likewise.
10279 * reload1.c (alter_reg): Likewise.
10280 (eliminate_regs_1): Likewise.
10281 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10283 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10285 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10286 back to if statements, including unpack.
10288 2017-08-30 Martin Liska <mliska@suse.cz>
10290 PR inline-asm/82001
10291 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10293 (func_checker::compare_asm_inputs_outputs): ... this function.
10294 (func_checker::compare_gimple_asm): Use the function to compare
10295 also ASM constrains.
10296 * ipa-icf-gimple.h: Rename the function.
10298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10299 Alan Hayward <alan.hayward@arm.com>
10300 David Sherwood <david.sherwood@arm.com>
10302 * coretypes.h (complex_mode): New type.
10303 * gdbhooks.py (build_pretty_printer): Handle it.
10304 * machmode.h (complex_mode): New class.
10305 (complex_mode::includes_p): New function.
10306 (is_complex_int_mode): Likewise.
10307 (is_complex_float_mode): Likewise.
10308 * genmodes.c (get_mode_class): Handle complex mode classes.
10309 * function.c (expand_function_end): Use is_complex_int_mode.
10311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10312 Alan Hayward <alan.hayward@arm.com>
10313 David Sherwood <david.sherwood@arm.com>
10315 * coretypes.h (scalar_mode_pod): New typedef.
10316 * gdbhooks.py (build_pretty_printer): Handle it.
10317 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10318 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10319 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10320 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10321 as_a <scalar_mode>.
10323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10324 Alan Hayward <alan.hayward@arm.com>
10325 David Sherwood <david.sherwood@arm.com>
10327 * machmode.h (mode_for_vector): Take a scalar_mode instead
10329 * stor-layout.c (mode_for_vector): Likewise.
10330 * explow.c (promote_mode): Use as_a <scalar_mode>.
10331 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10334 Alan Hayward <alan.hayward@arm.com>
10335 David Sherwood <david.sherwood@arm.com>
10337 * target.def (preferred_simd_mode): Take a scalar_mode
10338 instead of a machine_mode.
10339 * targhooks.h (default_preferred_simd_mode): Likewise.
10340 * targhooks.c (default_preferred_simd_mode): Likewise.
10341 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10342 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10343 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10344 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10345 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10346 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10347 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10348 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10350 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10351 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10352 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10353 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10354 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10355 * doc/tm.texi: Regenerate.
10356 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10360 Alan Hayward <alan.hayward@arm.com>
10361 David Sherwood <david.sherwood@arm.com>
10363 * target.def (scalar_mode_supported_p): Take a scalar_mode
10364 instead of a machine_mode.
10365 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10366 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10367 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10368 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10369 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10370 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10371 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10372 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10373 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10374 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10375 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10376 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10377 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10378 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10380 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10381 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10382 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10383 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10384 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10386 * doc/tm.texi: Regenerate.
10388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10389 Alan Hayward <alan.hayward@arm.com>
10390 David Sherwood <david.sherwood@arm.com>
10392 * coretypes.h (opt_scalar_mode): New typedef.
10393 * gdbhooks.py (build_pretty_printers): Handle it.
10394 * machmode.h (mode_iterator::get_2xwider): Add overload for
10396 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10398 * expr.c (convert_mode_scalar): Likewise.
10399 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10400 * optabs.c (expand_float): Likewise.
10401 (expand_fix): Likewise.
10402 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10405 Alan Hayward <alan.hayward@arm.com>
10406 David Sherwood <david.sherwood@arm.com>
10408 * optabs.c (expand_float): Explicitly check for scalars before
10409 using a branching expansion.
10410 (expand_fix): Likewise.
10412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10413 Alan Hayward <alan.hayward@arm.com>
10414 David Sherwood <david.sherwood@arm.com>
10416 * expr.c (convert_mode): Split scalar handling out into...
10417 (convert_mode_scalar): ...this new function. Treat the modes
10420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10421 Alan Hayward <alan.hayward@arm.com>
10422 David Sherwood <david.sherwood@arm.com>
10424 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10426 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10429 Alan Hayward <alan.hayward@arm.com>
10430 David Sherwood <david.sherwood@arm.com>
10432 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10433 rather than a machine_mode.
10434 (fixed_from_string): Likewise.
10435 (fixed_convert): Likewise.
10436 (fixed_convert_from_int): Likewise.
10437 (fixed_convert_from_real): Likewise.
10438 (real_convert_from_fixed): Likewise.
10439 * fixed-value.c (fixed_from_double_int): Likewise.
10440 (fixed_from_string): Likewise.
10441 (fixed_convert): Likewise.
10442 (fixed_convert_from_int): Likewise.
10443 (fixed_convert_from_real): Likewise.
10444 (real_convert_from_fixed): Likewise.
10445 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10448 Alan Hayward <alan.hayward@arm.com>
10449 David Sherwood <david.sherwood@arm.com>
10451 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10452 of separate mode class checks. Do not allow vector modes here.
10453 (immed_wide_int_const): Use as_a <scalar_mode>.
10454 * explow.c (trunc_int_for_mode): Likewise.
10455 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10456 (wi::shwi): Likewise.
10457 (wi::min_value): Likewise.
10458 (wi::max_value): Likewise.
10459 * dwarf2out.c (loc_descriptor): Likewise.
10460 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10461 for CONST_WIDE_INT.
10463 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10464 Alan Hayward <alan.hayward@arm.com>
10465 David Sherwood <david.sherwood@arm.com>
10467 * tree.h (SCALAR_TYPE_MODE): New macro.
10468 * expr.c (expand_expr_addr_expr_1): Use it.
10469 (expand_expr_real_2): Likewise.
10470 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10471 (fold_convert_const_fixed_from_int): Likewise.
10472 (fold_convert_const_fixed_from_real): Likewise.
10473 (native_encode_fixed): Likewise
10474 (native_encode_complex): Likewise
10475 (native_encode_vector): Likewise.
10476 (native_interpret_fixed): Likewise.
10477 (native_interpret_real): Likewise.
10478 (native_interpret_complex): Likewise.
10479 (native_interpret_vector): Likewise.
10480 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10481 (simd_clone_adjust_argument_types): Likewise.
10482 (simd_clone_init_simd_arrays): Likewise.
10483 (simd_clone_adjust): Likewise.
10484 * stor-layout.c (layout_type): Likewise.
10485 * tree.c (build_minus_one_cst): Likewise.
10486 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10487 * tree-inline.c (estimate_move_cost): Likewise.
10488 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10489 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10490 (vectorizable_reduction): Likewise.
10491 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10492 (vect_recog_mixed_size_cond_pattern): Likewise.
10493 (check_bool_pattern): Likewise.
10494 (adjust_bool_pattern): Likewise.
10495 (search_type_for_mask_1): Likewise.
10496 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10497 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10498 (vectorizable_load): Likewise.
10499 (vectorizable_store): Likewise.
10500 * ubsan.c (ubsan_encode_value): Likewise.
10501 * varasm.c (output_constant): Likewise.
10503 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10504 Alan Hayward <alan.hayward@arm.com>
10505 David Sherwood <david.sherwood@arm.com>
10507 * coretypes.h (scalar_mode): New class.
10508 * machmode.h (scalar_mode): Likewise.
10509 (scalar_mode::includes_p): New function.
10510 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10511 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10512 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10513 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10514 * expmed.c (store_bit_field_1): Likewise.
10515 (extract_bit_field_1): Likewise.
10516 * expr.c (write_complex_part): Likewise.
10517 (read_complex_part): Likewise.
10518 (emit_move_complex_push): Likewise.
10519 (expand_expr_real_2): Likewise.
10520 * function.c (assign_parm_setup_reg): Likewise.
10521 (assign_parms_unsplit_complex): Likewise.
10522 * optabs.c (expand_binop): Likewise.
10523 * rtlanal.c (subreg_get_info): Likewise.
10524 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10525 * varasm.c (output_constant_pool_2): Likewise.
10527 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10528 Alan Hayward <alan.hayward@arm.com>
10529 David Sherwood <david.sherwood@arm.com>
10531 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10533 (expmed_mult_highpart_optab): Likewise.
10534 (expmed_mult_highpart): Likewise.
10536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10537 Alan Hayward <alan.hayward@arm.com>
10538 David Sherwood <david.sherwood@arm.com>
10540 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10541 instead of a machine_mode.
10542 (builtin_memset_read_str): Likewise.
10543 * builtins.c (c_readstr): Likewise.
10544 (builtin_memcpy_read_str): Likewise.
10545 (builtin_strncpy_read_str): Likewise.
10546 (builtin_memset_read_str): Likewise.
10547 (builtin_memset_gen_str): Likewise.
10548 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10549 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10550 instead of a machine_mode.
10551 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10553 (make_extraction): Likewise.
10554 (try_widen_shift_mode): Take and return scalar_int_modes instead
10556 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10557 a scalar_int_mode instead of a machine_mode.
10558 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10559 (avr_addr_space_pointer_mode): Likewise.
10560 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10561 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10562 (msp430_unwind_word_mode): Likewise.
10563 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10564 (spu_addr_space_pointer_mode): Likewise.
10565 (spu_addr_space_address_mode): Likewise.
10566 (spu_libgcc_cmp_return_mode): Likewise.
10567 (spu_libgcc_shift_count_mode): Likewise.
10568 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10569 (rl78_addr_space_pointer_mode): Likewise.
10570 (fl78_unwind_word_mode): Likewise.
10571 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10573 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10574 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10575 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10576 (mips_valid_pointer_mode): Likewise.
10577 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10578 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10579 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10581 (ft32_addr_space_address_mode): Likewise.
10582 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10583 scalar_int_mode instead of a machine_mode.
10584 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10586 (m32c_addr_space_address_mode): Likewise.
10587 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10588 (rs6000_eh_return_filter_mode): Likewise.
10589 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10590 (rs6000_eh_return_filter_mode): Likewise.
10591 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10592 (s390_libgcc_shift_count_mode): Likewise.
10593 (s390_unwind_word_mode): Likewise.
10594 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10596 * target.def (mode_rep_extended): Likewise.
10597 (valid_pointer_mode): Likewise.
10598 (addr_space.valid_pointer_mode): Likewise.
10599 (eh_return_filter_mode): Return a scalar_int_mode rather than
10601 (libgcc_cmp_return_mode): Likewise.
10602 (libgcc_shift_count_mode): Likewise.
10603 (unwind_word_mode): Likewise.
10604 (addr_space.pointer_mode): Likewise.
10605 (addr_space.address_mode): Likewise.
10606 * doc/tm.texi: Regenerate.
10607 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10609 (do_jump): Use scalar_int_mode for local variables.
10610 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10611 rather than a machine_mode.
10612 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10613 (scompare_loc_descriptor_wide): Likewise.
10614 (scompare_loc_descriptor_narrow): Likewise.
10615 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10617 * except.c (sjlj_emit_dispatch_table): Likewise.
10618 (expand_builtin_eh_copy_values): Likewise.
10619 * explow.c (convert_memory_address_addr_space_1): Likewise.
10620 Take a scalar_int_mode rather than a machine_mode.
10621 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10622 than a machine_mode.
10623 (memory_address_addr_space): Use scalar_int_mode for local variables.
10624 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10625 rather than a machine_mode.
10626 * expmed.c (mask_rtx): Likewise.
10627 (init_expmed_one_conv): Likewise.
10628 (expand_mult_highpart_adjust): Likewise.
10629 (extract_high_half): Likewise.
10630 (expmed_mult_highpart_optab): Likewise.
10631 (expmed_mult_highpart): Likewise.
10632 (expand_smod_pow2): Likewise.
10633 (expand_sdiv_pow2): Likewise.
10634 (emit_store_flag_int): Likewise.
10635 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10637 (extract_low_bits): Likewise.
10638 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10640 * expr.c (pieces_addr::adjust): Likewise.
10641 (can_store_by_pieces): Likewise.
10642 (store_by_pieces): Likewise.
10643 (clear_by_pieces_1): Likewise.
10644 (expand_expr_addr_expr_1): Likewise.
10645 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10646 (expand_expr_real_1): Likewise.
10647 (try_casesi): Likewise.
10648 * final.c (shorten_branches): Likewise.
10649 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10650 type of "mode" to machine_mode.
10651 * internal-fn.c (expand_arith_overflow_result_store): Take a
10652 scalar_int_mode rather than a machine_mode.
10653 (expand_mul_overflow): Use scalar_int_mode for local variables.
10654 * loop-doloop.c (doloop_modify): Likewise.
10655 (doloop_optimize): Likewise.
10656 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10657 than a machine_mode.
10658 (expand_doubleword_shift_condmove): Likewise.
10659 (expand_doubleword_shift): Likewise.
10660 (expand_doubleword_clz): Likewise.
10661 (expand_doubleword_popcount): Likewise.
10662 (expand_doubleword_parity): Likewise.
10663 (expand_absneg_bit): Use scalar_int_mode for local variables.
10664 (prepare_float_lib_cmp): Likewise.
10665 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10666 rather than a machine_mode.
10667 (convert_memory_address_addr_space): Likewise.
10668 (get_mode_bounds): Likewise.
10669 (get_address_mode): Return a scalar_int_mode rather than a
10671 * rtlanal.c (get_address_mode): Likewise.
10672 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10673 than a machine_mode.
10674 * targhooks.c (default_mode_rep_extended): Likewise.
10675 (default_valid_pointer_mode): Likewise.
10676 (default_addr_space_valid_pointer_mode): Likewise.
10677 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10678 than a machine_mode.
10679 (default_libgcc_cmp_return_mode): Likewise.
10680 (default_libgcc_shift_count_mode): Likewise.
10681 (default_unwind_word_mode): Likewise.
10682 (default_addr_space_pointer_mode): Likewise.
10683 (default_addr_space_address_mode): Likewise.
10684 * targhooks.h (default_eh_return_filter_mode): Likewise.
10685 (default_libgcc_cmp_return_mode): Likewise.
10686 (default_libgcc_shift_count_mode): Likewise.
10687 (default_unwind_word_mode): Likewise.
10688 (default_addr_space_pointer_mode): Likewise.
10689 (default_addr_space_address_mode): Likewise.
10690 (default_mode_rep_extended): Take a scalar_int_mode rather than
10692 (default_valid_pointer_mode): Likewise.
10693 (default_addr_space_valid_pointer_mode): Likewise.
10694 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10696 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10697 rather than a machine_mode.
10698 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10699 for local variables.
10700 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10701 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10702 than a machine_mode.
10704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10705 Alan Hayward <alan.hayward@arm.com>
10706 David Sherwood <david.sherwood@arm.com>
10708 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10709 the mode argument to scalar_int_mode.
10710 (do_jump_by_parts_zero_rtx): Likewise.
10711 (do_jump_by_parts_equality_rtx): Likewise.
10712 (do_jump_by_parts_greater): Take a mode argument.
10713 (do_jump_by_parts_equality): Likewise.
10714 (do_jump_1): Update calls accordingly.
10716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10717 Alan Hayward <alan.hayward@arm.com>
10718 David Sherwood <david.sherwood@arm.com>
10720 * is-a.h (safe_dyn_cast): New function.
10721 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10722 (jump_table_for_label): Likewise.
10723 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10724 instead of an rtx_insn *.
10725 (shorten_branches): Use dyn_cast instead of LABEL_P and
10726 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10727 rtx_jump_table_data::get_data_mode.
10728 (final_scan_insn): Likewise.
10730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10731 Alan Hayward <alan.hayward@arm.com>
10732 David Sherwood <david.sherwood@arm.com>
10734 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10735 trying to combine a full-register integer set with a subreg
10738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10739 Alan Hayward <alan.hayward@arm.com>
10740 David Sherwood <david.sherwood@arm.com>
10742 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10743 that is always either address_mode or pointer_mode.
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 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10750 when the two are known to be equal.
10752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10753 Alan Hayward <alan.hayward@arm.com>
10754 David Sherwood <david.sherwood@arm.com>
10756 * simplify-rtx.c (simplify_const_unary_operation): Use
10757 is_a <scalar_int_mode> instead of checking for a nonzero
10758 precision. Forcibly convert op_mode to a scalar_int_mode
10759 in that case. More clearly differentiate the operand and
10760 result modes and use the former when deciding what the value
10761 of a count-bits operation should be. Use is_int_mode instead
10762 of checking for a MODE_INT. Remove redundant check for whether
10763 this mode has a zero precision.
10765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10766 Alan Hayward <alan.hayward@arm.com>
10767 David Sherwood <david.sherwood@arm.com>
10769 * optabs.c (widen_leading): Change the type of the mode argument
10770 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10771 (widen_bswap): Likewise.
10772 (expand_parity): Likewise.
10773 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10774 (expand_ffs): Likewise.
10775 (epand_unop): Check for scalar integer modes before calling the
10778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10779 Alan Hayward <alan.hayward@arm.com>
10780 David Sherwood <david.sherwood@arm.com>
10782 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10784 (expand_expr_real_1): Update call accordingly.
10786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10787 Alan Hayward <alan.hayward@arm.com>
10788 David Sherwood <david.sherwood@arm.com>
10790 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10791 value_mode arguments. Use scalar_int_mode internally.
10792 (store_bit_field_1): Rename the new integer mode from imode
10793 to op0_mode and use it instead of GET_MODE (op0). Update calls
10794 to store_split_bit_field, store_bit_field_using_insv and
10795 store_fixed_bit_field.
10796 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10797 Use scalar_int_mode internally. Use a bit count rather than a mode
10798 when calculating the largest bit size for get_best_mode.
10799 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10800 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10801 Remove assertion that OP0 has a scalar integer mode.
10802 (store_split_bit_field): Add op0_mode and value_mode arguments.
10803 Update calls to extract_fixed_bit_field.
10804 (extract_bit_field_using_extv): Add an op0_mode argument.
10805 Use scalar_int_mode internally.
10806 (extract_bit_field_1): Rename the new integer mode from imode to
10807 op0_mode and use it instead of GET_MODE (op0). Update calls to
10808 extract_split_bit_field, extract_bit_field_using_extv and
10809 extract_fixed_bit_field.
10810 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10811 to extract_split_bit_field and extract_fixed_bit_field_1.
10812 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10813 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10814 on the target mode.
10815 (extract_split_bit_field): Add an op0_mode argument. Update call
10816 to extract_fixed_bit_field.
10818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10819 Alan Hayward <alan.hayward@arm.com>
10820 David Sherwood <david.sherwood@arm.com>
10822 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10823 * explow.c (hard_function_value): Likewise.
10824 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10825 convert_to_mode call outside the loop.
10826 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10827 for the mode iterator. Require the mode specified by max_pieces
10829 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10831 (copy_blkmode_to_reg): Likewise.
10832 (set_storage_via_setmem): Likewise.
10833 * optabs.c (prepare_cmp_insn): Likewise.
10834 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10835 * stor-layout.c (finish_bitfield_representative): Likewise.
10837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10838 Alan Hayward <alan.hayward@arm.com>
10839 David Sherwood <david.sherwood@arm.com>
10841 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10842 * expr.c (convert_move): Use them.
10843 (convert_modes): Likewise.
10844 (store_expr_with_bounds): Likewise.
10846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10847 Alan Hayward <alan.hayward@arm.com>
10848 David Sherwood <david.sherwood@arm.com>
10850 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10851 parameter for the mode of "x". Remove the "known_x", "known_mode"
10852 and "known_ret" arguments. Change the type of the mode argument
10853 to scalar_int_mode.
10854 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10855 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10856 (reg_num_sign_bit_copies_for_combine): Likewise.
10857 * rtlanal.c (nonzero_bits1): Likewise.
10858 (num_sign_bit_copies1): Likewise.
10859 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10860 (reg_num_sign_bit_copies_general): Likewise.
10861 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10862 (reg_nonzero_bits_general): Likewise.
10864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10865 Alan Hayward <alan.hayward@arm.com>
10866 David Sherwood <david.sherwood@arm.com>
10868 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10869 than in subroutines. Return 1 for non-integer modes.
10870 (cached_num_sign_bit_copies): Change the type of the mode parameter
10871 to scalar_int_mode.
10872 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10873 classes. Handle CONST_INT_P first and then check whether X also
10874 has a scalar integer mode. Check the same thing for inner registers
10875 of a SUBREG and for values that are being extended or truncated.
10877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10878 Alan Hayward <alan.hayward@arm.com>
10879 David Sherwood <david.sherwood@arm.com>
10881 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10882 in subroutines. Return the mode mask for non-integer modes.
10883 (cached_nonzero_bits): Change the type of the mode parameter
10884 to scalar_int_mode.
10885 (nonzero_bits1): Likewise. Remove early exit for other mode
10886 classes. Handle CONST_INT_P first and then check whether X
10887 also has a scalar integer mode.
10889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10890 Alan Hayward <alan.hayward@arm.com>
10891 David Sherwood <david.sherwood@arm.com>
10893 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10894 Return a scalar_int_mode and assert that the size is greater than
10896 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10897 (op_by_pieces_d::op_by_pieces_d): Likewise.
10898 (op_by_pieces_d::run): Likewise.
10899 (can_store_by_pieces): Likewise.
10901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10902 Alan Hayward <alan.hayward@arm.com>
10903 David Sherwood <david.sherwood@arm.com>
10905 * combine.c (extract_left_shift): Add a mode argument and update
10907 (make_compound_operation_int): Change the type of the mode parameter
10908 to scalar_int_mode and update the call to extract_left_shift.
10910 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10911 Alan Hayward <alan.hayward@arm.com>
10912 David Sherwood <david.sherwood@arm.com>
10914 * combine.c (simplify_and_const_int): Change the type of the mode
10915 parameter to scalar_int_mode.
10916 (simplify_and_const_int_1): Likewise. Update recursive call.
10918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10919 Alan Hayward <alan.hayward@arm.com>
10920 David Sherwood <david.sherwood@arm.com>
10922 * combine.c (simplify_compare_const): Check that the mode is a
10923 scalar_int_mode (rather than VOIDmode) before testing its
10925 (simplify_comparison): Move COMPARISON_P handling out of the
10926 loop and restrict the latter part of the loop to scalar_int_modes.
10927 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10928 and when considering SUBREG_REGs. Use is_int_mode instead of
10929 checking GET_MODE_CLASS against MODE_INT.
10931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10932 Alan Hayward <alan.hayward@arm.com>
10933 David Sherwood <david.sherwood@arm.com>
10935 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10936 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10937 shift_unit_mode and for modes involved in scalar shifts.
10939 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10940 Alan Hayward <alan.hayward@arm.com>
10941 David Sherwood <david.sherwood@arm.com>
10943 * combine.c (force_int_to_mode): New function, split out from...
10944 (force_to_mode): ...here. Keep xmode up-to-date and use it
10945 instead of GET_MODE (x).
10947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10948 Alan Hayward <alan.hayward@arm.com>
10949 David Sherwood <david.sherwood@arm.com>
10951 * optabs-query.h (extraction_insn::struct_mode): Change type to
10952 opt_scalar_int_mode and update comment.
10953 (extraction_insn::field_mode): Change type to scalar_int_mode.
10954 (extraction_insn::pos_mode): Likewise.
10955 * combine.c (make_extraction): Update accordingly.
10956 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10957 (get_optab_extraction_insn): Likewise.
10958 * recog.c (simplify_while_replacing): Likewise.
10959 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10960 parameter to opt_scalar_int_mode.
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 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10967 to a scalar_int_mode instead of a machine_mode.
10968 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10969 (get_best_mode): Return a boolean and use a pointer argument to store
10970 the selected mode. Replace the limit mode parameter with a bit limit.
10971 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10972 for the values returned by bit_field_mode_iterator::next_mode.
10973 (store_bit_field): Update call to get_best_mode.
10974 (store_fixed_bit_field): Likewise.
10975 (extract_fixed_bit_field): Likewise.
10976 * expr.c (optimize_bitfield_assignment_op): Likewise.
10977 * fold-const.c (optimize_bit_field_compare): Likewise.
10978 (fold_truth_andor_1): Likewise.
10979 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10980 Update for new type of m_mode.
10981 (get_best_mode): As above.
10983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10984 Alan Hayward <alan.hayward@arm.com>
10985 David Sherwood <david.sherwood@arm.com>
10987 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10988 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10989 (store_bit_field): Check is_a <scalar_int_mode> before calling
10990 strict_volatile_bitfield_p.
10991 (extract_bit_field): Likewise.
10993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10994 Alan Hayward <alan.hayward@arm.com>
10995 David Sherwood <david.sherwood@arm.com>
10997 * target.def (cstore_mode): Return a scalar_int_mode.
10998 * doc/tm.texi: Regenerate.
10999 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11000 * targhooks.h (default_cstore_mode): Likewise.
11001 * targhooks.c (default_cstore_mode): Likewise, using a forced
11003 * expmed.c (emit_cstore): Expect the target of the cstore to be
11006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11007 Alan Hayward <alan.hayward@arm.com>
11008 David Sherwood <david.sherwood@arm.com>
11010 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11012 (niter_desc): Likewise mode.
11013 (iv_analyze): Add a mode parameter.
11015 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11016 and change its type to scalar_int_mode.
11017 * loop-iv.c: Update commentary at head of file.
11018 (iv_constant): Pass the mode paraeter before the rtx it describes
11019 and change its type to scalar_int_mode. Remove VOIDmode handling.
11020 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11021 (iv_extend): Likewise.
11022 (shorten_into_mode): Likewise.
11023 (iv_add): Use scalar_int_mode.
11024 (iv_mult): Likewise.
11025 (iv_shift): Likewise.
11026 (canonicalize_iv_subregs): Likewise.
11027 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11028 it describes and change its mode to scalar_int_mode. Also change
11029 the type of the returned inner_mode to scalar_int_mode.
11030 (get_biv_step): Likewise, turning outer_mode from a pointer
11031 into a direct parameter. Update call to get_biv_step_1.
11032 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11033 iv_constant and get_biv_step.
11034 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11035 and change its type to scalar_int_mode. Don't initialise iv->mode
11036 to VOIDmode and remove later checks for its still being VOIDmode.
11037 Update calls to iv_analyze_op and iv_analyze_expr. Check
11038 is_a <scalar_int_mode> when changing the mode under consideration.
11039 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11040 Update call to iv_analyze_expr.
11041 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11042 inner register is not also a scalar_int_mode. Update call to
11044 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11045 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11046 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11047 separate mode class checks. Update calls to iv_analyze. Remove
11048 fix-up of VOIDmodes after iv_analyze_biv.
11049 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11050 don't have a scalar_int_mode. Update call to biv_p.
11052 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11053 Alan Hayward <alan.hayward@arm.com>
11054 David Sherwood <david.sherwood@arm.com>
11056 * cfgexpand.c (convert_debug_memory_address): Use
11057 as_a <scalar_int_mode>.
11058 * combine.c (expand_compound_operation): Likewise.
11059 (make_extraction): Likewise.
11060 (change_zero_ext): Likewise.
11061 (simplify_comparison): Likewise.
11062 * cse.c (cse_insn): Likewise.
11063 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11064 (mem_loc_descriptor): Likewise.
11065 (loc_descriptor): Likewise.
11066 * expmed.c (init_expmed_one_mode): Likewise.
11067 (synth_mult): Likewise.
11068 (emit_store_flag_1): Likewise.
11069 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11070 of a comparison with size.
11071 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11072 (reduce_to_bit_field_precision): Likewise.
11073 * function.c (expand_function_end): Likewise.
11074 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11075 * loop-doloop.c (doloop_modify): Likewise.
11076 * optabs.c (expand_binop): Likewise.
11077 (expand_unop): Likewise.
11078 (expand_copysign_absneg): Likewise.
11079 (prepare_cmp_insn): Likewise.
11080 (maybe_legitimize_operand): Likewise.
11081 * recog.c (const_scalar_int_operand): Likewise.
11082 * rtlanal.c (get_address_mode): Likewise.
11083 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11084 (simplify_cond_clz_ctz): Likewise.
11085 * tree-nested.c (get_nl_goto_field): Likewise.
11086 * tree.c (build_vector_type_for_mode): Likewise.
11087 * var-tracking.c (use_narrower_mode): Likewise.
11089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11090 Alan Hayward <alan.hayward@arm.com>
11091 David Sherwood <david.sherwood@arm.com>
11093 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11094 * builtins.c (expand_builtin_signbit): Use it.
11095 * cfgexpand.c (expand_debug_expr): Likewise.
11096 * dojump.c (do_jump): Likewise.
11097 (do_compare_and_jump): Likewise.
11098 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11099 * expmed.c (make_tree): Likewise.
11100 * expr.c (expand_expr_real_2): Likewise.
11101 (expand_expr_real_1): Likewise.
11102 (try_casesi): Likewise.
11103 * fold-const-call.c (fold_const_call_ss): Likewise.
11104 * fold-const.c (unextend): Likewise.
11105 (extract_muldiv_1): Likewise.
11106 (fold_single_bit_test): Likewise.
11107 (native_encode_int): Likewise.
11108 (native_encode_string): Likewise.
11109 (native_interpret_int): Likewise.
11110 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11111 * internal-fn.c (expand_addsub_overflow): Likewise.
11112 (expand_neg_overflow): Likewise.
11113 (expand_mul_overflow): Likewise.
11114 (expand_arith_overflow): Likewise.
11115 * match.pd: Likewise.
11116 * stor-layout.c (layout_type): Likewise.
11117 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11118 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11119 * tree-ssanames.c (get_range_info): Likewise.
11120 * tree-switch-conversion.c (array_value_type) Likewise.
11121 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11122 (vect_recog_divmod_pattern): Likewise.
11123 (vect_recog_mixed_size_cond_pattern): Likewise.
11124 * tree-vrp.c (extract_range_basic): Likewise.
11125 (simplify_float_conversion_using_ranges): Likewise.
11126 * tree.c (int_fits_type_p): Likewise.
11127 * ubsan.c (instrument_bool_enum_load): Likewise.
11128 * varasm.c (mergeable_string_section): Likewise.
11129 (narrowing_initializer_constant_valid_p): Likewise.
11130 (output_constant): Likewise.
11132 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11133 Alan Hayward <alan.hayward@arm.com>
11134 David Sherwood <david.sherwood@arm.com>
11136 * machmode.h (NARROWEST_INT_MODE): New macro.
11137 * expr.c (alignment_for_piecewise_move): Use it instead of
11138 GET_CLASS_NARROWEST_MODE (MODE_INT).
11139 (push_block): Likewise.
11140 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11142 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11145 Alan Hayward <alan.hayward@arm.com>
11146 David Sherwood <david.sherwood@arm.com>
11148 * postreload.c (move2add_valid_value_p): Change the type of the
11149 mode parameter to scalar_int_mode.
11150 (move2add_use_add2_insn): Add a mode parameter and use it instead
11152 (move2add_use_add3_insn): Likewise.
11153 (reload_cse_move2add): Update accordingly.
11155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11156 Alan Hayward <alan.hayward@arm.com>
11157 David Sherwood <david.sherwood@arm.com>
11159 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11161 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11162 * optabs.c (expand_unop): Likewise.
11164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11165 Alan Hayward <alan.hayward@arm.com>
11166 David Sherwood <david.sherwood@arm.com>
11168 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11169 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11170 (popcount_loc_descriptor): Likewise.
11171 (bswap_loc_descriptor): Likewise.
11172 (rotate_loc_descriptor): Likewise.
11173 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11174 calling the functions above.
11176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11177 Alan Hayward <alan.hayward@arm.com>
11178 David Sherwood <david.sherwood@arm.com>
11180 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11182 (try_combine): Likewise.
11183 (simplify_if_then_else): Likewise.
11184 * cse.c (cse_insn): Likewise.
11185 * dwarf2out.c (mem_loc_descriptor): Likewise.
11186 * emit-rtl.c (gen_lowpart_common): Likewise.
11187 * simplify-rtx.c (simplify_truncation): Likewise.
11188 (simplify_binary_operation_1): Likewise.
11189 (simplify_const_relational_operation): Likewise.
11190 (simplify_ternary_operation): Likewise.
11191 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11194 Alan Hayward <alan.hayward@arm.com>
11195 David Sherwood <david.sherwood@arm.com>
11197 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11198 * reload.c (push_reload): Likewise.
11199 (find_reloads): Likewise.
11201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11202 Alan Hayward <alan.hayward@arm.com>
11203 David Sherwood <david.sherwood@arm.com>
11205 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11206 (make_compound_operation_int): Likewise.
11207 (change_zero_ext): Likewise.
11208 * expr.c (convert_move): Likewise.
11209 (convert_modes): Likewise.
11210 * fwprop.c (forward_propagate_subreg): Likewise.
11211 * loop-iv.c (get_biv_step_1): Likewise.
11212 * optabs.c (widen_operand): Likewise.
11213 * postreload.c (move2add_valid_value_p): Likewise.
11214 * recog.c (simplify_while_replacing): Likewise.
11215 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11216 (simplify_binary_operation_1): Likewise. Remove redundant
11217 mode equality check.
11219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11220 Alan Hayward <alan.hayward@arm.com>
11221 David Sherwood <david.sherwood@arm.com>
11223 * combine.c (combine_simplify_rtx): Add checks for
11224 is_a <scalar_int_mode>.
11225 (simplify_if_then_else): Likewise.
11226 (make_field_assignment): Likewise.
11227 (simplify_comparison): Likewise.
11228 * ifcvt.c (noce_try_bitop): Likewise.
11229 * loop-invariant.c (canonicalize_address_mult): Likewise.
11230 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11233 Alan Hayward <alan.hayward@arm.com>
11234 David Sherwood <david.sherwood@arm.com>
11236 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11237 is_a <scalar_int_mode> instead of != BLKmode.
11239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11240 Alan Hayward <alan.hayward@arm.com>
11241 David Sherwood <david.sherwood@arm.com>
11243 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11244 instead of != VOIDmode.
11245 * combine.c (if_then_else_cond): Likewise.
11246 (change_zero_ext): Likewise.
11247 * dwarf2out.c (mem_loc_descriptor): Likewise.
11248 (loc_descriptor): Likewise.
11249 * rtlanal.c (canonicalize_condition): Likewise.
11250 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11253 Alan Hayward <alan.hayward@arm.com>
11254 David Sherwood <david.sherwood@arm.com>
11256 * simplify-rtx.c (simplify_binary_operation_1): Use
11257 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11260 Alan Hayward <alan.hayward@arm.com>
11261 David Sherwood <david.sherwood@arm.com>
11263 * wide-int.h (int_traits<unsigned char>) New class.
11264 (int_traits<unsigned short>) Likewise.
11265 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11266 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11268 * combine.c (set_nonzero_bits_and_sign_copies): Use
11269 is_a <scalar_int_mode>.
11270 (find_split_point): Likewise.
11271 (combine_simplify_rtx): Likewise.
11272 (simplify_logical): Likewise.
11273 (expand_compound_operation): Likewise.
11274 (expand_field_assignment): Likewise.
11275 (make_compound_operation): Likewise.
11276 (extended_count): Likewise.
11277 (change_zero_ext): Likewise.
11278 (simplify_comparison): Likewise.
11279 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11280 (ucompare_loc_descriptor): Likewise.
11281 (minmax_loc_descriptor): Likewise.
11282 (mem_loc_descriptor): Likewise.
11283 (loc_descriptor): Likewise.
11284 * expmed.c (init_expmed_one_mode): Likewise.
11285 * lra-constraints.c (lra_constraint_offset): Likewise.
11286 * optabs.c (prepare_libcall_arg): Likewise.
11287 * postreload.c (move2add_note_store): Likewise.
11288 * reload.c (operands_match_p): Likewise.
11289 * rtl.h (load_extend_op): Likewise.
11290 * rtlhooks.c (gen_lowpart_general): Likewise.
11291 * simplify-rtx.c (simplify_truncation): Likewise.
11292 (simplify_unary_operation_1): Likewise.
11293 (simplify_binary_operation_1): Likewise.
11294 (simplify_const_binary_operation): Likewise.
11295 (simplify_const_relational_operation): Likewise.
11296 (simplify_subreg): Likewise.
11297 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11298 * var-tracking.c (adjust_mems): Likewise.
11299 (prepare_call_arguments): Likewise.
11301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11302 Alan Hayward <alan.hayward@arm.com>
11303 David Sherwood <david.sherwood@arm.com>
11305 * machmode.h (is_int_mode): New fuction.
11306 * combine.c (find_split_point): Use it.
11307 (combine_simplify_rtx): Likewise.
11308 (simplify_if_then_else): Likewise.
11309 (simplify_set): Likewise.
11310 (simplify_shift_const_1): Likewise.
11311 (simplify_comparison): Likewise.
11312 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11313 * cse.c (notreg_cost): Likewise.
11314 (cse_insn): Likewise.
11315 * cselib.c (cselib_lookup_1): Likewise.
11316 * dojump.c (do_jump_1): Likewise.
11317 (do_compare_rtx_and_jump): Likewise.
11318 * dse.c (get_call_args): Likewise.
11319 * dwarf2out.c (rtl_for_decl_init): Likewise.
11320 (native_encode_initializer): Likewise.
11321 * expmed.c (emit_store_flag_1): Likewise.
11322 (emit_store_flag): Likewise.
11323 * expr.c (convert_modes): Likewise.
11324 (store_field): Likewise.
11325 (expand_expr_real_1): Likewise.
11326 * fold-const.c (fold_read_from_constant_string): Likewise.
11327 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11328 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11329 * optabs.c (expand_binop): Likewise.
11330 (expand_unop): Likewise.
11331 (expand_abs_nojump): Likewise.
11332 (expand_one_cmpl_abs_nojump): Likewise.
11333 * simplify-rtx.c (mode_signbit_p): Likewise.
11334 (val_signbit_p): Likewise.
11335 (val_signbit_known_set_p): Likewise.
11336 (val_signbit_known_clear_p): Likewise.
11337 (simplify_relational_operation_1): Likewise.
11338 * tree.c (vector_type_mode): Likewise.
11340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11341 Alan Hayward <alan.hayward@arm.com>
11342 David Sherwood <david.sherwood@arm.com>
11344 * machmode.h (smallest_mode_for_size): Fix formatting.
11345 (smallest_int_mode_for_size): New function.
11346 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11347 instead of smallest_mode_for_size.
11348 * combine.c (make_extraction): Likewise.
11349 * config/arc/arc.c (arc_expand_movmem): Likewise.
11350 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11351 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11352 * config/s390/s390.c (s390_expand_insv): Likewise.
11353 * config/sparc/sparc.c (assign_int_registers): Likewise.
11354 * config/spu/spu.c (spu_function_value): Likewise.
11355 (spu_function_arg): Likewise.
11356 * coverage.c (get_gcov_type): Likewise.
11357 (get_gcov_unsigned_t): Likewise.
11358 * dse.c (find_shift_sequence): Likewise.
11359 * expmed.c (store_bit_field_1): Likewise.
11360 * expr.c (convert_move): Likewise.
11361 (store_field): Likewise.
11362 * internal-fn.c (expand_arith_overflow): Likewise.
11363 * optabs-query.c (get_best_extraction_insn): Likewise.
11364 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11365 * stor-layout.c (layout_type): Likewise.
11366 (initialize_sizetypes): Likewise.
11367 * targhooks.c (default_get_mask_mode): Likewise.
11368 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11370 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11371 Alan Hayward <alan.hayward@arm.com>
11372 David Sherwood <david.sherwood@arm.com>
11374 * machmode.h (opt_mode::else_blk): New function.
11375 (int_mode_for_mode): Declare.
11376 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11377 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11379 * cfgexpand.c (expand_debug_expr): Likewise.
11380 * combine.c (gen_lowpart_or_truncate): Likewise.
11381 (gen_lowpart_for_combine): Likewise.
11382 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11383 * config/avr/avr.c (avr_to_int_mode): Likewise.
11384 (avr_out_plus_1): Likewise.
11385 (avr_out_plus): Likewise.
11386 (avr_out_round): Likewise.
11387 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11388 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11389 (s390_expand_vcond): Likewise.
11390 * config/spu/spu.c (spu_split_immediate): Likewise.
11391 (spu_expand_mov): Likewise.
11392 * dse.c (get_stored_val): Likewise.
11393 * expmed.c (store_bit_field_1): Likewise.
11394 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11396 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11397 (extract_low_bits): Likewise.
11398 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11399 handling rather than repeating the check.
11400 (emit_group_store): Likewise.
11401 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11402 * optabs.c (expand_absneg_bit): Likewise.
11403 (expand_copysign_absneg): Likewise.
11404 (expand_copysign_bit): Likewise.
11405 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11406 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11407 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11408 * var-tracking.c (prepare_call_arguments): Likewise.
11409 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11410 int_mode_for_mode instead of mode_for_size.
11411 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11414 Alan Hayward <alan.hayward@arm.com>
11415 David Sherwood <david.sherwood@arm.com>
11417 * machmode.h (int_mode_for_size): New function.
11418 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11419 instead of mode_for_size.
11420 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11422 * combine.c (expand_field_assignment): Use int_mode_for_size
11423 instead of mode_for_size.
11424 (make_extraction): Likewise.
11425 (simplify_shift_const_1): Likewise.
11426 (simplify_comparison): Likewise.
11427 * dojump.c (do_jump): Likewise.
11428 * dwarf2out.c (mem_loc_descriptor): Likewise.
11429 * emit-rtl.c (init_derived_machine_modes): Likewise.
11430 * expmed.c (flip_storage_order): Likewise.
11431 (convert_extracted_bit_field): Likewise.
11432 * expr.c (copy_blkmode_from_reg): Likewise.
11433 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11434 * internal-fn.c (expand_mul_overflow): Likewise.
11435 * lower-subreg.c (simple_move): Likewise.
11436 * optabs-libfuncs.c (init_optabs): Likewise.
11437 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11438 * tree.c (vector_type_mode): Likewise.
11439 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11440 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11441 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11442 * tree-vect-stmts.c (vectorizable_load): Likewise.
11443 (vectorizable_store): Likewise.
11445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11446 Alan Hayward <alan.hayward@arm.com>
11447 David Sherwood <david.sherwood@arm.com>
11449 * coretypes.h (pod_mode): New type.
11450 (scalar_int_mode_pod): New typedef.
11451 * machmode.h (pod_mode): New class.
11452 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11453 * genmodes.c (emit_mode_int_n): Update accordingly.
11454 * lower-subreg.h (target_lower_subreg): Change type to
11455 scalar_int_mode_pod.
11456 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11457 scalar_int_mode_pod.
11459 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11460 Alan Hayward <alan.hayward@arm.com>
11461 David Sherwood <david.sherwood@arm.com>
11463 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11464 machine_mode to scalar_int_mode.
11465 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11466 (rs6000_option_override_internal): Remove cast to int.
11467 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11468 machine_mode to scalar_int_mode.
11469 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11470 (rs6000_option_override_internal): Remove cast to int.
11471 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11472 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11474 * config/s390/s390.c (s390_expand_builtin): Likewise.
11475 * coretypes.h (scalar_int_mode): New type.
11476 (opt_scalar_int_mode): New typedef.
11477 * machmode.h (scalar_int_mode): New class.
11478 (scalar_int_mode::includes_p): New function.
11479 (byte_mode): Change type to scalar_int_mode.
11480 (word_mode): Likewise.
11481 (ptr_mode): Likewise.
11482 * emit-rtl.c (byte_mode): Likewise.
11483 (word_mode): Likewise.
11484 (ptr_mode): Likewise.
11485 (init_derived_machine_modes): Update accordingly.
11486 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11487 and MODE_PARTIAL_INT.
11488 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11489 opt_scalar_int_mode.
11491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11492 Alan Hayward <alan.hayward@arm.com>
11493 David Sherwood <david.sherwood@arm.com>
11495 * target.def (libgcc_floating_mode_supported_p): Take a
11497 * doc/tm.texi: Regenerate.
11498 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11500 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11501 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11504 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11505 Alan Hayward <alan.hayward@arm.com>
11506 David Sherwood <david.sherwood@arm.com>
11508 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11509 * doc/tm.texi: Regenerate.
11510 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11511 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11512 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11513 * targhooks.h (default_floatn_mode): Likewise.
11514 * targhooks.c (default_floatn_mode): Likewise.
11515 * tree.c (build_common_tree_nodes): Update accordingly.
11517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11518 Alan Hayward <alan.hayward@arm.com>
11519 David Sherwood <david.sherwood@arm.com>
11521 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11522 (mode_iterator::iterate_p): Likewise.
11523 (mode_iterator::get_wider): Likewise.
11524 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11527 Alan Hayward <alan.hayward@arm.com>
11528 David Sherwood <david.sherwood@arm.com>
11530 * coretypes.h (opt_scalar_float_mode): New typedef.
11531 * machmode.h (float_mode_for_size): New function.
11532 * emit-rtl.c (double_mode): Delete.
11533 (init_emit_once): Use float_mode_for_size.
11534 * stor-layout.c (layout_type): Likewise.
11535 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11538 Alan Hayward <alan.hayward@arm.com>
11539 David Sherwood <david.sherwood@arm.com>
11541 * output.h (assemble_real): Take a scalar_float_mode.
11542 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11543 * config/arm/arm.md (consttable_4): Likewise.
11544 (consttable_8): Likewise.
11545 (consttable_16): Likewise.
11546 * config/mips/mips.md (consttable_float): Likewise.
11547 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11548 * varasm.c (assemble_real): Take a scalar_float_mode.
11549 (output_constant_pool_2): Update accordingly.
11550 (output_constant): Likewise.
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 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11557 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11558 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11559 (native_encode_real): Likewise.
11560 (native_interpret_real): Likewise.
11561 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11562 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11565 Alan Hayward <alan.hayward@arm.com>
11566 David Sherwood <david.sherwood@arm.com>
11568 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11569 <scalar_float_mode>. Simplify.
11570 (gen_extend_conv_libfunc): Likewise.
11572 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11573 Alan Hayward <alan.hayward@arm.com>
11574 David Sherwood <david.sherwood@arm.com>
11576 * coretypes.h (scalar_float_mode): New type.
11577 * machmode.h (mode_traits::from_int): Use machine_mode if
11578 USE_ENUM_MODES is defined.
11579 (is_a): New function.
11581 (dyn_cast): Likewise.
11582 (scalar_float_mode): New class.
11583 (scalar_float_mode::includes_p): New function.
11584 (is_float_mode): Likewise.
11585 * gdbhooks.py (MachineModePrinter): New class.
11586 (build_pretty_printer): Use it for scalar_float_mode.
11587 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11588 (format_helper::format_helper): Turn into a template.
11589 * genmodes.c (get_mode_class): New function.
11590 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11591 or machine_mode if none.
11592 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11593 as_a <scalar_float_mode>.
11594 * dwarf2out.c (mem_loc_descriptor): Likewise.
11595 (insert_float): Likewise.
11596 (add_const_value_attribute): Likewise.
11597 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11598 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11599 (expand_unop): Update accordingly.
11600 (expand_abs_nojump): Likewise.
11601 (expand_copysign_absneg): Take a scalar_float_mode.
11602 (expand_copysign_bit): Likewise.
11603 (expand_copysign): Update accordingly.
11605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11606 Alan Hayward <alan.hayward@arm.com>
11607 David Sherwood <david.sherwood@arm.com>
11609 * coretypes.h (opt_mode): New class.
11610 * machmode.h (opt_mode): Likewise.
11611 (opt_mode::else_void): New function.
11612 (opt_mode::require): Likewise.
11613 (opt_mode::exists): Likewise.
11614 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11615 (GET_MODE_2XWIDER_MODE): Likewise.
11616 (mode_iterator::get_wider): Update accordingly.
11617 (mode_iterator::get_2xwider): Likewise.
11618 (mode_iterator::get_known_wider): Likewise, turning into a template.
11619 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11620 forcing a wider mode to exist.
11621 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11622 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11623 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11624 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11625 * lower-subreg.c (init_lower_subreg): Likewise.
11626 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11627 on the final iteration.
11628 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11629 a wider mode exists before asking for a move pattern.
11630 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11631 forcing a wider mode to exist.
11632 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11633 returning false if no such mode exists.
11634 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11635 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11636 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11637 Avoid checking for a MODE_INT if we already know the mode is not a
11639 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11640 forcing a wider mode to exist.
11641 (expmed_mult_highpart_optab): Likewise.
11642 (expmed_mult_highpart): Likewise.
11643 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11645 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11646 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11647 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11648 * internal-fn.c (expand_mul_overflow): Update use of
11649 GET_MODE_2XWIDER_MODE.
11650 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11651 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11652 GET_MODE_WIDER_MODE.
11653 (convert_plusminus_to_widen): Likewise.
11654 * tree-switch-conversion.c (array_value_type): Likewise.
11655 * var-tracking.c (emit_note_insn_var_location): Likewise.
11656 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11657 Return false inside rather than outside the loop if no wider mode
11659 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11660 and GET_MODE_2XWIDER_MODE
11661 (can_compare_p): Use else_void.
11662 * gdbhooks.py (OptMachineModePrinter): New class.
11663 (build_pretty_printer): Use it for opt_mode.
11665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11666 Alan Hayward <alan.hayward@arm.com>
11667 David Sherwood <david.sherwood@arm.com>
11669 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11670 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11673 Alan Hayward <alan.hayward@arm.com>
11674 David Sherwood <david.sherwood@arm.com>
11676 * machmode.h (mode_traits): New structure.
11677 (get_narrowest_mode): New function.
11678 (mode_iterator::start): Likewise.
11679 (mode_iterator::iterate_p): Likewise.
11680 (mode_iterator::get_wider): Likewise.
11681 (mode_iterator::get_known_wider): Likewise.
11682 (mode_iterator::get_2xwider): Likewise.
11683 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11684 (FOR_EACH_MODE): Likewise.
11685 (FOR_EACH_MODE_FROM): Likewise.
11686 (FOR_EACH_MODE_UNTIL): Likewise.
11687 (FOR_EACH_WIDER_MODE): Likewise.
11688 (FOR_EACH_2XWIDER_MODE): Likewise.
11689 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11690 * combine.c (simplify_comparison): Likewise
11691 * config/i386/i386.c (type_natural_mode): Likewise.
11692 * cse.c (cse_insn): Likewise.
11693 * dse.c (find_shift_sequence): Likewise.
11694 * emit-rtl.c (init_derived_machine_modes): Likewise.
11695 (init_emit_once): Likewise.
11696 * explow.c (hard_function_value): Likewise.
11697 * expmed.c (extract_fixed_bit_field_1): Likewise.
11698 (extract_bit_field_1): Likewise.
11699 (expand_divmod): Likewise.
11700 (emit_store_flag_1): Likewise.
11701 * expr.c (init_expr_target): Likewise.
11702 (convert_move): Likewise.
11703 (alignment_for_piecewise_move): Likewise.
11704 (widest_int_mode_for_size): Likewise.
11705 (emit_block_move_via_movmem): Likewise.
11706 (copy_blkmode_to_reg): Likewise.
11707 (set_storage_via_setmem): Likewise.
11708 (compress_float_constant): Likewise.
11709 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11710 * optabs-query.c (get_best_extraction_insn): Likewise.
11711 * optabs.c (expand_binop): Likewise.
11712 (expand_twoval_unop): Likewise.
11713 (expand_twoval_binop): Likewise.
11714 (widen_leading): Likewise.
11715 (widen_bswap): Likewise.
11716 (expand_parity): Likewise.
11717 (expand_unop): Likewise.
11718 (prepare_cmp_insn): Likewise.
11719 (prepare_float_lib_cmp): Likewise.
11720 (expand_float): Likewise.
11721 (expand_fix): Likewise.
11722 (expand_sfix_optab): Likewise.
11723 * postreload.c (move2add_use_add2_insn): Likewise.
11724 * reg-stack.c (reg_to_stack): Likewise.
11725 * reginfo.c (choose_hard_reg_mode): Likewise.
11726 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11727 * stor-layout.c (mode_for_size): Likewise.
11728 (smallest_mode_for_size): Likewise.
11729 (mode_for_vector): Likewise.
11730 (finish_bitfield_representative): Likewise.
11731 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11732 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11733 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11734 * var-tracking.c (prepare_call_arguments): Likewise.
11736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11737 Alan Hayward <alan.hayward@arm.com>
11738 David Sherwood <david.sherwood@arm.com>
11740 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11741 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11742 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11743 * machmode.h (mode_size): Move earlier in file.
11744 (mode_precision): Likewise.
11745 (mode_inner): Likewise.
11746 (mode_nunits): Likewise.
11747 (mode_unit_size): Likewise.
11748 (unit_unit_precision): Likewise.
11749 (mode_wider): Likewise.
11750 (mode_2xwider): Likewise.
11751 (machine_mode): New class.
11752 (mode_to_bytes): New function.
11753 (mode_to_bits): Likewise.
11754 (mode_to_precision): Likewise.
11755 (mode_to_inner): Likewise.
11756 (mode_to_unit_size): Likewise.
11757 (mode_to_unit_precision): Likewise.
11758 (mode_to_nunits): Likewise.
11759 (GET_MODE_SIZE): Use mode_to_bytes.
11760 (GET_MODE_BITSIZE): Use mode_to_bits.
11761 (GET_MODE_PRECISION): Use mode_to_precision.
11762 (GET_MODE_INNER): Use mode_to_inner.
11763 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11764 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11765 (GET_MODE_NUNITS): Use mode_to_nunits.
11766 * system.h (ALWAYS_INLINE): New macro.
11767 * config/powerpcspe/powerpcspe-c.c
11768 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11769 int for arg1_mode and arg2_mode.
11771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11772 Alan Hayward <alan.hayward@arm.com>
11773 David Sherwood <david.sherwood@arm.com>
11775 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11776 Prefix mode names with E_ in case statements.
11777 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11778 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11779 (aarch64_split_simd_move): Likewise.
11780 (aarch64_gen_storewb_pair): Likewise.
11781 (aarch64_gen_loadwb_pair): Likewise.
11782 (aarch64_gen_store_pair): Likewise.
11783 (aarch64_gen_load_pair): Likewise.
11784 (aarch64_get_condition_code_1): Likewise.
11785 (aarch64_constant_pool_reload_icode): Likewise.
11786 (get_rsqrte_type): Likewise.
11787 (get_rsqrts_type): Likewise.
11788 (get_recpe_type): Likewise.
11789 (get_recps_type): Likewise.
11790 (aarch64_gimplify_va_arg_expr): Likewise.
11791 (aarch64_simd_container_mode): Likewise.
11792 (aarch64_emit_load_exclusive): Likewise.
11793 (aarch64_emit_store_exclusive): Likewise.
11794 (aarch64_expand_compare_and_swap): Likewise.
11795 (aarch64_gen_atomic_cas): Likewise.
11796 (aarch64_emit_bic): Likewise.
11797 (aarch64_emit_atomic_swap): Likewise.
11798 (aarch64_emit_atomic_load_op): Likewise.
11799 (aarch64_evpc_trn): Likewise.
11800 (aarch64_evpc_uzp): Likewise.
11801 (aarch64_evpc_zip): Likewise.
11802 (aarch64_evpc_ext): Likewise.
11803 (aarch64_evpc_rev): Likewise.
11804 (aarch64_evpc_dup): Likewise.
11805 (aarch64_gen_ccmp_first): Likewise.
11806 (aarch64_gen_ccmp_next): Likewise.
11807 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11808 (alpha_emit_xfloating_libcall): Likewise.
11809 (emit_insxl): Likewise.
11810 (alpha_arg_type): Likewise.
11811 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11812 (arc_preferred_simd_mode): Likewise.
11813 (arc_secondary_reload): Likewise.
11814 (get_arc_condition_code): Likewise.
11815 (arc_print_operand): Likewise.
11816 (arc_legitimate_constant_p): Likewise.
11817 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11818 * config/arc/arc.md (casesi_load): Likewise.
11819 (casesi_compact_jump): Likewise.
11820 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11821 (cc_use_register): Likewise.
11822 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11823 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11824 (arm_init_iwmmxt_builtins): Likewise.
11825 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11826 (neon_expand_vector_init): Likewise.
11827 (arm_attr_length_move_neon): Likewise.
11828 (maybe_get_arm_condition_code): Likewise.
11829 (arm_emit_vector_const): Likewise.
11830 (arm_preferred_simd_mode): Likewise.
11831 (arm_output_iwmmxt_tinsr): Likewise.
11832 (thumb1_output_casesi): Likewise.
11833 (thumb2_output_casesi): Likewise.
11834 (arm_emit_load_exclusive): Likewise.
11835 (arm_emit_store_exclusive): Likewise.
11836 (arm_expand_compare_and_swap): Likewise.
11837 (arm_evpc_neon_vuzp): Likewise.
11838 (arm_evpc_neon_vzip): Likewise.
11839 (arm_evpc_neon_vrev): Likewise.
11840 (arm_evpc_neon_vtrn): Likewise.
11841 (arm_evpc_neon_vext): Likewise.
11842 (arm_validize_comparison): Likewise.
11843 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11844 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11845 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11846 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11847 (c6x_preferred_simd_mode): Likewise.
11848 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11849 (epiphany_rtx_costs): Likewise.
11850 * config/epiphany/predicates.md (proper_comparison_operator):
11852 * config/frv/frv.c (condexec_memory_operand): Likewise.
11853 (frv_emit_move): Likewise.
11854 (output_move_single): Likewise.
11855 (output_condmove_single): Likewise.
11856 (frv_hard_regno_mode_ok): Likewise.
11857 (frv_matching_accg_mode): Likewise.
11858 * config/h8300/h8300.c (split_adds_subs): Likewise.
11859 (h8300_rtx_costs): Likewise.
11860 (h8300_print_operand): Likewise.
11861 (compute_mov_length): Likewise.
11862 (output_logical_op): Likewise.
11863 (compute_logical_op_length): Likewise.
11864 (compute_logical_op_cc): Likewise.
11865 (h8300_shift_needs_scratch_p): Likewise.
11866 (output_a_shift): Likewise.
11867 (compute_a_shift_length): Likewise.
11868 (compute_a_shift_cc): Likewise.
11869 (expand_a_rotate): Likewise.
11870 (output_a_rotate): Likewise.
11871 * config/i386/i386.c (classify_argument): Likewise.
11872 (function_arg_advance_32): Likewise.
11873 (function_arg_32): Likewise.
11874 (function_arg_64): Likewise.
11875 (function_value_64): Likewise.
11876 (ix86_gimplify_va_arg): Likewise.
11877 (ix86_legitimate_constant_p): Likewise.
11878 (put_condition_code): Likewise.
11879 (split_double_mode): Likewise.
11880 (ix86_avx256_split_vector_move_misalign): Likewise.
11881 (ix86_expand_vector_logical_operator): Likewise.
11882 (ix86_split_idivmod): Likewise.
11883 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11884 (ix86_build_const_vector): Likewise.
11885 (ix86_build_signbit_mask): Likewise.
11886 (ix86_match_ccmode): Likewise.
11887 (ix86_cc_modes_compatible): Likewise.
11888 (ix86_expand_branch): Likewise.
11889 (ix86_expand_sse_cmp): Likewise.
11890 (ix86_expand_sse_movcc): Likewise.
11891 (ix86_expand_int_sse_cmp): Likewise.
11892 (ix86_expand_vec_perm_vpermi2): Likewise.
11893 (ix86_expand_vec_perm): Likewise.
11894 (ix86_expand_sse_unpack): Likewise.
11895 (ix86_expand_int_addcc): Likewise.
11896 (ix86_split_to_parts): Likewise.
11897 (ix86_vectorize_builtin_gather): Likewise.
11898 (ix86_vectorize_builtin_scatter): Likewise.
11899 (avx_vpermilp_parallel): Likewise.
11900 (inline_memory_move_cost): Likewise.
11901 (ix86_tieable_integer_mode_p): Likewise.
11902 (x86_maybe_negate_const_int): Likewise.
11903 (ix86_expand_vector_init_duplicate): Likewise.
11904 (ix86_expand_vector_init_one_nonzero): Likewise.
11905 (ix86_expand_vector_init_one_var): Likewise.
11906 (ix86_expand_vector_init_concat): Likewise.
11907 (ix86_expand_vector_init_interleave): Likewise.
11908 (ix86_expand_vector_init_general): Likewise.
11909 (ix86_expand_vector_set): Likewise.
11910 (ix86_expand_vector_extract): Likewise.
11911 (emit_reduc_half): Likewise.
11912 (ix86_emit_i387_round): Likewise.
11913 (ix86_mangle_type): Likewise.
11914 (ix86_expand_round_sse4): Likewise.
11915 (expand_vec_perm_blend): Likewise.
11916 (canonicalize_vector_int_perm): Likewise.
11917 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11918 (expand_vec_perm_1): Likewise.
11919 (expand_vec_perm_interleave3): Likewise.
11920 (expand_vec_perm_even_odd_pack): Likewise.
11921 (expand_vec_perm_even_odd_1): Likewise.
11922 (expand_vec_perm_broadcast_1): Likewise.
11923 (ix86_vectorize_vec_perm_const_ok): Likewise.
11924 (ix86_expand_vecop_qihi): Likewise.
11925 (ix86_expand_mul_widen_hilo): Likewise.
11926 (ix86_expand_sse2_abs): Likewise.
11927 (ix86_expand_pextr): Likewise.
11928 (ix86_expand_pinsr): Likewise.
11929 (ix86_preferred_simd_mode): Likewise.
11930 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11931 * config/i386/sse.md (*andnot<mode>3): Likewise.
11932 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11933 (*<code><mode>3): Likewise.
11934 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11935 (ia64_expand_atomic_op): Likewise.
11936 (ia64_arg_type): Likewise.
11937 (ia64_mode_to_int): Likewise.
11938 (ia64_scalar_mode_supported_p): Likewise.
11939 (ia64_vector_mode_supported_p): Likewise.
11940 (expand_vec_perm_broadcast): Likewise.
11941 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11942 (iq2000_function_arg_advance): Likewise.
11943 (iq2000_function_arg): Likewise.
11944 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11945 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11946 (m68k_libcall_value): Likewise.
11947 (m68k_function_value): Likewise.
11948 (sched_attr_op_type): Likewise.
11949 * config/mcore/mcore.c (mcore_output_move): Likewise.
11950 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11952 (microblaze_function_arg): Likewise.
11953 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11954 (mips_print_operand): Likewise.
11955 (mips_mode_ok_for_mov_fmt_p): Likewise.
11956 (mips_vector_mode_supported_p): Likewise.
11957 (mips_preferred_simd_mode): Likewise.
11958 (mips_expand_vpc_loongson_even_odd): Likewise.
11959 (mips_expand_vec_unpack): Likewise.
11960 (mips_expand_vi_broadcast): Likewise.
11961 (mips_expand_vector_init): Likewise.
11962 (mips_expand_vec_reduc): Likewise.
11963 (mips_expand_msa_cmp): Likewise.
11964 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11965 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11966 (cc_flags_for_mode): Likewise.
11967 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11968 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11969 (nds32_output_casesi_pc_relative): Likewise.
11970 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11971 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11972 (nvptx_gen_unpack): Likewise.
11973 (nvptx_gen_pack): Likewise.
11974 (nvptx_gen_shuffle): Likewise.
11975 (nvptx_gen_wcast): Likewise.
11976 (nvptx_preferred_simd_mode): Likewise.
11977 * config/pa/pa.c (pa_secondary_reload): Likewise.
11978 * config/pa/predicates.md (base14_operand): Likewise.
11979 * config/powerpcspe/powerpcspe-c.c
11980 (altivec_resolve_overloaded_builtin): Likewise.
11981 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11983 (rs6000_preferred_simd_mode): Likewise.
11984 (output_vec_const_move): Likewise.
11985 (rs6000_expand_vector_extract): Likewise.
11986 (rs6000_split_vec_extract_var): Likewise.
11987 (reg_offset_addressing_ok_p): Likewise.
11988 (rs6000_legitimate_offset_address_p): Likewise.
11989 (rs6000_legitimize_address): Likewise.
11990 (rs6000_emit_set_const): Likewise.
11991 (rs6000_const_vec): Likewise.
11992 (rs6000_emit_move): Likewise.
11993 (spe_build_register_parallel): Likewise.
11994 (rs6000_darwin64_record_arg_recurse): Likewise.
11995 (swap_selector_for_mode): Likewise.
11996 (spe_init_builtins): Likewise.
11997 (paired_init_builtins): Likewise.
11998 (altivec_init_builtins): Likewise.
11999 (do_load_for_compare): Likewise.
12000 (rs6000_generate_compare): Likewise.
12001 (rs6000_expand_float128_convert): Likewise.
12002 (emit_load_locked): Likewise.
12003 (emit_store_conditional): Likewise.
12004 (rs6000_output_function_epilogue): Likewise.
12005 (rs6000_handle_altivec_attribute): Likewise.
12006 (rs6000_function_value): Likewise.
12007 (emit_fusion_gpr_load): Likewise.
12008 (emit_fusion_p9_load): Likewise.
12009 (emit_fusion_p9_store): Likewise.
12010 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12011 (fusion_gpr_mem_load): Likewise.
12012 (fusion_addis_mem_combo_load): Likewise.
12013 (fusion_addis_mem_combo_store): Likewise.
12014 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12015 (fusion_gpr_mem_load): Likewise.
12016 (fusion_addis_mem_combo_load): Likewise.
12017 (fusion_addis_mem_combo_store): Likewise.
12018 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12020 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12021 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12022 (rs6000_preferred_simd_mode): Likewise.
12023 (output_vec_const_move): Likewise.
12024 (rs6000_expand_vector_extract): Likewise.
12025 (rs6000_split_vec_extract_var): Likewise.
12026 (reg_offset_addressing_ok_p): Likewise.
12027 (rs6000_legitimate_offset_address_p): Likewise.
12028 (rs6000_legitimize_address): Likewise.
12029 (rs6000_emit_set_const): Likewise.
12030 (rs6000_const_vec): Likewise.
12031 (rs6000_emit_move): Likewise.
12032 (rs6000_darwin64_record_arg_recurse): Likewise.
12033 (swap_selector_for_mode): Likewise.
12034 (paired_init_builtins): Likewise.
12035 (altivec_init_builtins): Likewise.
12036 (rs6000_expand_float128_convert): Likewise.
12037 (emit_load_locked): Likewise.
12038 (emit_store_conditional): Likewise.
12039 (rs6000_output_function_epilogue): Likewise.
12040 (rs6000_handle_altivec_attribute): Likewise.
12041 (rs6000_function_value): Likewise.
12042 (emit_fusion_gpr_load): Likewise.
12043 (emit_fusion_p9_load): Likewise.
12044 (emit_fusion_p9_store): Likewise.
12045 * config/rx/rx.c (rx_gen_move_template): Likewise.
12046 (flags_from_mode): Likewise.
12047 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12048 (s390_slb_comparison): Likewise.
12049 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12050 (s390_vector_mode_supported_p): Likewise.
12051 (s390_cc_modes_compatible): Likewise.
12052 (s390_match_ccmode_set): Likewise.
12053 (s390_canonicalize_comparison): Likewise.
12054 (s390_emit_compare_and_swap): Likewise.
12055 (s390_branch_condition_mask): Likewise.
12056 (s390_rtx_costs): Likewise.
12057 (s390_secondary_reload): Likewise.
12058 (__SECONDARY_RELOAD_CASE): Likewise.
12059 (s390_expand_cs): Likewise.
12060 (s390_preferred_simd_mode): Likewise.
12061 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12062 * config/sh/sh.c (sh_print_operand): Likewise.
12063 (dump_table): Likewise.
12064 (sh_secondary_reload): Likewise.
12065 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12066 * config/sh/sh.md (casesi_worker_1): Likewise.
12067 (casesi_worker_2): Likewise.
12068 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12069 (fcc_comparison_operator): Likewise.
12070 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12071 (emit_soft_tfmode_cvt): Likewise.
12072 (sparc_preferred_simd_mode): Likewise.
12073 (output_cbranch): Likewise.
12074 (sparc_print_operand): Likewise.
12075 (sparc_expand_vec_perm_bmask): Likewise.
12076 (vector_init_bshuffle): Likewise.
12077 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12078 (spu_vector_mode_supported_p): Likewise.
12079 (spu_expand_insv): Likewise.
12080 (spu_emit_branch_or_set): Likewise.
12081 (spu_handle_vector_attribute): Likewise.
12082 (spu_builtin_splats): Likewise.
12083 (spu_builtin_extract): Likewise.
12084 (spu_builtin_promote): Likewise.
12085 (spu_expand_sign_extend): Likewise.
12086 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12087 (tilegx_simd_int): Likewise.
12088 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12089 (tilepro_simd_int): Likewise.
12090 * config/v850/v850.c (const_double_split): Likewise.
12091 (v850_print_operand): Likewise.
12092 (ep_memory_offset): Likewise.
12093 * config/vax/vax.c (vax_rtx_costs): Likewise.
12094 (vax_output_int_move): Likewise.
12095 (vax_output_int_add): Likewise.
12096 (vax_output_int_subtract): Likewise.
12097 * config/visium/predicates.md (visium_branch_operator): Likewise.
12098 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12099 (visium_print_operand_address): Likewise.
12100 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12101 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12102 (xtensa_expand_conditional_branch): Likewise.
12103 (xtensa_copy_incoming_a7): Likewise.
12104 (xtensa_output_literal): Likewise.
12105 * dfp.c (decimal_real_maxval): Likewise.
12106 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12109 Alan Hayward <alan.hayward@arm.com>
12110 David Sherwood <david.sherwood@arm.com>
12112 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12113 (mode_nunits_inline): Likewise.
12114 (mode_inner_inline): Likewise.
12115 (mode_unit_size_inline): Likewise.
12116 (mode_unit_precision_inline): Likewise.
12117 (emit_insn_modes_h): Likewise. Also emit a #define of the
12119 (emit_mode_wider): Add an E_ prefix to mode names.
12120 (emit_mode_complex): Likewise.
12121 (emit_mode_inner): Likewise.
12122 (emit_mode_adjustments): Likewise.
12123 (emit_mode_int_n): Likewise.
12124 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12125 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12126 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12127 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12128 (CRC32_BUILTIN, ENTRY): Likewise.
12129 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12130 (aarch64_pop_regs): Likewise.
12131 (aarch64_process_components): Likewise.
12132 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12133 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12134 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12135 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12136 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12137 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12139 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12140 (pcmpistr): Likewise.
12141 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12142 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12143 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12145 * config/rl78/rl78.c (mduc_regs): Likewise.
12146 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12147 (htm_expand_builtin): Likewise.
12148 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12149 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12150 * config/xtensa/xtensa.c (print_operand): Likewise.
12151 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12152 (NUM_MODE_VECTOR_INT): Likewise.
12153 * genoutput.c (null_operand): Likewise.
12154 (output_operand_data): Likewise.
12155 * genrecog.c (print_parameter_value): Likewise.
12156 * lra.c (debug_operand_data): Likewise.
12158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12160 * dwarf2out.c (scompare_loc_descriptor_wide)
12161 (scompare_loc_descriptor_narrow): New functions, split out from...
12162 (scompare_loc_descriptor): ...here.
12163 * expmed.c (emit_store_flag_int): New function, split out from...
12164 (emit_store_flag): ...here.
12166 2017-08-30 Richard Biener <rguenther@suse.de>
12168 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12169 (dwarf2out_early_finish): Move setting of AT_pubnames from
12170 early debug output to early finish.
12172 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12174 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12175 and -mdata-region to the assembler.
12177 2017-08-30 Richard Biener <rguenther@suse.de>
12179 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12181 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12182 (dwarf2out_early_global_decl): Only generate a DIE for the
12183 abstract origin if it doesn't already exist or is a declaration DIE.
12184 (resolve_addr): Do not add the linkage name twice when
12185 generating a stub DIE for the DW_TAG_GNU_call_site target.
12187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12189 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12190 Use machine_mode rather than int for arg1_mode.
12192 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12195 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12196 that the second argument of the built-in functions to unpack
12197 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12198 switch statement instead a lot of if statements.
12199 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12200 Allow 64-bit values to be in Altivec registers as well as
12201 traditional floating point registers.
12202 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12204 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12206 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12207 MAX_REGS_PER_ADDRESS == 1.
12209 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12211 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12212 * config/i386/i386.c (ix86_option_override_internal): Simplify
12213 setting of opts->x_flag_entry.
12215 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12216 Jakub Jelinek <jakub@redhat.com>
12217 Richard Biener <rguenther@suse.de>
12219 PR tree-optimization/81503
12220 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12221 folded constant fits in the target type; reorder tests for clarity.
12223 2017-08-29 Martin Liska <mliska@suse.cz>
12225 * passes.def: Include pass_lower_switch.
12226 * stmt.c (dump_case_nodes): Remove and move to
12227 tree-switch-conversion.
12228 (case_values_threshold): Likewise.
12229 (expand_switch_as_decision_tree_p): Likewise.
12230 (emit_case_decision_tree): Likewise.
12231 (expand_case): Likewise.
12232 (balance_case_nodes): Likewise.
12233 (node_has_low_bound): Likewise.
12234 (node_has_high_bound): Likewise.
12235 (node_is_bounded): Likewise.
12236 (emit_case_nodes): Likewise.
12237 (struct simple_case_node): New struct.
12238 (add_case_node): Remove.
12239 (emit_case_dispatch_table): Use vector instead of case_list.
12240 (reset_out_edges_aux): Remove.
12241 (compute_cases_per_edge): Likewise.
12242 (expand_case): Build list of simple_case_node.
12243 (expand_sjlj_dispatch_table): Use it.
12244 * tree-switch-conversion.c (struct case_node): Moved from
12245 stmt.c and adjusted.
12246 (emit_case_nodes): Likewise.
12247 (node_has_low_bound): Likewise.
12248 (node_has_high_bound): Likewise.
12249 (node_is_bounded): Likewise.
12250 (case_values_threshold): Likewise.
12251 (reset_out_edges_aux): Likewise.
12252 (compute_cases_per_edge): Likewise.
12253 (add_case_node): Likewise.
12254 (dump_case_nodes): Likewise.
12255 (balance_case_nodes): Likewise.
12256 (expand_switch_as_decision_tree_p): Likewise.
12257 (emit_jump): Likewise.
12258 (emit_case_decision_tree): Likewise.
12259 (try_switch_expansion): Likewise.
12260 (do_jump_if_equal): Likewise.
12261 (emit_cmp_and_jump_insns): Likewise.
12262 (fix_phi_operands_for_edge): New function.
12263 (record_phi_operand_mapping): Likewise.
12264 (class pass_lower_switch): New pass.
12265 (pass_lower_switch::execute): New function.
12266 (make_pass_lower_switch): Likewise.
12267 (conditional_probability):
12268 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12269 * tree-pass.h: Add make_pass_lower_switch.
12271 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12274 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12277 2017-08-29 Richard Biener <rguenther@suse.de>
12279 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12280 we do not add a DW_AT_inline attribute twice.
12281 (gen_subprogram_die): Remove code setting DW_AT_inline on
12282 DECL_ABSTRACT_P nodes.
12284 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12286 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12287 calls to internal functions.
12288 (gimplify_modify_expr): Likewise.
12289 * tree-call-cdce.c (use_internal_fn): Likewise.
12290 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12291 (convert_to_divmod): Set the nothrow flag.
12292 * tree-if-conv.c (predicate_mem_writes): Likewise.
12293 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12294 (vectorizable_call): Likewise.
12295 (vectorizable_store): Likewise.
12296 (vectorizable_load): Likewise.
12297 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12298 (vect_recog_mask_conversion_pattern): Likewise.
12300 2017-08-29 Martin Liska <mliska@suse.cz>
12303 * gcc.c (driver_handle_option): Add new argument.
12304 * opts-common.c (handle_option): Pass
12305 target_option_override_hook.
12306 * opts-global.c (lang_handle_option): Add new option.
12307 (set_default_handlers): Add new argument.
12308 (decode_options): Likewise.
12309 * opts.c (target_handle_option): Likewise.
12310 (common_handle_option): Call target_option_override_hook.
12311 * opts.h (struct cl_option_handler_func): Add hook for
12312 target option override.
12313 (struct cl_option_handlers): Likewise.
12314 (set_default_handlers): Add new argument.
12315 (decode_options): Likewise.
12316 (common_handle_option): Likewise.
12317 (target_handle_option): Likewise.
12318 * toplev.c (toplev::main): Pass targetm.target_option.override
12321 2017-08-29 Richard Biener <rguenther@suse.de>
12322 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12324 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12325 life to the active subtree.
12327 2017-08-28 Jeff Law <law@redhat.com>
12329 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12330 derive_equivalences.
12331 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12332 Code moved into....
12333 (edge_info::derive_equivalences): New private member function
12335 * tree-ssa-dom.c (class edge_info): Changed from a struct
12336 to a class. Add ctor/dtor, methods and data members.
12337 (edge_info::edge_info): Renamed from allocate_edge_info.
12338 Initialize additional members.
12339 (edge_info::~edge_info): New.
12340 (free_dom_edge_info): Delete the edge info.
12341 (record_edge_info): Use new class & associated member functions.
12342 Tighten forms for testing for edge equivalences.
12343 (record_temporary_equivalences): Iterate over the simple
12344 equivalences rather than assuming there's only one per edge.
12345 (cprop_into_successor_phis): Iterate over the simple
12346 equivalences rather than assuming there's only one per edge.
12347 (optimize_stmt): Use operand_equal_p rather than pointer
12348 equality for mini-DSE code.
12350 2017-08-28 Nathan Sidwell <nathan@acm.org>
12352 * gcc.c (execute): Fold SIGPIPE handling into switch
12353 statement. Adjust internal error message.
12355 2017-08-28 Richard Biener <rguenther@suse.de>
12358 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12359 Do nothing for removed DIEs.
12361 2017-08-28 Richard Biener <rguenther@suse.de>
12363 PR tree-optimization/81977
12364 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12367 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12370 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12371 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12372 using targetm.gen_mem_thread_fence.
12374 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12377 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12378 predicate to register_operand. Reorder operands.
12379 (*btr<mode>): Ditto.
12380 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12381 (*btr<mode>_mask): Ditto.
12383 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12385 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12386 * config/rs6000/xmmintrin.h: New file.
12387 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12389 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12392 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12393 parameter and_insn and return it.
12394 (recombine_lvx_pattern): Insert a copy to ensure availability of
12395 the base register of the copied masking operation at the point of
12396 the instruction replacement.
12397 (recombine_stvx_pattern): Likewise.
12399 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12401 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12402 undocumented switches.
12403 (-mpower9-dform-vector): Likewise.
12404 (-mpower9-dform): Likewise.
12405 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12406 comments to delete references to -mpower9-dform* switches.
12407 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12408 Delete reference to -mpower9-dform* switches, test for
12409 -mpower9-vector instead.
12410 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12411 (OTHER_P9_VECTOR_MASKS): Likewise.
12412 (POWERPC_MASKS): Likewise.
12413 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12414 tests against -mpower9-dform* to -mpower9-vector. Delete code
12415 that checked for -mpower9-dform* consistancy with other options.
12416 Add test for -mpower9-misc to enable other power9 switches.
12417 (rs6000_init_hard_regno_mode_ok): Likewise.
12418 (rs6000_option_override_internal): Likewise.
12419 (rs6000_emit_prologue): Likewise.
12420 (rs6000_emit_epilogue): Likewise.
12421 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12422 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12423 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12425 (emit_fusion_p9_store): Likewise.
12426 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12427 resetting these macros if the assembler does not support ISA 3.0
12429 (TARGET_P9_DFORM_VECTOR): Likewise.
12430 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12431 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12433 2017-08-25 Alan Modra <amodra@gmail.com>
12435 PR rtl-optimization/81747
12436 * cse.c (cse_extended_basic_block): Don't attempt to record
12437 equivalences for degenerate conditional branches that branch
12438 to their fall-through.
12440 2017-08-24 Martin Sebor <msebor@redhat.com>
12442 PR middle-end/81908
12443 * gimple-fold.c (size_must_be_zero_p): New function.
12444 (gimple_fold_builtin_memory_op): Call it.
12446 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12448 * config/rs6000/mm_malloc.h: New file.
12450 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12452 PR tree-optimization/81913
12453 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12454 analysis when either IVs in condition can wrap.
12456 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12458 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12459 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12461 2017-08-24 Richard Biener <rguenther@suse.de>
12464 * targhooks.c (default_target_can_inline_p): Properly
12465 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12466 is present and always compare.
12467 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12468 infer -mfpmath=sse from TARGET_SSE_P.
12469 (ix86_can_inline_p): Properly use target_option_default_node when
12470 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12472 2017-08-24 Richard Biener <rguenther@suse.de>
12475 * dwarf2out.c (output_die): Handle flag_generate_offload like
12477 (output_comp_unit): Likewise.
12478 (gen_array_type_die): Likewise.
12479 (dwarf2out_early_finish): Likewise.
12480 (note_variable_value_in_expr): Likewise.
12481 (dwarf2out_finish): Likewise. Adjust assert.
12482 * cgraphunit.c (symbol_table::compile): Move setting of
12483 flag_generate_offload earlier ...
12484 (symbol_table::finalize_compilation_unit): ... here, before
12485 early debug finalization.
12487 2017-08-24 Richard Biener <rguenther@suse.de>
12489 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12490 and ipa-fnsummary.h.
12491 (ix86_can_inline_p): When ix86_fpmath flags do not match
12492 check whether the callee uses FP math at all.
12494 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12496 PR middle-end/81931
12497 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12498 instead of TYPE_PRECISION.
12500 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12501 Alan Hayward <alan.hayward@arm.com>
12502 David Sherwood <david.sherwood@arm.com>
12504 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12506 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12508 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12510 (ix86_frame::stack_realign_allocate): New field.
12511 (struct machine_frame_state): Modify comments.
12512 (machine_frame_state::sp_realigned_fp_end): New field.
12513 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12514 layout calculation.
12515 (sp_valid_at): Add assertion to assure no attempt to access invalid
12516 offset of a realigned stack.
12517 (fp_valid_at): Likewise.
12518 (choose_baseaddr): Modify comments.
12519 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12520 ix86_expand_prologue.
12521 (ix86_expand_prologue): Modify stack realignment and allocation.
12522 (ix86_expand_epilogue): Modify comments.
12523 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12524 avx2_runtime, avx512f, and avx512f_runtime.
12526 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12528 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12529 (mstackrealign): Do not init to -1.
12530 * config/i386/i386.c (ix86_option_override_internal):
12531 Check opts_set, not opts when setting default value of
12532 opts->x_ix86_force_align_arg_pointer.
12534 2017-08-23 Richard Biener <rguenther@suse.de>
12536 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12537 lang_hooks.decl_printable_name.
12538 * print-rtl-function.c (print_rtx_function): Likewise.
12539 * tree-pretty-print.c (dump_function_header): Likewise.
12541 2017-08-23 Richard Biener <rguenther@suse.de>
12544 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12545 -g0 at compile-time.
12547 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12549 PR middle-end/19706
12550 * doc/sourcebuild.texi (Other hardware attributes):
12553 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12555 PR middle-end/19706
12556 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12557 Fix single-use check.
12559 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12561 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12563 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12565 * doc/install.texi: Modify to add more details on running selected
12568 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12570 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12571 is combined with -mabi=ms.
12572 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12575 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12577 PR tree-optimization/81488
12578 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12579 and cached_basis fields.
12580 (MAX_SPREAD): New constant.
12581 (alloc_cand_and_find_basis): Initialize new fields.
12582 (clear_visited): New function.
12583 (create_phi_basis_1): Rename from create_phi_basis, set visited
12584 and cached_basis fields.
12585 (create_phi_basis): New wrapper function.
12586 (phi_add_costs_1): Rename from phi_add_costs, add spread
12587 parameter, set visited field, short-circuit when limits reached.
12588 (phi_add_costs): New wrapper function.
12589 (record_phi_increments_1): Rename from record_phi_increments, set
12591 (record_phi_increments): New wrapper function.
12592 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12593 (phi_incr_cost): New wrapper function.
12594 (all_phi_incrs_profitable_1): Rename from
12595 all_phi_incrs_profitable, set visited field.
12596 (all_phi_incrs_profitable): New wrapper function.
12598 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12599 Alan Hayward <alan.hayward@arm.com>
12600 David Sherwood <david.sherwood@arm.com>
12602 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12603 that takes the outer and inner modes.
12604 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12605 comparison as the canonical test for a paradoxical subreg.
12606 * combine.c (simplify_set): Use paradoxical_subreg_p.
12607 (make_extraction): Likewise.
12608 (force_to_mode): Likewise.
12609 (rtx_equal_for_field_assignment_p): Likewise.
12610 (gen_lowpart_for_combine): Likewise.
12611 (simplify_comparison): Likewise.
12612 * cse.c (equiv_constant): Likewise.
12613 * expmed.c (store_bit_field_1): Likewise.
12614 * final.c (alter_subreg): Likewise.
12615 * fwprop.c (propagate_rtx): Likewise.
12616 (forward_propagate_subreg): Likewise.
12617 * ira-conflicts.c (ira_build_conflicts): Likewise.
12618 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12619 * lra-constraints.c (curr_insn_transform): Likewise.
12620 (split_reg): Likewise.
12621 * lra-eliminations.c (move_plus_up): Likewise.
12622 (lra_eliminate_regs_1): Likewise.
12623 * recog.c (general_operand): Likewise.
12624 * ree.c (combine_reaching_defs): Likewise.
12625 * reload.c (push_reload): Likewise.
12626 (find_reloads): Likewise.
12627 * reload1.c (elimination_effects): Likewise.
12628 (compute_reload_subreg_offset): Likewise.
12629 (choose_reload_regs): Likewise.
12630 * rtlanal.c (subreg_lsb_1): Likewise.
12631 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12632 (simplify_subreg): Likewise.
12633 * var-tracking.c (track_loc_p): Likewise.
12634 * emit-rtl.c (byte_lowpart_offset): Likewise.
12635 (paradoxical_subreg_p): Delete out-of-line definition.
12637 2017-08-22 Jeff Law <law@redhat.com>
12639 PR tree-optimization/81741
12640 PR tree-optimization/71947
12641 * tree-ssa-dom.c: Include tree-inline.h.
12642 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12643 equivalences if one is more expensive to compute than the other.
12644 * tree-ssa-scopedtables.h (class const_or_copies): Make
12645 record_const_or_copy_raw method private.
12646 (class avail_exprs_stack): New method simplify_binary_operation.
12647 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12648 avail_exprs_stack::simplify_binary_operation as needed.
12649 (avail_exprs_stack::simplify_binary_operation): New function.
12651 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12653 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12654 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12655 (DOT_SYMBOLS): Likewise.
12656 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12657 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12658 (RS6000_ABI_NAME): Likewise.
12659 (TARGET_CMODEL): Likewise.
12660 (TOC_SECTION_ASM_OP): Likewise.
12661 (SET_CMODEL): New macro.
12662 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12664 2017-08-22 Richard Biener <rguenther@suse.de>
12666 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12667 to allow for free-lang-data replacements similar to verify_type_variant.
12669 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12670 Alan Hayward <alan.hayward@arm.com>
12671 David Sherwood <david.sherwood@arm.com>
12673 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12674 VOIDmode for the LABEL_REF.
12676 2017-08-22 Richard Biener <rguenther@suse.de>
12678 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12680 2017-08-22 Richard Biener <rguenther@suse.de>
12682 * common.opt (feliminate-dwarf2-dups): Ignore.
12683 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12684 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12685 same_die_p_wrap, compute_section_prefix,
12686 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12687 (comdat_symbol_id, comdat_symbol_number): Likewise.
12688 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12690 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12691 (output_die): Mark unreachable path unreachable.
12692 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12693 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12694 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12695 (dwarf2out_early_finish): Likewise.
12697 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12699 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12701 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12704 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12705 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12706 (avr_attribute_table) <io, io_low, address>: Initialize
12707 .decl_required with true.
12709 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12711 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12712 undocumented debugging options.
12713 (-mvsx-scalar-double): Likewise.
12714 (-mallow-df-permute): Likewise.
12715 (-mvectorize-builtins): Likewise.
12716 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12717 (rs6000_builtin_vectorized_function): Likewise.
12718 (rs6000_builtin_md_vectorized_function): Likewise.
12719 (rs6000_opt_vars): Likewise.
12721 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12724 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12725 (*btrq_imm): Rename from *btrq.
12726 (*btcq_imm): Rename from *btcq.
12727 (btsc): New code attribute.
12728 (*<btsc><mode>): New insn pattern.
12729 (*btr<mode>): Ditto.
12730 (*<btsc><mode>_mask): New insn_and_split pattern.
12731 (*btr<mode>_mask): Ditto.
12733 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12734 Alan Hayward <alan.hayward@arm.com>
12735 David Sherwood <david.sherwood@arm.com>
12737 * function.c (pad_below): Simplify padding calculation.
12739 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12740 Alan Hayward <alan.hayward@arm.com>
12741 David Sherwood <david.sherwood@arm.com>
12743 * target.def (function_prologue): Remove frame size argument.
12744 (function_epilogue): Likewise.
12745 * doc/tm.texi: Regenerate.
12746 * final.c (final_start_function): Update call to function_prologue.
12747 (final_end_function): Update call to function_epilogue.
12748 (default_function_pro_epilogue): Remove frame size argument.
12749 * output.h (default_function_pro_epilogue): Likewise.
12750 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12751 (arm_output_function_prologue): Likewise.
12752 * config/frv/frv.c (frv_function_prologue): Likewise.
12753 (frv_function_epilogue): Likewise.
12754 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12755 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12756 (ia64_output_function_epilogue): Likewise.
12757 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12758 (m32r_output_function_epilogue): Likewise.
12759 * config/microblaze/microblaze.c (microblaze_function_prologue)
12760 (microblaze_function_epilogue): Likewise.
12761 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12762 (mips_output_function_epilogue): Likewise.
12763 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12764 (mmix_target_asm_function_epilogue): Likewise.
12765 * config/msp430/msp430.c (msp430_start_function): Likewise.
12766 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12767 (nds32_asm_function_epilogue): Likewise.
12768 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12769 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12770 (pa_output_function_epilogue): Likewise.
12771 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12772 (rs6000_output_function_epilogue): Likewise.
12773 * config/rl78/rl78.c (rl78_start_function): Likewise.
12774 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12775 (rs6000_output_function_epilogue): Likewise.
12776 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12777 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12778 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12779 (sparc_asm_function_epilogue): Likewise.
12781 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12783 * tree.h (type_has_mode_precision_p): New function.
12784 * convert.c (convert_to_integer_1): Use it.
12785 * expr.c (expand_expr_real_2): Likewise.
12786 (expand_expr_real_1): Likewise.
12787 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12788 * match.pd: Likewise.
12789 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12790 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12791 * tree-tailcall.c (process_assignment): Likewise.
12792 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12793 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12794 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12795 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12796 (vectorizable_assignment): Likewise.
12797 (vectorizable_shift): Likewise.
12798 (vectorizable_operation): Likewise.
12799 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12801 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12803 * match.pd: Add pow (C, x) simplification.
12805 2017-08-21 Richard Biener <rguenther@suse.de>
12807 PR tree-optimization/81900
12808 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12809 for blocks with abnormal predecessors.
12810 (compute_antic): Do not set visited flag prematurely.
12812 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12815 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12817 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12819 * stor-layout.h (vector_type_mode): Move to...
12820 * tree.h (vector_type_mode): ...here.
12821 * stor-layout.c (vector_type_mode): Move to...
12822 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12824 2017-08-21 Richard Biener <rguenther@suse.de>
12826 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12827 register_external_die hooks.
12828 (debug_false_tree_charstarstar_uhwistar): Declare.
12829 (debug_nothing_tree_charstar_uhwi): Likewise.
12830 * debug.c (do_nothing_debug_hooks): Adjust.
12831 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12832 (debug_nothing_tree_charstar_uhwi): Likewise.
12833 * dbxout.c (dbx_debug_hooks): Adjust.
12834 (xcoff_debug_hooks): Likewise.
12835 * sdbout.c (sdb_debug_hooks): Likewise.
12836 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12837 * dwarf2out.c (macinfo_label_base): New global.
12838 (dwarf2out_register_external_die): New function for the
12839 register_external_die hook.
12840 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12841 (dwarf2_debug_hooks): Use them.
12842 (dwarf2_lineno_debug_hooks): Adjust.
12843 (struct die_struct): Add with_offset flag.
12844 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12845 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12846 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12847 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12848 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12849 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12850 defining section names for the early LTO debug variants.
12851 (reset_indirect_string): New helper.
12852 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12853 (print_dw_val): Add support for offsetted symbol references.
12854 (get_ultimate_context): Split out from is_cxx.
12855 (is_cxx): Use get_ultimate_context.
12856 (is_fortran): Add decl overload.
12857 (compute_comp_unit_symbol): Split out worker from
12858 compute_section_prefix.
12859 (compute_section_prefix): Call compute_comp_unit_symbol and
12860 set comdat_type_p here.
12861 (output_die): Skip DIE symbol output for the LTO added one.
12862 Handle DIE symbol references with offset.
12863 (output_comp_unit): Guard section name mangling properly.
12864 For LTO debug sections emit a symbol at the section beginning
12865 which we use to refer to its DIEs.
12866 (add_abstract_origin_attribute): For DIEs registered via
12867 dwarf2out_register_external_die directly refer to the early
12868 DIE rather than indirectly through the shadow one we created.
12869 Remove obsolete call to dwarf2out_abstract_function for
12870 non-function/block origins.
12871 (gen_array_type_die): When generating early LTO debug do
12872 not emit DW_AT_string_length.
12873 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12874 late when in LTO. As suggested place a gcc_unreachable for
12875 the DECL_ABSTRACT_P case.
12876 (gen_subprogram_die): Avoid another specification DIE
12877 for early built declarations/definitions for the late LTO case.
12878 (gen_variable_die): Add type references for late duplicated VLA dies
12880 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12881 we have the abstract instance already.
12882 (process_scope_var): Adjust decl DIE contexts in LTO which
12883 first puts them in limbo.
12884 (gen_decl_die): Do not generate type DIEs late apart from
12885 types for VLAs or for decls we do not yet have a DIE. Do not
12886 call dwarf2out_abstract_function late.
12887 (dwarf2out_early_global_decl): Make sure to create DIEs
12888 for abstract instances of a decl first.
12889 (dwarf2out_late_global_decl): Adjust comment.
12890 (output_macinfo_op): With multiple macro sections use
12891 macinfo_label_base to distinguish labels.
12892 (output_macinfo): Likewise. Update macinfo_label_base.
12893 Pass in the line info label.
12894 (note_variable_value_in_expr): When generating LTO resolve
12895 all variable values here by generating DIEs as needed.
12896 (init_sections_and_labels): Add early LTO debug flag parameter
12897 and generate different sections and names if set. Add generation
12898 counter for the labels so we can have multiple of them.
12899 (reset_dies): Helper to allow DIEs to be output multiple times.
12900 (dwarf2out_finish): When outputting DIEs to the fat part of an
12901 LTO object first reset DIEs.
12902 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12903 (modified_type_die): Check for decl_ultimate_origin being self
12905 (gen_type_die_with_usage): Likewise.
12906 (gen_typedef_die): Allow decl_ultimate_origin being self.
12907 (set_decl_abstract_flags): Remove.
12908 (set_block_abstract_flags): Likewise.
12909 (dwarf2out_abstract_function): Treat the early generated DIEs
12910 as the abstract copy and only add DW_AT_inline and
12911 DW_AT_artificial here and call set_decl_origin_self.
12912 If the DIE has an abstract origin don't do anything.
12913 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12914 if we have none yet (Go fails to build one, PR78628).
12915 (variably_modified_type_p): Prevent endless recursion for Ada
12916 cyclic pointer types.
12917 * lto-streamer-in.c: Include debug.h.
12918 (dref_queue): New global.
12919 (lto_read_tree_1): Stream in DIE references.
12920 (lto_input_tree): Register DIE references.
12921 (input_function): Stream DECL_DEBUG_ARGS.
12922 * lto-streamer-out.c: Include debug.h.
12923 (lto_write_tree_1): Output DIE references.
12924 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12925 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12926 (output_function): Stream DECL_DEBUG_ARGS.
12927 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12928 Stream DECL_ABSTRACT_ORIGIN.
12929 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12930 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12931 DECL_CONTEXT for file-scope decls.
12932 * lto-streamer.h (struct dref_entry): Declare.
12933 (dref_queue): Likewise.
12934 * cfgexpand.c (pass_expand::execute): Do not call the
12935 outlining_inline_function hook here.
12936 * lto-wrapper.c (debug_obj): New global.
12937 (tool_cleanup): Unlink it if required.
12938 (debug_objcopy): New function.
12939 (run_gcc): Handle early debug sections in the IL files by
12940 extracting them to separate files, partially linkin them and
12941 feeding the result back as result to the linker.
12942 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12943 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12944 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12945 sections into a separate segment.
12946 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12948 (darwin_asm_dwarf_section): Likewise.
12949 (darwin_asm_output_dwarf_offset): Likewise.
12950 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12952 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12953 Alan Hayward <alan.hayward@arm.com>
12954 David Sherwood <david.sherwood@arm.com>
12956 * read-md.h (md_reader::record_potential_iterator_use): Replace
12957 pointer argument with an rtx and an index.
12958 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12959 (apply_mode_iterator): Likewise.
12960 (apply_code_iterator): Likewise.
12961 (apply_int_iterator): Likewise.
12962 (apply_subst_iterator): Likewise.
12963 (record_iterator_use): Likewise.
12964 (record_attribute_use): Likewise.
12965 (md_reader::record_potential_iterator_use): Likewise. Update calls
12966 to record_iterator_use and apply_iterator.
12967 (iterator_use): Replace ptr with x and index.
12968 (attribute_use): Likewise.
12969 (apply_attribute_uses): Update calls to apply_iterator.
12970 (apply_iterators): Likewise. Update initialization of iterator_use.
12971 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12972 and record_potential_iterator_use.
12973 (rtx_reader::read_rtx_operand): Likewise.
12975 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12976 Alan Hayward <alan.hayward@arm.com>
12977 David Sherwood <david.sherwood@arm.com>
12979 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12982 2017-08-21 Richard Biener <rguenther@suse.de>
12984 PR middle-end/81884
12985 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12986 at struct end conservatively when comparing common bases.
12988 2017-08-21 Richard Biener <rguenther@suse.de>
12990 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12991 (mem_ref_in_stmt): Remove.
12992 (determine_max_movement): Use ref index to get at the reference.
12993 (invariantness_dom_walker::before_dom_children): Deal with
12994 lim data already initialized.
12995 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12997 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12999 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13000 (TARGET_ISA_ROUND): Ditto.
13001 (TARGET_ROUND): Ditto.
13002 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13003 * config/i386/i386.md: Ditto.
13004 * config/i386/sse.md: Ditto.
13005 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13006 with OPTION_MASK_ISA_SSE4_1.
13008 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13011 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13012 __builtin_ia32_lzcnt_u16.
13014 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13017 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13018 (rs6000_set_current_function): Rewrite function to use it.
13020 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13023 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13024 and TYPE_WARN_IF_NOT_ALIGN.
13025 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13026 (handle_warn_if_not_align): New.
13027 (place_union_field): Call handle_warn_if_not_align.
13028 (place_field): Call handle_warn_if_not_align.
13029 Copy TYPE_WARN_IF_NOT_ALIGN.
13030 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13031 (layout_type): Likewise.
13032 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13034 (tree_decl_common): Add warn_if_not_align.
13035 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13036 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13037 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13038 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13039 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13040 * doc/extend.texi: Document warn_if_not_aligned attribute.
13041 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13043 2017-08-17 Martin Liska <mliska@suse.cz>
13046 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13047 (get_data_dependence): Use it as pointer type.
13048 (distribute_loop): Likewise.
13050 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13052 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13053 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13054 (p8_vmrgow_<mode>_direct): New define_insn.
13055 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13056 handle endianness for vmrgew and vmrgow permute patterns.
13058 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13060 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13061 * config/rs6000/rs6000-cpus.def: Remove comment.
13062 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13063 (POWERPC_MASKS): Likewise.
13064 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13065 use of TARGET_VSX_TIMODE.
13066 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13067 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13068 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13069 (rs6000_option_override_internal): Remove dead code.
13070 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13071 (rs6000_legitimize_reload_address): Likewise.
13072 (rs6000_legitimate_address_p): Likewise.
13073 (rs6000_opt_masks): Delete "vsx-timode".
13074 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13075 from function comment.
13076 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13077 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13078 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13080 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13081 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13082 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13083 (VSX_TI): Likewise.
13085 (define_peephole2): Likewise.
13087 2017-08-17 Martin Sebor <msebor@redhat.com>
13090 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13091 past the end of an array.
13092 (test_pp_format): Add test cases.
13094 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13096 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13097 missing ECF_NOTHROW flags.
13099 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13102 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13103 operands residing in integer registers.
13104 (*vsx_le_perm_load_<mode>): Likewise.
13105 (*vsx_le_perm_store_<mode>): Likewise.
13106 (define_peephole2): Add peepholes to optimize the above.
13108 2017-08-17 Marek Polacek <polacek@redhat.com>
13110 PR middle-end/81814
13111 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13112 to mimic what shorten_compare did. Change the return type to bool.
13113 (fold_cond_expr_with_comparison): Update call to
13114 operand_equal_for_comparison_p.
13115 (fold_ternary_loc): Likewise.
13117 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13119 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13121 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13123 2017-08-17 Richard Biener <rguenther@suse.de>
13125 * tree-ssa-structalias.c (solve_graph): When propagating
13126 to successors update the graphs succ edges and avoid duplicate work.
13128 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13131 * config/i386/i386.c (ix86_option_override_internal): Save target
13132 specific options after ix86_stack_protector_guard_reg was changed.
13134 2017-08-17 Richard Biener <rguenther@suse.de>
13136 PR tree-optimization/81827
13137 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13138 (new_var_info): Initialize it conservatively.
13139 (get_call_vi): Mark register vars.
13140 (new_scalar_tmp_constraint_exp): Likewise.
13141 (handle_rhs_call): Likewise.
13142 (handle_const_call): Likewise.
13143 (create_function_info_for): Likewise.
13144 (solve_constraints): Sort varinfos to separate register from
13145 non-register vars to pack points-to solution bitmaps during
13148 2017-08-17 Marek Polacek <polacek@redhat.com>
13150 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13152 2017-08-17 Richard Biener <rguenther@suse.de>
13154 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13155 to true when overflow is undefined and we saturated the result.
13157 2017-08-17 Alan Modra <amodra@gmail.com>
13160 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13161 Don't use store multiple if only one reg needs saving.
13162 (interesting_frame_related_regno): New function.
13163 (rs6000_frame_related): Don't emit frame info for regs that
13165 (rs6000_emit_epilogue): Likewise.
13167 2017-08-16 Nathan Sidwell <nathan@acm.org>
13169 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13170 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13171 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13172 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13173 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13174 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13175 (verify_type): Adjust for TYPE_BINFO move.
13176 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13177 process TYPE_BINFO directly.
13178 (hash_tree): Likewise.
13179 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13181 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13184 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13186 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13188 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13191 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13192 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13193 Add ix86_binary_operator_ok to insn constraint.
13194 (*iordi_1_bts): Ditto.
13195 (*xordi_1_btc): Ditto.
13196 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13197 Update corresponding peephole2 pattern.
13201 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13203 PR tree-optimization/81832
13204 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13205 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13207 2017-08-16 Marek Polacek <polacek@redhat.com>
13210 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13211 regarding size_zero_node.
13213 2017-08-16 Martin Liska <mliska@suse.cz>
13216 * config.gcc: Respect previously set extra_objs in case
13219 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13221 PR tree-optimization/81835
13222 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13223 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13224 not depend on the phi.
13226 2017-08-16 Alan Modra <amodra@gmail.com>
13228 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13231 2017-08-16 Alan Modra <amodra@gmail.com>
13233 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13234 (save_reg_p): ..into this. Update all callers.
13235 (first_reg_to_save): Simplify.
13237 2017-08-16 Alan Modra <amodra@gmail.com>
13239 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13242 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13246 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13247 constant count if that count is less than 32.
13249 2017-08-15 Nathan Sidwell <nathan@acm.org>
13251 * gcc.c (execute): Emit friendlier message if inferior is killed
13252 by an external cause.
13254 2017-08-15 Richard Biener <rguenther@suse.de>
13256 PR tree-optimization/81790
13257 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13258 CONSTRUCTORs from simplifying and VN.
13260 2017-08-14 Martin Sebor <msebor@redhat.com>
13262 * builtin-attrs.def: Add comments.
13264 2017-08-14 Martin Sebor <msebor@redhat.com>
13267 * doc/extend.texi (attribute nonstring): Document new attribute.
13269 2017-08-14 Martin Sebor <msebor@redhat.com>
13272 * tree-diagnostic.c (default_tree_printer): Handle %G.
13273 * gimple-pretty-print.h (percent_G_format): Declare new function.
13274 * gimple-pretty-print.c (percent_G_format): Define.
13275 * tree-pretty-print.c (percent_K_format): Add argument.
13277 2017-08-14 Martin Sebor <msebor@redhat.com>
13279 PR translation/79998
13280 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13281 Remove a stray space.
13283 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13286 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13287 (*iordi_1_bts): Ditto.
13288 (*xordi_1_btc): Ditto.
13290 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13293 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13294 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13296 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13297 quoted strings, and make more translator-friendly.
13298 (darwin_rs6000_override_options): Likewise.
13299 (rs6000_option_override_internal): Likewise.
13300 (rs6000_return_in_memory): Fix overlong line.
13301 (init_cmulative_args): Use quoted strings, and make more
13302 translator-friendly.
13303 (rs6000_pass_by_reference): Fix overlong line.
13304 (def_builtin): Use quoted strings.
13305 (altivec_expand_predicate_builtin): Use quoted strings, and make
13306 more translator-friendly.
13307 (htm_expand_builtin): Use quoted strings.
13308 (cpu_expand_builtin): Use quoted strings, and make more
13309 translator-friendly.
13310 (altivec_expand_builtin): Likewise.
13311 (paired_expand_predicate_builtin): Likewise.
13312 (rs6000_invalid_builtin): Likewise.
13313 (builtin_function_type): Use quoted strings.
13314 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13315 more translator-friendly.
13316 (rs6000_trampoline_init): Likewise.
13317 (rs6000_handle_altivec_attribute): Likewise.
13318 (rs6000_inner_target_options): Use quoted strings.
13319 (rs6000_disable_incompatible_switches): Likewise.
13320 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13321 strings, and make more translator-friendly.
13322 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13324 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13326 PR tree-optimization/81799
13327 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13328 cond_expr to simple gimple operand.
13330 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13332 PR middle-end/46932
13333 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13335 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13339 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13340 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13341 TARGET_GASISR_PROLOGUES.
13342 * config/avr/avr.c (avr_option_override): Same.
13343 (avr_pass_pre_proep::execute): Same.
13345 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13348 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13349 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13351 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13353 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13354 define_insn_and_split. Split to a memory load from 0 in
13355 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13356 using PTR mode iterator.
13357 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13358 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13359 (*add_tp_<mode>): Redefine as define_insn_and_split.
13360 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13361 address space. Merge with *add_tp_x32 using PTR mode iterator.
13362 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13363 Split to an add with a memory load from 0 in
13364 DEFAULT_TLS_SEG_REG address space.
13366 2017-08-12 Andrew Pinski <apinski@cavium.com>
13368 * config/aarch64/aarch64-option-extensions.def (rdma):
13369 Fix feature string to what Linux prints out in /proc/cpuinfo.
13371 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13374 * dwarf2out.c (modified_type_die): For C typedef types that have
13375 an ultimate origin, process the ultimate origin instead of the
13377 (gen_typedef_die): Assert that input DECLs have no ultimate
13379 (gen_type_die_with_usage): For typedef variants that have an
13380 ultimate origin, just call gen_decl_die on the original DECL.
13381 (process_scope_var): Avoid creating DIEs for local typedefs and
13382 concrete static variables.
13384 2017-08-12 Alan Modra <amodra@gmail.com>
13388 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13389 match gnu-user.h startfile.
13390 (ENDFILE_LINUX_SPEC): Similarly.
13392 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13395 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13397 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13399 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13400 * config/aarch64/aarch64.md (mov<mode>): Change.
13401 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13402 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13403 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13405 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13407 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13408 for storage order barriers.
13410 2017-08-11 Martin Liska <mliska@suse.cz>
13413 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13414 variables of void type.
13416 2017-08-11 Martin Liska <mliska@suse.cz>
13418 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13419 TARGET_SUPPORTS_ALIASES.
13420 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13421 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13422 (optimize_weakref): Likewise.
13423 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13424 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13425 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13427 2017-08-11 Martin Liska <mliska@suse.cz>
13430 * config/i386/i386.c (make_resolver_func): Do complete
13431 refactoring of the function.
13433 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13436 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13437 * config/i386/i386.c (ix86_stack_protect_guard): Use
13438 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13439 * doc/invoke.texi (x86 Options): Document
13440 -mstack-protector-guard-symbol= option.
13442 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13444 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13445 * config/i386/i386.c (ix86_split_stack_guard): New function.
13446 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13447 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13448 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13449 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13450 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13451 (split_stack_space_check): Call ix86_split_stack_guard.
13453 2017-08-10 Martin Sebor <msebor@redhat.com>
13455 * print-tree.c (print_node): Print location using the established
13457 Replace spaces with colons.
13458 (debug_raw, debug): Ditto.
13460 2017-08-10 Martin Sebor <msebor@redhat.com>
13463 * pretty-print.c (pp_format): Correct the handling of %s precision.
13465 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13468 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13470 (ix86_finalize_stack_frame_flags): This. Also clear
13471 frame_pointer_needed if -fno-omit-frame-pointer is used without
13473 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13474 with ix86_finalize_stack_frame_flags.
13475 (ix86_expand_epilogue): Likewise.
13476 (ix86_expand_split_stack_prologue): Likewise.
13477 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13479 2017-08-10 Martin Liska <mliska@suse.cz>
13482 * c-attribs.c (handle_target_attribute):
13483 Report warning for an empty string argument of target attribute.
13485 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13488 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13490 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13491 or DECL_NONLOCAL labels.
13492 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13493 or DECL_NONLOCAL labels here.
13495 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13497 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13498 to indicate when early gimple folding has been disabled.
13499 (rs6000_gimple_fold_builtin): Add debug content.
13500 (rs6000_invalid_builtin): Fix whitespace.
13501 (rs6000_expand_builtin): Fix whitespace.
13502 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13504 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13507 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13508 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13510 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13512 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13513 (qdf24xx): Likewise.
13514 * config/aarch64/aarch64.md: Include falkor.md.
13515 * config/aarch64/falkor.md: New.
13517 2017-08-09 Marek Polacek <polacek@redhat.com>
13520 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13521 * diagnostic.c (emit_diagnostic): Add a comment.
13522 (emit_diagnostic_valist): New function.
13524 2017-08-09 Marek Polacek <polacek@redhat.com>
13527 * input.c (make_location): New overload.
13528 * input.h (make_location): Declare.
13530 2017-08-08 Alan Modra <amodra@gmail.com>
13531 H.J. Lu <hongjiu.lu@intel.com>
13534 * gcc.c (NO_PIE_SPEC): Delete.
13535 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13537 (LINK_PIE_SPEC): ..to here.
13538 (LINK_COMMAND_SPEC): Support -no-pie.
13539 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13540 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13542 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13543 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13544 (ENDFILE_CRTEND_SPEC): Similarly.
13546 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13549 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13550 (mstack-protector-guard-offset=): Ditto.
13551 * config/i386/i386.c (ix86_option_override): Handle
13552 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13554 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13555 ix86_stack_protect_guard_offset variables.
13556 (TARGET_STACK_PROTECT_GUARD): Always define.
13557 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13558 and -mstack-protector-guard-offset= options.
13560 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13562 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13563 boundary case for the last candidate.
13565 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13567 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13568 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13570 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13572 PR middle-end/19706
13573 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13574 * config/aarch64/aarch64-builtins.c
13575 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13576 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13577 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13579 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13580 Andrew Pinski <pinskia@gmail.com>
13582 PR middle-end/19706
13583 * internal-fn.def (XORSIGN): New.
13584 * optabs.def (xorsign_optab): New.
13585 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13586 (convert_expand_mult_copysign): New.
13587 (pass_optimize_widening_mul::execute): Call
13588 convert_expand_mult_copysign.
13590 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13592 PR tree-optimization/81354
13593 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13594 Insert on edges rather than explicitly creating landing pads.
13595 (analyze_candidates_and_replace): Commit edge inserts.
13597 2017-08-08 Richard Biener <rguenther@suse.de>
13599 PR middle-end/81719
13600 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13601 (expand_simple_operations): Also look through ADDR_EXPRs with
13602 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13604 2017-08-08 Richard Biener <rguenther@suse.de>
13606 PR tree-optimization/81723
13607 * tree-vect-slp.c (struct bst_traits): New hash traits.
13608 (bst_fail): New global.
13609 (vect_build_slp_tree_2): New worker, split out from ...
13610 (vect_build_slp_tree): ... this now wrapping it with using
13611 bst_fail set to cache SLP tree build fails. Properly handle
13613 (vect_analyze_slp_instance): Allocate and free bst_fail.
13615 2017-08-08 Martin Liska <mliska@suse.cz>
13618 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13619 LABEL_DECLs that can be from a different function.
13621 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13623 PR tree-optimization/81744
13624 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13625 loop's number of iterations.
13627 2017-08-08 Martin Liska <mliska@suse.cz>
13629 * asan.c: Include header files.
13630 * attribs.c (build_decl_attribute_variant): New function moved
13632 (build_type_attribute_qual_variant): Likewise.
13633 (cmp_attrib_identifiers): Likewise.
13634 (simple_cst_list_equal): Likewise.
13635 (omp_declare_simd_clauses_equal): Likewise.
13636 (attribute_value_equal): Likewise.
13637 (comp_type_attributes): Likewise.
13638 (build_type_attribute_variant): Likewise.
13639 (lookup_ident_attribute): Likewise.
13640 (remove_attribute): Likewise.
13641 (merge_attributes): Likewise.
13642 (merge_type_attributes): Likewise.
13643 (merge_decl_attributes): Likewise.
13644 (merge_dllimport_decl_attributes): Likewise.
13645 (handle_dll_attribute): Likewise.
13646 (attribute_list_equal): Likewise.
13647 (attribute_list_contained): Likewise.
13648 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13649 (lookup_attribute_by_prefix): Likewise.
13650 * bb-reorder.c: Include header files.
13651 * builtins.c: Likewise.
13652 * calls.c: Likewise.
13653 * cfgexpand.c: Likewise.
13654 * cgraph.c: Likewise.
13655 * cgraphunit.c: Likewise.
13656 * convert.c: Likewise.
13657 * dwarf2out.c: Likewise.
13658 * final.c: Likewise.
13659 * fold-const.c: Likewise.
13660 * function.c: Likewise.
13661 * gimple-expr.c: Likewise.
13662 * gimple-fold.c: Likewise.
13663 * gimple-pretty-print.c: Likewise.
13664 * gimple.c: Likewise.
13665 * gimplify.c: Likewise.
13666 * hsa-common.c: Likewise.
13667 * hsa-gen.c: Likewise.
13668 * internal-fn.c: Likewise.
13669 * ipa-chkp.c: Likewise.
13670 * ipa-cp.c: Likewise.
13671 * ipa-devirt.c: Likewise.
13672 * ipa-fnsummary.c: Likewise.
13673 * ipa-inline.c: Likewise.
13674 * ipa-visibility.c: Likewise.
13676 * lto-cgraph.c: Likewise.
13677 * omp-expand.c: Likewise.
13678 * omp-general.c: Likewise.
13679 * omp-low.c: Likewise.
13680 * omp-offload.c: Likewise.
13681 * omp-simd-clone.c: Likewise.
13682 * opts-global.c: Likewise.
13683 * passes.c: Likewise.
13684 * predict.c: Likewise.
13685 * sancov.c: Likewise.
13686 * sanopt.c: Likewise.
13687 * symtab.c: Likewise.
13688 * toplev.c: Likewise.
13689 * trans-mem.c: Likewise.
13690 * tree-chkp.c: Likewise.
13691 * tree-eh.c: Likewise.
13692 * tree-into-ssa.c: Likewise.
13693 * tree-object-size.c: Likewise.
13694 * tree-parloops.c: Likewise.
13695 * tree-profile.c: Likewise.
13696 * tree-ssa-ccp.c: Likewise.
13697 * tree-ssa-live.c: Likewise.
13698 * tree-ssa-loop.c: Likewise.
13699 * tree-ssa-sccvn.c: Likewise.
13700 * tree-ssa-structalias.c: Likewise.
13701 * tree-ssa.c: Likewise.
13702 * tree-streamer-in.c: Likewise.
13703 * tree-vectorizer.c: Likewise.
13704 * tree-vrp.c: Likewise.
13705 * tsan.c: Likewise.
13706 * ubsan.c: Likewise.
13707 * varasm.c: Likewise.
13708 * varpool.c: Likewise.
13709 * tree.c: Remove functions moved to attribs.[ch].
13710 * tree.h: Likewise.
13711 * config/aarch64/aarch64.c: Add attrs.h header file.
13712 * config/alpha/alpha.c: Likewise.
13713 * config/arc/arc.c: Likewise.
13714 * config/arm/arm.c: Likewise.
13715 * config/avr/avr.c: Likewise.
13716 * config/bfin/bfin.c: Likewise.
13717 * config/c6x/c6x.c: Likewise.
13718 * config/cr16/cr16.c: Likewise.
13719 * config/cris/cris.c: Likewise.
13720 * config/darwin.c: Likewise.
13721 * config/epiphany/epiphany.c: Likewise.
13722 * config/fr30/fr30.c: Likewise.
13723 * config/frv/frv.c: Likewise.
13724 * config/ft32/ft32.c: Likewise.
13725 * config/h8300/h8300.c: Likewise.
13726 * config/i386/winnt.c: Likewise.
13727 * config/ia64/ia64.c: Likewise.
13728 * config/iq2000/iq2000.c: Likewise.
13729 * config/lm32/lm32.c: Likewise.
13730 * config/m32c/m32c.c: Likewise.
13731 * config/m32r/m32r.c: Likewise.
13732 * config/m68k/m68k.c: Likewise.
13733 * config/mcore/mcore.c: Likewise.
13734 * config/microblaze/microblaze.c: Likewise.
13735 * config/mips/mips.c: Likewise.
13736 * config/mmix/mmix.c: Likewise.
13737 * config/mn10300/mn10300.c: Likewise.
13738 * config/moxie/moxie.c: Likewise.
13739 * config/msp430/msp430.c: Likewise.
13740 * config/nds32/nds32-isr.c: Likewise.
13741 * config/nds32/nds32.c: Likewise.
13742 * config/nios2/nios2.c: Likewise.
13743 * config/nvptx/nvptx.c: Likewise.
13744 * config/pa/pa.c: Likewise.
13745 * config/pdp11/pdp11.c: Likewise.
13746 * config/powerpcspe/powerpcspe.c: Likewise.
13747 * config/riscv/riscv.c: Likewise.
13748 * config/rl78/rl78.c: Likewise.
13749 * config/rx/rx.c: Likewise.
13750 * config/s390/s390.c: Likewise.
13751 * config/sh/sh.c: Likewise.
13752 * config/sol2.c: Likewise.
13753 * config/sparc/sparc.c: Likewise.
13754 * config/spu/spu.c: Likewise.
13755 * config/stormy16/stormy16.c: Likewise.
13756 * config/tilegx/tilegx.c: Likewise.
13757 * config/tilepro/tilepro.c: Likewise.
13758 * config/v850/v850.c: Likewise.
13759 * config/vax/vax.c: Likewise.
13760 * config/visium/visium.c: Likewise.
13761 * config/xtensa/xtensa.c: Likewise.
13763 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13766 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13767 constraints since the -mupper-regs-* switches have been
13769 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13770 into a vector from a double word element that was extracted from
13771 another vector, and eliminate extra XXPERMDI instructions.
13772 (vsx_concat_<mode>_2): Likewise.
13773 (vsx_concat_<mode>_3): Likewise.
13774 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13775 concat to allow optimizing inserts from previous extracts.
13777 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13779 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13780 memory reference to a SSP offset in TLS address space.
13781 (ix86_print_operand) <case '@'>: Remove.
13782 (ix86_print_operand_punct_valid_p): Remove '@' code.
13783 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13784 UNSPEC_SP_TLS_TEST.
13785 (stack_tls_protect_set_<mode>): Remove.
13786 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13787 (stack_tls_protect_test_<mode>): Remove.
13788 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13790 2017-08-07 Olivier Hainque <hainque@adacore.com>
13793 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13795 2017-08-07 Douglas Rupp <rupp@adacore.com>
13797 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13798 variable was referenced as multidir in command.
13800 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13803 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13806 2017-08-07 Martin Liska <mliska@suse.cz>
13808 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13809 * config/rl78/rl78.c: Add include of attribs.h.
13810 * config/sh/sh.c: Likewise.
13811 * config/v850/v850.c: Likewise.
13813 2017-08-07 Tom de Vries <tom@codesourcery.com>
13815 PR middle-end/78266
13816 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13818 2017-08-07 Martin Liska <mliska@suse.cz>
13820 * config/mips/mips.c: Include attribs.h.
13822 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13825 * doc/invoke.texi: Document change in behvaior for -Ofast for
13828 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13830 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13832 (aarch64_pop_regs): Likewise.
13833 (aarch64_gen_load_pair): Likewise.
13834 (aarch64_save_callee_saves): Likewise.
13835 (aarch64_restore_callee_saves): Likewise.
13837 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13839 * config/i386/i386.c: Revert the last change.
13841 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13844 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13846 (ix86_finalize_stack_frame_flags): This. Also clear
13847 frame_pointer_needed if -fno-omit-frame-pointer is used without
13849 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13850 with ix86_finalize_stack_frame_flags.
13851 (ix86_expand_epilogue): Likewise.
13852 (ix86_expand_split_stack_prologue): Likewise.
13854 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13857 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13858 to P_AES for Westmere.
13860 2017-08-07 Jonathan Yong <10walls@gmail.com>
13862 * config/i386/mingw.opt (fset-stack-executable): Removed.
13863 * config/i386/cygming.opt (fset-stack-executable): Moved
13865 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13867 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13869 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13871 2017-08-07 Marek Polacek <polacek@redhat.com>
13873 PR middle-end/81737
13874 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13876 2017-08-07 Martin Liska <mliska@suse.cz>
13878 * attribs.h (canonicalize_attr_name): New function.
13879 (cmp_attribs): Move from c-format.c and adjusted.
13880 (is_attribute_p): Moved from tree.h.
13881 * tree-inline.c: Add new includes.
13882 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13883 (private_is_attribute_p): Remove.
13884 (private_lookup_attribute): Likewise.
13885 (private_lookup_attribute_by_prefix): Simplify.
13886 (remove_attribute): Use is_attribute_p.
13887 * tree.h: Remove removed declarations.
13889 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13891 PR middle-end/81698
13892 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13893 instead of computing it in the function. Formatting fix.
13894 (expand_case): Don't rely on default_edge being the first edge,
13895 clear it if removing it, pass default_edge to
13896 emit_case_dispatch_table.
13897 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13900 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13902 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13903 insn in the function, emit NOP after the insn.
13905 2017-08-06 Tom de Vries <tom@codesourcery.com>
13907 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13910 2017-08-06 Tom de Vries <tom@codesourcery.com>
13912 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13915 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13917 PR tree-optimization/57371
13918 * match.pd: New pattern.
13920 2017-08-04 Marek Polacek <polacek@redhat.com>
13922 PR middle-end/81695
13923 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13924 perform the computation in offset_int.
13926 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13928 PR tree-optimization/81136
13929 * tree-vectorizer.h: Include tree-hash-traits.h.
13930 (vec_base_alignments): New typedef.
13931 (vec_info): Add a base_alignments field.
13932 (vect_record_base_alignments): Declare.
13933 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13935 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13936 (create_data_ref): Add an is_conditional_in_stmt argument.
13937 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13938 the is_conditional_in_stmt field.
13939 (data_ref_loc): Add an is_conditional_in_stmt field.
13940 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13941 (find_data_references_in_stmt): Update call to create_data_ref.
13942 (graphite_find_data_references_in_stmt): Likewise.
13943 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13944 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13945 (vect_record_base_alignment): New function.
13946 (vect_record_base_alignments): Likewise.
13947 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13948 for nested statements even if we fail to compute a misalignment.
13949 Use pooled base alignments for unconditional references.
13950 (vect_find_same_alignment_drs): Compare base addresses instead
13952 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13953 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13955 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13957 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13958 Add an explicit name for the enum. Use auto_vec for slp_instances
13959 and grouped_stores.
13960 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13962 (_bb_vec_info): Add a constructor and destructor.
13963 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13964 (destroy_loop_vec_info): Delete.
13965 (vect_destroy_datarefs): Likewise.
13966 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13967 (vec_info::vec_info): New function.
13968 (vec_info::~vec_info): Likewise.
13969 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13970 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13971 destroy_loop_vec_info.
13972 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13973 (_loop_vec_info::_loop_vec_info): ...this.
13974 (destroy_loop_vec_info): Replace with...
13975 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13976 the stmt_vec_infos. Leave handling of vec_info information to its
13977 destructor. Remove explicit vector releases.
13978 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13979 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13980 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13981 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13982 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13983 (destroy_bb_vec_info): Replace with...
13984 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13985 information to its destructor.
13986 (vect_slp_analyze_bb_1): Use new and delete instead of
13987 new_bb_vec_info and destroy_bb_vec_info.
13988 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13991 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13993 * tree-data-ref.h (subscript): Add access_fn field.
13994 (data_dependence_relation): Add could_be_independent_p.
13995 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13996 (same_access_functions): Move to tree-data-ref.c.
13997 * tree-data-ref.c (ref_contains_union_access_p): New function.
13998 (access_fn_component_p): Likewise.
13999 (access_fn_components_comparable_p): Likewise.
14000 (dr_analyze_indices): Add a reference to access_fn_component_p.
14001 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14003 (constant_access_functions): Likewise.
14004 (add_other_self_distances): Likewise.
14005 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14006 (initialize_data_dependence_relation): Use XCNEW and remove
14007 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14008 of access functions that have the same type. Allow the
14009 subsequence to end with different bases in some circumstances.
14010 Record the chosen access functions in SUB_ACCESS_FN.
14011 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14012 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14013 (subscript_dependence_tester_1): Likewise dra and drb.
14014 (build_classic_dist_vector): Update calls accordingly.
14015 (subscript_dependence_tester): Likewise.
14016 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14017 DDR_COULD_BE_INDEPENDENT_P.
14018 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14019 comp_alias_ddrs instead of may_alias_ddrs.
14020 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14022 (vect_analyze_data_ref_dependence): Use it if
14023 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14024 distance vectors if that fails.
14025 (dependence_distance_ge_vf): New function.
14026 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14027 LOOP_VINFO_MAY_ALIAS_DDRS.
14029 2017-08-04 Richard Biener <rguenther@suse.de>
14031 PR middle-end/81705
14032 * fold-const.c (fold_binary_loc): Properly restrict
14033 minus_var0 && minus_var1 case when associating undefined overflow
14036 2017-08-04 Tom de Vries <tom@codesourcery.com>
14038 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14040 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14042 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14043 Don't start diagnostic messages with a capital letter.
14044 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14046 (rs6000_invalid_builtin): Likewise.
14047 (rs6000_trampoline_init): Likewise.
14049 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14052 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14053 after setting changeable df flags.
14055 2017-08-03 Richard Biener <rguenther@suse.de>
14057 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14058 up if the use is in USE - X.
14060 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14062 * toplev.c (dumpfile.h): New include.
14063 (internal_error_reentered): New static function. Use it...
14064 (internal_error_function): ...here to handle reentered internal_error.
14066 2017-08-03 Richard Biener <rguenther@suse.de>
14068 PR middle-end/81148
14069 * fold-const.c (split_tree): Add minus_var and minus_con
14070 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14071 here but always use minus_*.
14072 (associate_trees): Assert we never associate with MINUS_EXPR
14073 and NULL first operand. Do not recurse for PLUS_EXPR operands
14074 when associating as MINUS_EXPR either.
14075 (fold_binary_loc): Track minus_var and minus_con.
14077 2017-08-03 Tom de Vries <tom@codesourcery.com>
14080 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14081 ACCEL_COMPILER, apply finish_options on
14082 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14084 2017-08-03 Tom de Vries <tom@codesourcery.com>
14087 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14088 function_entry_patch_area_size > 0.
14090 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14093 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14094 instead of 10??LU, perform unit multiplication in wide_int,
14095 don't change alloc_object_size_limit if the limit is larger
14098 PR tree-optimization/81655
14099 PR tree-optimization/81588
14100 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14101 the case when ranges[i].low and high are 1 for unsigned type with
14104 PR middle-end/81052
14105 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14106 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14108 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14110 * tree-vrp.h: Add include guard.
14112 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14115 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14116 (ud2): New insn pattern.
14117 * config/i386/i386.c (ix86_expand_epilogue):
14118 For naked functions, generate ud2 instead of trap insn.
14120 2017-08-02 Marek Polacek <polacek@redhat.com>
14123 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14125 2017-08-02 Tom de Vries <tom@codesourcery.com>
14126 Cesar Philippidis <cesar@codesourcery.com>
14128 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14129 Add missing edge probabilities.
14131 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14133 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14134 Correct endianness.
14136 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14138 PR middle-end/79499
14139 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14140 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14141 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14142 prologue_seq sequences - if any.
14144 2017-08-02 Richard Biener <rguenther@suse.de>
14146 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14147 via vectors if supported, integer extracts via punning if supported
14148 or otherwise vector extracts.
14150 2017-08-02 Richard Biener <rguenther@suse.de>
14152 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14154 (bitmap_insert_into_set): ... this.
14156 2017-08-02 Richard Biener <rguenther@suse.de>
14158 PR tree-optimization/81633
14160 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14162 PR tree-optimization/71752
14163 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14165 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14167 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14168 (machine_function::call_ms2sysv_pad_out): Remove field.
14169 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14170 (ix86_compute_frame_layout): Likewise.
14172 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14175 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14176 attribute with interrupt attribute.
14178 2017-08-01 Andrew Pinski <apinski@cavium.com>
14180 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14181 BIT_INSERT_EXPR's operand 1
14182 to see if the types precision matches.
14184 2017-08-01 Martin Liska <mliska@suse.cz>
14186 PR middle-end/70140
14187 * builtins.c (expand_builtin_memcpy_args): Remove.
14188 (expand_builtin_memcpy): Call newly added function
14189 expand_builtin_memory_copy_args.
14190 (expand_builtin_memcpy_with_bounds): Likewise.
14191 (expand_builtin_mempcpy): Remove last argument.
14192 (expand_builtin_mempcpy_with_bounds): Likewise.
14193 (expand_builtin_memory_copy_args): New function created from
14194 expand_builtin_mempcpy_args with small modifications.
14195 (expand_builtin_mempcpy_args): Remove.
14196 (expand_builtin_stpcpy): Remove unused argument.
14197 (expand_builtin): Likewise.
14198 (expand_builtin_with_bounds): Likewise.
14200 2017-08-01 Martin Liska <mliska@suse.cz>
14203 Make mempcpy more optimal (PR middle-end/70140).
14205 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14208 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14209 __builtin_vec_cmpne verify both arguments are compatible vectors
14210 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14211 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14212 move computation of aligned to after checking the argument types.
14216 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14217 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14220 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14221 Doug Rupp <rupp@adacore.com>
14222 Olivier Hainque <hainque@adacore.com>
14224 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14225 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14227 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14228 for TARGET_OS_CPP_BUILTIN.
14229 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14230 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14232 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14233 passing required abi options to the assembler for EABI configurations.
14234 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14235 of .text.hot and .text.unlikely sections for kernel modules when
14236 using ARM style exceptions.
14237 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14238 options. Add EXTRA_CC1_SPEC.
14239 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14241 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14243 (ARM_TARGET2_DWARF_FORMAT): Define.
14244 * config/arm/t-vxworks: Adjust multilib control to removal of the
14245 Diab command line options.
14247 2017-08-01 Martin Liska <mliska@suse.cz>
14249 PR gcov-profile/81561
14250 * gcov.c (unblock): Make unblocking safe as we need to preserve
14251 index correspondence of blocks and block_lists.
14253 2017-08-01 Richard Biener <rguenther@suse.de>
14255 PR tree-optimization/81181
14256 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14257 (compute_antic): ... end of iteration here.
14259 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14261 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14262 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14263 (ftree-slp-vectorize): Likewise.
14264 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14265 can no longer be set independent of flag_tree_loop_vectorize.
14266 * omp-general.c (emp_max_vf): Likewise.
14267 * opts.c (enable_fdo_optimizations): Remove references to
14268 flag_tree_vectorize, these are now implicit.
14269 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14270 and leave it for the options machinery.
14272 2017-08-01 Martin Liska <mliska@suse.cz>
14274 PR middle-end/70140
14275 * builtins.c (expand_builtin_memcpy_args): Remove.
14276 (expand_builtin_memcpy): Call newly added function
14277 expand_builtin_memory_copy_args.
14278 (expand_builtin_memcpy_with_bounds): Likewise.
14279 (expand_builtin_mempcpy): Remove last argument.
14280 (expand_builtin_mempcpy_with_bounds): Likewise.
14281 (expand_builtin_memory_copy_args): New function created from
14282 expand_builtin_mempcpy_args with small modifications.
14283 (expand_builtin_mempcpy_args): Remove.
14284 (expand_builtin_stpcpy): Remove unused argument.
14285 (expand_builtin): Likewise.
14286 (expand_builtin_with_bounds): Likewise.
14288 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14291 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14292 print "ds:" only for immediates in generic address space.
14294 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14297 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14298 (ix86_function_ok_for_sibcall): Return false for naked functions.
14300 2017-08-01 Richard Biener <rguenther@suse.de>
14302 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14303 (compute_antic): Seed worklist with exit block predecessors.
14304 * cfganal.c (dfs_find_deadend): For a cycle return the source
14305 of the edge closing it.
14307 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14309 * config/aarch64/aarch64.c
14310 (aarch64_can_const_movi_rtx_p): Move 0 check.
14312 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14314 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14315 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14317 * match.pd: Ditto in address comparison pattern.
14319 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14321 PR tree-optimization/81627
14322 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14323 closed ssa form for store-store chain.
14325 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14327 PR tree-optimization/81620
14328 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14329 for store-store chain.
14331 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14333 PR tree-optimization/81588
14334 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14335 ranges[i].in_p, invert comparison code ccode. For >/>=,
14336 swap rhs1 and rhs2 and comparison code unconditionally,
14337 for </<= don't do that. Don't swap rhs1/rhs2 again if
14338 ranges[i].in_p, instead invert comparison code ccode if
14339 opcode or oe->rank is BIT_IOR_EXPR.
14342 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14343 a direct optab to conversion optab.
14344 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14345 with GET_MODE_INNER as last argument instead of optab_handler.
14346 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14347 vector extraction if possible and optab is available.
14348 * expr.c (store_constructor): Use convert_optab_handler instead
14349 of optab_handler. Use vector initialization from smaller
14350 vectors if possible and optab is available.
14351 * tree-vect-stmts.c (vectorizable_load): Likewise.
14352 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14353 now have two modes.
14354 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14355 of vec_init from half-sized vectors with the same element mode.
14356 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14357 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14358 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14359 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14360 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14361 after mode in gen_vec_extract* calls.
14362 (vec_extract<mode>): Renamed to ...
14363 (vec_extract<mode><ssescalarmodelower>): ... this.
14364 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14365 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14366 element mode after mode in gen_vec_init* calls.
14367 (VEC_INIT_HALF_MODE): New mode iterator.
14368 (vec_init<mode>): Renamed to ...
14369 (vec_init<mode><ssescalarmodelower>): ... this.
14370 (vec_init<mode><ssehalfvecmodelower>): New expander.
14371 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14372 (vec_extractv2sfsf): ... this.
14373 (vec_initv2sf): Renamed to ...
14374 (vec_initv2sfsf): ... this.
14375 (vec_extractv2si): Renamed to ...
14376 (vec_extractv2sisi): ... this.
14377 (vec_initv2si): Renamed to ...
14378 (vec_initv2sisi): ... this.
14379 (vec_extractv4hi): Renamed to ...
14380 (vec_extractv4hihi): ... this.
14381 (vec_initv4hi): Renamed to ...
14382 (vec_initv4hihi): ... this.
14383 (vec_extractv8qi): Renamed to ...
14384 (vec_extractv8qiqi): ... this.
14385 (vec_initv8qi): Renamed to ...
14386 (vec_initv8qiqi): ... this.
14387 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14388 (vec_init<mode>): Renamed to ...
14389 (vec_init<mode><VEC_base_l>): ... this.
14390 (vec_extract<mode>): Renamed to ...
14391 (vec_extract<mode><VEC_base_l>): ... this.
14392 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14393 (vec_initv2sfsf): ... this.
14394 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14395 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14396 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14397 element mode after mode in gen_vec_init* calls.
14398 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14399 (vec_init<mode><Vel>): ... this.
14400 (vec_extract<mode>): Renamed to ...
14401 (vec_extract<mode><Vel>): ... this.
14402 * config/aarch64/iterators.md (Vel): New mode attribute.
14403 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14404 Add element mode after mode in gen_vec_extract* calls.
14405 * config/s390/vector.md (non_vec_l): New mode attribute.
14406 (vec_extract<mode>): Renamed to ...
14407 (vec_extract<mode><non_vec_l>): ... this.
14408 (vec_init<mode>): Renamed to ...
14409 (vec_init<mode><non_vec_l>): ... this.
14410 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14411 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14413 * config/arm/iterators.md (V_elem_l): New mode attribute.
14414 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14415 (vec_extract<mode><V_elem_l>): ... this.
14416 (vec_extractv2di): Renamed to ...
14417 (vec_extractv2didi): ... this.
14418 (vec_init<mode>): Renamed to ...
14419 (vec_init<mode><V_elem_l>): ... this.
14420 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14421 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14422 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14423 Add element mode after gen_vec_extract* calls.
14424 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14425 (vec_init<mode><unitmode>): ... this.
14426 (vec_extract<mode>): Renamed to ...
14427 (vec_extract<mode><unitmode>): ... this.
14428 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14429 (vec_init<mode><unitmode>): ... this.
14430 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14431 (vec_initv2sfsf): ... this.
14432 (vec_extractv2sf): Renamed to ...
14433 (vec_extractv2sfsf): ... this.
14434 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14435 Add element mode after gen_vec_extract* calls.
14436 * config/mips/mips.md (unitmode): New mode iterator.
14437 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14438 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14439 * config/spu/spu.md (inner_l): New mode attribute.
14440 (vec_init<mode>): Renamed to ...
14441 (vec_init<mode><inner_l>): ... this.
14442 (vec_extract<mode>): Renamed to ...
14443 (vec_extract<mode><inner_l>): ... this.
14444 * config/sparc/sparc.md (veltmode): New mode iterator.
14445 (vec_init<VMALL:mode>): Renamed to ...
14446 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14447 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14448 (vec_initv2sisi): ... this.
14449 (vec_initv2sf): Renamed to ...
14450 (vec_initv2sfsf): ... this.
14451 (vec_extractv2sf): Renamed to ...
14452 (vec_extractv2sfsf): ... this.
14453 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14454 (vec_init<mode>): Renamed to ...
14455 (vec_init<mode><VEC_base_l>): ... this.
14456 (vec_extract<mode>): Renamed to ...
14457 (vec_extract<mode><VEC_base_l>): ... this.
14458 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14459 (vec_initv2sfsf): ... this.
14460 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14461 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14462 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14463 gen_vec_init* calls.
14465 2017-08-01 Richard Biener <rguenther@suse.de>
14467 PR tree-optimization/81297
14468 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14469 TREE_OVERFLOW from INTEGER_CSTs.
14471 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14473 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14475 2017-07-31 Carl Love <cel@us.ibm.com>
14477 * config/rs6000/rs6000-c: Add support for built-in functions
14478 vector signed char vec_xl_be (signed long long, signed char *);
14479 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14480 vector signed int vec_xl_be (signed long long, signed int *);
14481 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14482 vector signed long long vec_xl_be (signed long long, signed long long *);
14483 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14484 vector signed short vec_xl_be (signed long long, signed short *);
14485 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14486 vector double vec_xl_be (signed long long, double *);
14487 vector float vec_xl_be (signed long long, float *);
14488 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14489 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14490 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14492 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14493 (altivec_expand_builtin): Add switch statement to call
14494 altivec_expand_xl_be for each builtin.
14495 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14496 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14497 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14498 __builtin_vsx_le_be_v16qi.
14499 * doc/extend.texi: Update the built-in documentation file for the
14500 new built-in functions.
14502 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14505 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14507 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14509 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14511 * config.gcc: Add z14.
14512 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14513 CPU model numbers for z13s and z14.
14514 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14516 * config/s390/s390-opts.h (enum processor_type): Rename
14517 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14518 * config/s390/s390.c (processor_table): Add field for CPU name to
14519 be passed to Binutils.
14520 (s390_asm_output_machine_for_arch): Use the new field in
14521 processor_table for Binutils.
14522 (s390_expand_builtin): Replace arch12 with z14.
14523 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14524 (s390_get_sched_attrmask): Likewise.
14525 (s390_get_unit_mask): Likewise.
14526 * config/s390/s390.opt: Add z14 to processor_type enum.
14528 2017-07-31 Martin Jambor <mjambor@suse.cz>
14531 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14532 regardless of optimization level.
14534 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14535 Martin Liska <mliska@suse.cz>
14537 * predict.def: Remove old comment and adjust probability.
14538 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14539 PREDICT statements.
14541 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14544 * config/i386/i386.c (ix86_function_naked): New function.
14545 (ix86_can_use_return_insn_p): Return false for naked functions.
14546 (ix86_expand_prologue): Skip prologue for naked functions.
14547 (ix86_expand_epilogue): Skip epilogue for naked functions
14548 and emit trap instruction.
14549 (ix86_warn_func_return): New function.
14550 (ix86_attribute_table): Add "naked" attribute specification.
14551 (TARGET_WARN_FUNC_RETURN): Define.
14552 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14554 2017-07-31 Martin Liska <mliska@suse.cz>
14556 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14557 (dump_gimple_bb_header): Always dump BB info.
14558 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14560 2017-07-31 Martin Liska <mliska@suse.cz>
14563 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14564 also with current_function_decl non-null equality.
14566 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14569 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14570 change type to the element type, instead add eltype variable and
14571 use it where we are interested in the element type.
14573 PR tree-optimization/81603
14574 * ipa-polymorphic-call.c
14575 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14576 offset arithmetic in offset_int, bail out if the resulting bit offset
14577 doesn't fit into shwi.
14579 2017-07-31 Martin Liska <mliska@suse.cz>
14581 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14582 (gimplify_save_expr): Fix comment.
14584 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14587 * config/i386/i386.c (ix86_function_arg): Update arguments for
14589 (ix86_compute_frame_layout): Set the initial stack offset to
14590 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14591 INCOMING_FRAME_SP_OFFSET.
14592 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14593 stack before exception handler returns.
14594 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14595 the 'ERROR_CODE' for exception handler.
14597 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14599 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14600 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14601 (ASM_OUTPUT_REG_POP): Ditto.
14602 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14603 instead of asm_fprintf to output pure string.
14605 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14607 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14608 to imported_module_or_decl hook.
14609 (debug_nothing_tree_tree_tree_bool): Remove.
14610 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14611 * debug.c (do_nothing_debug_hooks): Use
14612 debug_nothing_tree_tree_tree_bool_bool instead of
14613 debug_nothing_tree_tree_tree_bool.
14614 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14615 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14616 * sdbout.c (sdb_debug_hooks): Likewise.
14617 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14618 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14620 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14621 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14622 attribute, don't add anything.
14624 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14626 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14627 (fold_build2_stat_loc): Likewise.
14628 (fold_build3_stat_loc): Likewise.
14629 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14630 (fold_build1_loc): Remove macro.
14631 (fold_build2_loc): Likewise.
14632 (fold_build3_loc): Likewise.
14634 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14636 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14637 (gimple_build_debug_bind_source_stat): Likewise.
14638 * gimple.h (gimple_build_debug_bind): Remove macro.
14639 (gimple_build_debug_bind_source): Likewise.
14641 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14643 * bitmap.c (bitmap_alloc): Adjust.
14644 (bitmap_gc_alloc): Likewise.
14645 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14647 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14649 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14650 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14651 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14652 (bitmap_gc_alloc_stat): Likewise.
14653 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14655 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14657 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14658 * rtl.h (shallow_copy_rtx): Remove macro.
14660 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14662 * emit-rtl.c (gen_raw_REG): Adjust.
14663 * gengenrtl.c (gendef): Likewise.
14664 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14665 * rtl.h (rtx_alloc): Remove macro.
14667 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14669 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14670 (build_tree_list_stat): Likewise.
14671 * tree.h (build_tree_list): Remove macro.
14672 (build_tree_list_vec): Likewise.
14674 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14676 * tree.c (make_vector_stat): Remove _stat from name.
14677 (build_vector_stat): Likewise.
14678 * tree.h (make_vector_stat): Remove macro.
14679 (build_vector_stat): Likewise.
14681 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14683 * tree.h (build_var_debug_value): Remove prototype.
14685 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14687 * tree.c (tree_cons_stat): Remove _stat from name.
14688 * tree.h (tree_cons): Remove macro.
14690 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14692 * tree.c (build_vl_exp_stat): Remove _stat from name.
14693 * tree.h (build_vl_exp): Remove macro.
14695 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14697 * tree.c (build_decl_stat): Remove _stat from name.
14698 * tree.h (build_decl): Remove macro.
14700 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14702 * gimple.c (gimple_build_with_ops_stat): Adjust.
14703 (gimple_alloc_stat): Remove _stat from name.
14704 * gimple.h (gimple_alloc): Remove macro.
14706 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14708 * tree.c (make_tree_vec_stat): Remove _stat from name.
14709 (grow_tree_vec_stat): Likewise.
14710 * tree.h (make_tree_vec_stat): Adjust prototype.
14711 (grow_tree_vec_stat): Likewise.
14712 (make_tree_vec): Remove macro.
14713 (grow_tree_vec): Likewise.
14715 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14717 * fold-const.c (fold_build1_stat_loc): Adjust.
14718 (fold_build2_stat_loc): Likewise.
14719 (fold_build3_stat_loc): Likewise.
14720 * tree.c (build0_stat): Remove _stat from name.
14721 (build1_stat): Likewise.
14722 (build2_stat): Likewise.
14723 (build3_stat): Likewise.
14724 (build4_stat): Likewise.
14725 (build5_stat): Likewise.
14726 * tree.h (build1_loc): Remove macro, and rename _stat function
14728 (build2_loc): Likewise.
14729 (build3_loc): Likewise.
14730 (build4_loc): Likewise.
14731 (build5_loc): Likewise.
14733 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14735 * tree.c (make_int_cst_stat): Remove _stat from name.
14736 * tree.h (make_int_cst_stat): Adjust prototype.
14737 (make_int_cst): Remove macro.
14739 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14741 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14742 * tree.h (make_tree_binfo_stat): Adjust prototype.
14743 (make_tree_binfo): Remove.
14745 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14747 * tree.c (copy_node_stat): Rename to copy_node.
14748 (build_distinct_type_copy): Adjust.
14749 * tree.h (copy_node_stat): Adjust prototype.
14750 (copy_node): Remove macro.
14752 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14754 * tree.c (make_node_stat): rename to make_node.
14755 (build_tree_list_stat): Adjust.
14756 (build0_stat): Likewise.
14757 (build2_stat): Likewise.
14758 (build3_stat): Likewise.
14759 (build4_stat): Likewise.
14760 (build5_stat): Likewise.
14761 (build_decl_stat): Likewise.
14762 * tree.h (make_node_stat): Adjust prototype.
14763 (make_node): remove macro.
14765 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14767 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14768 (PPC_FEATURE2_SCV): Likewise.
14769 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14771 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14773 * config/aarch64/aarch64.c
14774 (aarch64_internal_mov_immediate): Add new special pattern.
14775 * config/aarch64/aarch64.md (*movdi_aarch64):
14776 Add reg/32bit const mov case.
14778 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14779 Richard Sandiford <richard.sandiford@linaro.org>
14781 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14782 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14783 Add integer and movi cases.
14784 (movi-split-hf-df-sf split, fp16): New.
14785 (enabled): Added TARGET_FP_F16INST.
14786 * config/aarch64/iterators.md (GPF_HF): New.
14787 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14789 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14791 * config/aarch64/aarch64.c
14792 (aarch64_simd_container_mode): Add prototype.
14793 (aarch64_expand_mov_immediate): Add HI support.
14794 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14795 (aarch64_can_const_movi_rtx_p): New.
14796 (aarch64_preferred_reload_class):
14797 Remove restrictions of using FP registers for certain SIMD operations.
14798 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14799 (aarch64_valid_floating_const): Add integer move validation.
14800 (aarch64_simd_imm_scalar_p): Remove.
14801 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14802 (aarch64_legitimate_constant_p): Expand list of supported cases.
14803 * config/aarch64/aarch64-protos.h
14804 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14805 (aarch64_reinterpret_float_as_int): New.
14806 (aarch64_simd_imm_scalar_p): Remove.
14807 * config/aarch64/constraints.md (Uvi): New.
14808 (Dd): Split into Ds and new Dd.
14809 * config/aarch64/aarch64.md (*movsi_aarch64):
14811 (*movdi_aarch64): Add SIMD mov case.
14813 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14815 * tree-predcom.c: (struct chain): Handle store-store chain in which
14816 stores for elimination only store loop invariant values.
14817 (execute_pred_commoning_chain): Ditto.
14818 (prepare_initializers_chain_store_elim): Ditto.
14819 (prepare_finalizers): Ditto.
14820 (is_inv_store_elimination_chain): New function.
14821 (initialize_root_vars_store_elim_1): New function.
14823 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14825 * tree-predcom.c: Revise general description of the pass.
14826 (enum chain_type): New enum type for store elimination.
14827 (struct chain): New field supporting store elimination.
14828 (struct component): Ditto.
14829 (dump_chain): Dump store-stores chain.
14830 (release_chain): Release resources.
14831 (split_data_refs_to_components): Compute and create component
14832 contains only stores for elimination.
14833 (get_chain_last_ref_at): New function.
14834 (make_invariant_chain): Initialization.
14835 (make_rooted_chain): Specify chain type in parameter and record it.
14836 (add_looparound_copies): Skip for store-stores chain.
14837 (determine_roots_comp): Compute type of chain and pass it to
14839 (initialize_root_vars_store_elim_2): New function.
14840 (finalize_eliminated_stores): New function.
14841 (remove_stmt): Handle store for elimination.
14842 (execute_pred_commoning_chain): Execute predictive commoning on
14843 store-store chains.
14844 (determine_unroll_factor): Skip unroll for store-stores chain.
14845 (prepare_initializers_chain_store_elim): New function.
14846 (prepare_initializers_chain): Hanlde store-store chain.
14847 (prepare_finalizers_chain, prepare_finalizers): New function.
14848 (tree_predictive_commoning_loop): Return integer value indicating
14849 if loop is unrolled or lcssa form is corrupted.
14850 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14852 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14854 * tree-predcom.c (initialize_root): Delete.
14855 (execute_pred_commoning_chain): Initialize root vars and replace
14856 reference of non-combined chain directly, rather than call above
14859 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14861 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14862 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14864 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14866 * tree-predcom.c (struct chain): New field init_seq.
14867 (release_chain): Release init_seq.
14868 (prepare_initializers_chain): Record intialization stmts in above
14870 (insert_init_seqs): New function.
14871 (tree_predictive_commoning_loop): Call insert_init_seqs.
14873 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14875 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14877 2017-07-28 Richard Biener <rguenther@suse.de>
14879 * match.pd: Remove superfluous :c.
14880 * genmatch.c (simplify::id): Add member.
14881 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14883 (current_id): New global.
14884 (dt_node::parent): Move from ...
14885 (dt_operand::parent): ... here. Add for_id member.
14886 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14887 (decision_tree::find_node): Relax order requirement when
14888 merging DT_TRUE nodes to ones inbetween the current simplify
14889 and the one we try to merge with. Add diagnostic whenever
14890 we need to enforce pattern order by not merging.
14891 (decision_tree::insert): Set current_id.
14892 (decision_tree::print_node): Dump parent node and for_id.
14893 (parser::last_id): Add member.
14894 (parser::push_simplify): Assign unique id.
14895 (parser::parser): Initialize last_id.
14897 2017-07-28 Martin Liska <mliska@suse.cz>
14900 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14901 gimple_build_debug_bind.
14903 2017-07-28 Richard Biener <rguenther@suse.de>
14905 PR tree-optimization/81502
14906 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14908 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14910 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14911 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14912 for BIT_FIELD_REF args.
14913 * fold-const.c (make_bit_field_ref): Likewise.
14914 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14916 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14919 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14920 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14921 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14922 Or it into SANITIZER_UNDEFINED.
14923 * ubsan.c: Include gimple-fold.h and varasm.h.
14924 (ubsan_expand_ptr_ifn): New function.
14925 (instrument_pointer_overflow): New function.
14926 (maybe_instrument_pointer_overflow): New function.
14927 (instrument_object_size): Formatting fix.
14928 (pass_ubsan::execute): Call instrument_pointer_overflow
14929 and maybe_instrument_pointer_overflow.
14930 * internal-fn.c (expand_UBSAN_PTR): New function.
14931 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14932 * sanitizer.def (__ubsan_handle_pointer_overflow,
14933 __ubsan_handle_pointer_overflow_abort): New builtins.
14934 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14935 * internal-fn.def (UBSAN_PTR): New internal function.
14936 * opts.c (sanitizer_opts): Add pointer-overflow.
14937 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14938 * fold-const.c (build_range_check): Compute pointer range check in
14939 integral type if pointer arithmetics would be needed. Formatting
14942 2017-07-28 Martin Liska <mliska@suse.cz>
14945 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14946 parameters that are of a variable-length.
14948 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14950 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14952 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14953 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14954 (CRT_CALL_STATIC_FUNCTION): Likewise.
14955 (ASM_DEFAULT_SPEC): New define.
14956 (ASM_SPEC32): Likewise.
14957 (ASM_SPEC64): Likewise.
14958 (ASM_SPEC_COMMON): Likewise.
14959 (ASM_SPEC): Likewise.
14960 (INVALID_64BIT): Likewise.
14961 (LINK_OS_DEFAULT_SPEC): Likewise.
14962 (LINK_OS_SPEC32): Likewise.
14963 (LINK_OS_SPEC64): Likewise.
14964 (POWERPC_LINUX): Likewise.
14965 (PTRDIFF_TYPE): Likewise.
14966 (RESTORE_FP_PREFIX): Likewise.
14967 (RESTORE_FP_SUFFIX): Likewise.
14968 (SAVE_FP_PREFIX): Likewise.
14969 (SAVE_FP_SUFFIX): Likewise.
14970 (SIZE_TYPE): Likewise.
14971 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14972 (TARGET_64BIT): Likewise.
14973 (TARGET_64BIT): Likewise.
14974 (TARGET_AIX): Likewise.
14975 (WCHAR_TYPE_SIZE): Likewise.
14976 (WCHAR_TYPE): Undefine.
14977 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14978 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14979 (CPP_OS_RTEMS_SPEC): Delete.
14980 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14981 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14983 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14985 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14987 PR tree-optimization/81578
14988 * tree-parloops.c (build_new_reduction): Bail out if
14989 reduction_code isn't one of the standard OpenMP reductions.
14990 Move the details printing after that decision.
14992 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14994 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14995 related to reload_in_progress.
14996 (splat_input_operand): Likewise.
14997 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14999 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15001 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15002 (TARGET_INSTANTIATE_DECLS): Likewise.
15003 (legitimate_indexed_address_p): Delete reload_in_progress code.
15004 (rs6000_debug_legitimate_address_p): Likewise.
15005 (rs6000_eliminate_indexed_memrefs): Likewise.
15006 (rs6000_emit_le_vsx_store): Likewise.
15007 (rs6000_emit_move_si_sf_subreg): Likewise.
15008 (rs6000_emit_move): Likewise.
15009 (register_to_reg_type): Likewise.
15010 (rs6000_pre_atomic_barrier): Likewise.
15011 (rs6000_machopic_legitimize_pic_address): Likewise.
15012 (rs6000_allocate_stack_temp): Likewise.
15013 (rs6000_address_for_fpconvert): Likewise.
15014 (rs6000_address_for_altivec): Likewise.
15015 (rs6000_secondary_memory_needed_rtx): Delete function.
15016 (rs6000_check_sdmode): Likewise.
15017 (rs6000_alloc_sdmode_stack_slot): Likewise.
15018 (rs6000_instantiate_decls): Likewise.
15019 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15020 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15021 Delete reload_in_progress.
15022 (*vec_reload_and_plus_<mptrsize>): Likewise.
15023 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15024 (vsx_div_v2di): Likewise.
15025 (vsx_udiv_v2di): Likewise.
15027 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15029 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15030 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15031 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15032 (rs6000_debug_reg_global): Delete print of LRA status.
15033 (rs6000_option_override_internal): Delete dead LRA related code.
15034 (rs6000_lra_p): Delete function.
15035 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15037 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15039 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15040 * config/riscv/rtems.h: New file.
15042 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15043 Sudakshina Das <sudi.das@arm.com>
15045 * config/aarch64/aarch64.md
15046 (define_split for and<mode>3nr_compare): Move
15047 non aarch64_logical_operand to a register.
15048 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15049 register immediate operand to a register.
15050 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15052 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15054 PR middle-end/81564
15055 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15057 2017-07-27 Richard Biener <rguenther@suse.de>
15059 PR tree-optimization/81573
15060 PR tree-optimization/81494
15061 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15062 multi defuse cycle case.
15064 2017-07-27 Richard Biener <rguenther@suse.de>
15066 PR tree-optimization/81571
15067 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15070 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15072 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15073 earlier and only if MASK_FPU is set. Adjust formatting.
15075 2017-07-27 Martin Liska <mliska@suse.cz>
15077 * opt-functions.awk: Add validation of value of Init.
15078 * optc-gen.awk: Pass new argument.
15080 2017-07-27 Martin Liska <mliska@suse.cz>
15082 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15083 Fix wrong condition.
15085 2017-07-27 Martin Liska <mliska@suse.cz>
15087 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15088 BBs and edges seen by autoFDO.
15090 2017-07-27 Richard Biener <rguenther@suse.de>
15092 PR tree-optimization/81502
15093 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15094 with incompatible but same sized type.
15095 (execute_update_addresses_taken): Likewise.
15097 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15099 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15100 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15102 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15105 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15106 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15107 Change s_operand to memory_operand.
15109 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15111 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15112 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15113 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15114 Emit instructions rather than returning an expression. Handle TFmode
15115 and KFmode by casting to TImode.
15116 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15117 (rs6000_emit_le_vsx_store): Likewise.
15118 * config/rs6000/vsx.md (VSX_TI): New iterator.
15119 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15120 (*vsx_le_undo_permute_<mode>): Likewise.
15121 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15122 emit the split sequence.
15123 (*vsx_le_perm_store_<mode>): Likewise.
15125 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15127 PR tree-optimization/81555
15128 PR tree-optimization/81556
15129 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15130 if true, force CHANGED for the recursive invocation.
15131 (reassociate_bb): Remember original length of ops array, pass
15132 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15134 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15135 attributes for noipa attribute. For naked attribute use
15136 lookup_attribute first before lookup_attribute_spec.
15137 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15139 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15140 for functions with noipa attribute.
15141 (cgraph_externally_visible_p): Return true for functions with noipa
15143 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15144 for functions with noipa attribute.
15145 * doc/extend.texi: Document noipa function attribute.
15146 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15147 also for functions with noipa attribute.
15148 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15150 2017-07-26 Andrew Pinski <apinski@cavium.com>
15152 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15153 vec_unalign_load_cost and vec_unalign_store_cost.
15155 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15157 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15158 -mvsx-small-integer option.
15159 (ISA_3_0_MASKS_IEEE): Likewise.
15160 (OTHER_VSX_VECTOR_MASKS): Likewise.
15161 (POWERPC_MASKS): Likewise.
15162 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15163 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15164 code, only testing for DImode being allowed in non-VSX floating
15166 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15167 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15169 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15170 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15171 TARGET_P8_VECTOR test.
15172 (rs6000_secondary_reload_simple_move): Likewise.
15173 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15174 since TARGET_P9_VECTOR was already tested.
15175 (rs6000_opt_masks): Remove -mvsx-small-integer.
15176 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15177 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15179 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15180 test for TARGET_VEXTRACTUB was used, and that uses
15182 (p9 extract splitter): Likewise.
15183 (vsx_extract_<mode>_di_p9): Likewise.
15184 (vsx_extract_<mode>_store_p9): Likewise.
15185 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15186 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15187 the elimination of TARGET_VSX_SMALL_INTEGER.
15188 (vsx_extract_<mode>_p8): Likewise.
15189 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15190 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15191 (vsx_set_<mode>_p9): Likewise.
15192 (vsx_set_v4sf_p9): Likewise.
15193 (vsx_set_v4sf_p9_zero): Likewise.
15194 (vsx_insert_extract_v4sf_p9): Likewise.
15195 (vsx_insert_extract_v4sf_p9_2): Likewise.
15196 * config/rs6000/rs6000.md (sign extend splitter): Change
15197 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15198 (floatsi<mode>2_lfiwax_mem): Likewise.
15199 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15200 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15201 since a test for TARGET_P9_VECTOR was used.
15202 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15203 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15204 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15205 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15206 TARGET_P8_VECTOR test.
15207 (fix_trunc<mode>si2_stfiwx): Likewise.
15208 (fix_trunc<mode>si2_internal): Likewise.
15209 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15210 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15212 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15213 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15214 TARGET_P8_VECTOR test.
15215 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15216 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15217 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15219 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15220 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15221 since a test for TARGET_P9_VECTOR was used.
15222 (splitter for loading small constants): Likewise.
15224 2017-07-26 Andrew Pinski <apinski@cavium.com>
15226 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15229 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15232 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15233 (fp_valid_at): Likewise.
15235 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15237 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15238 (qdf24xx_addrcost_table): Likewise.
15239 (cortexa57_tunings): Update to use generic_branch_cost.
15240 (cortexa72_tunings): Likewise.
15241 (cortexa73_tunings): Likewise.
15242 (qdf24xx_tunings): Likewise.
15244 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15246 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15247 (thunderx2t99_branch_cost): Likewise.
15248 (cortexa35_tunings): Update to use generic_branch_cost.
15249 (cortexa53_tunings): Likewise.
15250 (cortexa57_tunings): Likewise.
15251 (cortexa72_tunings): Likewise.
15252 (cortexa73_tunings): Likewise.
15253 (thunderx2t99_tunings): Likewise.
15255 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15257 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15258 (sparc_option_override): Honour MASK_FSMULD.
15259 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15260 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15261 * config/sparc/sparc.opt (mfsmuld): New option.
15262 * doc/invoke.texi (mfsmuld): Document option.
15264 2017-07-26 Marek Polacek <polacek@redhat.com>
15266 PR middle-end/70992
15267 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15269 2017-07-26 Richard Biener <rguenther@suse.de>
15271 * gimple-match-head.c (do_valueize): Return OP if valueize
15273 (get_def): New helper to get at the def stmt of a SSA name
15274 if valueize allows.
15275 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15276 do_valueize to get at the def stmt.
15277 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15279 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15281 PR middle-end/46932
15282 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15284 2017-07-26 Martin Liska <mliska@suse.cz>
15287 * function.c (expand_function_start): Make expansion of
15288 nonlocal_goto_save_area after parm_birth_insn.
15290 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15292 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15293 from all CPU target flags enable members.
15295 2017-07-26 Richard Biener <rguenther@suse.de>
15297 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15298 (decision_tree::gen): Make 'type' const.
15299 (write_predicate): Likewise.
15301 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15303 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15304 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15305 (rs6000_option_override_internal): Likewise.
15306 (rs6000_expand_vector_set): Likewise.
15307 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15308 (TARGET_UPPER_REGS_SF): Likewise.
15309 (TARGET_UPPER_REGS_DI): Likewise.
15310 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15311 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15312 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15313 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15314 (Splitters for DI constants in Altivec registers): Likewise.
15315 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15316 (vsx_set_v4sf_p9): Likewise.
15317 (vsx_set_v4sf_p9_zero): Likewise.
15318 (vsx_insert_extract_v4sf_p9): Likewise.
15319 (vsx_insert_extract_v4sf_p9_2): Likewise.
15321 2017-07-25 Carl Love <cel@us.ibm.com>
15323 * doc/extend.texi: Update the built-in documentation file for the
15324 existing built-in functions
15325 vector signed char vec_cnttz (vector signed char);
15326 vector unsigned char vec_cnttz (vector unsigned char);
15327 vector signed short vec_cnttz (vector signed short);
15328 vector unsigned short vec_cnttz (vector unsigned short);
15329 vector signed int vec_cnttz (vector signed int);
15330 vector unsigned int vec_cnttz (vector unsigned int);
15331 vector signed long long vec_cnttz (vector signed long long);
15332 vector unsigned long long vec_cnttz (vector unsigned long long);
15334 2017-07-25 Andrew Pinski <apinski@cavium.com>
15336 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15337 accesses where the use is for the first operand of a BIT_INSERT.
15339 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15342 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15343 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15345 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15347 * config/i386/gstabs.h: Delete.
15348 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15350 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15352 * config/i386/i386.c (ix86_decompose_address): Do not check for
15353 register RTX when looking at index_reg or base_reg.
15354 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15356 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15358 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15359 to update EH info here.
15361 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15363 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15365 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15367 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15369 2017-07-25 Torsten Duwe <duwe@suse.de>
15371 * common.opt: Introduce -fpatchable-function-entry
15372 command line option, and its variables function_entry_patch_area_size
15373 and function_entry_patch_area_start.
15374 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15375 including a two-value parser.
15376 * target.def (print_patchable_function_entry): New target hook.
15377 * targhooks.h (default_print_patchable_function_entry): New function.
15378 * targhooks.c (default_print_patchable_function_entry): Likewise.
15379 * toplev.c (process_options): Switch off IPA-RA if
15380 patchable function entries are being generated.
15381 * varasm.c (assemble_start_function): Look at the
15382 patchable-function-entry command line switch and current
15383 function attributes and maybe generate NOP instructions by
15384 calling the print_patchable_function_entry hook.
15385 * doc/extend.texi: Document patchable_function_entry attribute.
15386 * doc/invoke.texi: Document -fpatchable_function_entry
15387 command line option.
15388 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15390 * doc/tm.texi: Re-generate.
15392 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15395 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15396 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15398 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15400 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15402 2017-07-25 Richard Biener <rguenther@suse.de>
15404 PR tree-optimization/81455
15405 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15406 not walk in cycles when looking for guards.
15408 2017-07-25 Richard Biener <rguenther@suse.de>
15410 PR tree-optimization/81529
15411 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15412 when optimizing backedge uses.
15414 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15416 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15418 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15419 to dl_section_ref. On AIX, append an expression to subtract
15420 the size of the section length to dl_section_ref.
15422 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15424 * configure.ac: If any of the config.* scripts fail, exit 1.
15425 * configure: Regenerate.
15427 2017-07-25 Richard Biener <rguenther@suse.de>
15429 PR middle-end/81546
15430 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15431 of immediate uses, be more verbose on errors.
15433 2017-07-25 Richard Biener <rguenther@suse.de>
15435 PR tree-optimization/81510
15436 * tree-vect-loop.c (vect_is_simple_reduction): When the
15437 reduction stmt is not inside the loop bail out.
15439 2017-07-25 Richard Biener <rguenther@suse.de>
15441 PR tree-optimization/81303
15442 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15443 profitability check against LOOP_VINFO_NITERSM1.
15445 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15447 * domwalk.c (cmp_bb_postorder): Simplify.
15448 (sort_bbs_postorder): New function. Use it...
15449 (dom_walker::walk): ...here to optimize common cases.
15451 2017-07-25 Martin Liska <mliska@suse.cz>
15454 * ipa-visibility.c (function_and_variable_visibility): Make the
15455 redirection just on target that supports aliasing.
15456 Fix GNU coding style.
15458 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15461 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15463 * config/arm/rtems.h: Likewise.
15464 * config/bfin/rtems.h: Likewise.
15465 * config/i386/rtemself.h: Likewise.
15466 * config/lm32/rtems.h: Likewise.
15467 * config/m32c/rtems.h: Likewise.
15468 * config/m68k/rtemself.h: Likewise.
15469 * config/microblaze/rtems.h: Likewise.
15470 * config/mips/rtems.h: Likewise.
15471 * config/moxie/rtems.h: Likewise.
15472 * config/nios2/rtems.h: Likewise.
15473 * config/powerpcspe/rtems.h: Likewise.
15474 * config/rs6000/rtems.h: Likewise.
15475 * config/rtems.h: Likewise.
15476 * config/sh/rtems.h: Likewise.
15477 * config/sh/rtemself.h: Likewise.
15478 * config/sparc/rtemself.h: Likewise.
15480 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15483 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15484 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15485 * tree-ssa-structalias.c (alias_get_name): Same.
15487 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15490 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15491 instructions if no du chain is found.
15493 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15495 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15497 2017-07-25 Richard Biener <rguenther@suse.de>
15499 PR middle-end/81505
15500 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15503 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15505 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15506 upper-regs options.
15507 (ISA_2_7_MASKS_SERVER): Likewise.
15508 (ISA_3_0_MASKS_IEEE): Likewise.
15509 (OTHER_P8_VECTOR_MASKS): Likewise.
15510 (OTHER_VSX_VECTOR_MASKS): Likewise.
15511 (POWERPC_MASKS): Likewise.
15512 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15513 duplicate list of options.
15514 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15515 explicit -mupper-regs options.
15516 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15517 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15518 alias for -mupper-regs-df.
15519 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15520 (rs6000_init_hard_regno_mode_ok): Likewise.
15521 (rs6000_option_override_internal): Likewise.
15522 (rs6000_opt_masks): Likewise.
15523 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15524 options in terms of whether -mvsx or -mpower8-vector was used.
15525 (TARGET_UPPER_REGS_DI): Likewise.
15526 (TARGET_UPPER_REGS_SF): Likewise.
15527 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15528 -mupper-regs-* options.
15530 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15532 * passes.c (emergency_dump_function): Print some empty lines and a
15533 header before the RTL dump.
15535 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15537 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15539 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15542 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15543 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15545 2017-07-24 Carl Love <cel@us.ibm.com>
15547 * config/rs6000/rs6000-c.c: Add support for built-in functions
15548 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15549 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15550 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15551 vec_extract_fp_from_shortl): Add defines for the two builtins.
15552 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15553 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15555 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15556 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15557 * doc/extend.texi: Update the built-in documentation file for the
15558 new built-in function.
15560 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15563 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15565 * doc/generic.texi: Likewise.
15566 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15567 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15569 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15571 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15572 (aarch64_mls_elt_merge<mode>): Likewise.
15574 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15576 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15577 having __cxa_atexit.
15579 2017-07-23 Michael Collison <michael.collison@arm.com>
15581 * config/arm/arm.c (arm_option_override): Deprecate
15582 use of -mstructure-size-boundary.
15583 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15584 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15586 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15589 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15590 Reduce cost estimate for direct moves.
15592 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15595 * config/i386/i386.c (ix86_option_override_internal): Disable
15596 BMI, BMI2 and TBM instructions for -m16.
15598 2017-07-21 Carl Love <cel@us.ibm.com>
15600 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15601 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15602 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15603 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15604 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15605 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15606 VMULOSW): New enum "unspec" values.
15607 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15608 altivec_vmulosw): New patterns.
15609 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15610 VMULOSW): Add definitions.
15612 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15614 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15615 (qdf24xx): Likewise.
15616 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15617 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15618 (AARCH64_FL_V8_1): Renumber.
15619 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15620 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15621 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15622 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15623 rdma to feature modifiers list.
15625 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15627 PR middle-end/56727
15628 * ipa-visibility (function_and_variable_visibility): Convert
15629 recursive PLT call to direct call if appropriate.
15631 2017-07-21 Andrew Pinski <apinski@cavium.com>
15633 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15634 operand 1 to see if the types precision matches.
15635 * fold-const.c (operand_equal_p): Likewise.
15637 2017-07-21 Richard Biener <rguenther@suse.de>
15639 PR tree-optimization/81303
15640 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15641 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15642 (vect_peeling_hash_get_lowest_cost): Adjust.
15643 (vect_enhance_data_refs_alignment): Likewise. Use
15644 vect_get_peeling_costs_all_drs to compute the penalty for no
15645 peeling to match up costs.
15647 2017-07-21 Richard Biener <rguenther@suse.de>
15649 PR tree-optimization/81500
15650 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15651 we didn't identify a reduction path.
15653 2017-07-21 Tom de Vries <tom@codesourcery.com>
15654 Cesar Philippidis <cesar@codesourcery.com>
15656 PR gcov-profile/81442
15657 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15660 2017-07-21 Tom de Vries <tom@codesourcery.com>
15663 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15665 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15666 nvptx_override_options_after_change.
15668 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15670 * dwarf2out.c (output_file_names): Avoid double testing for
15671 dwarf_version >= 5.
15673 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15675 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15677 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15679 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15681 (try_crossjump_to_edge): Do not punt on partitioned functions.
15683 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15685 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15686 Put all BBs reachable only via paths crossing cold region to cold
15688 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15690 2016-07-21 Richard Biener <rguenther@suse.de>
15692 PR tree-optimization/81303
15693 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15694 into account prologue and epilogue iterations when raising
15695 min_profitable_iters to sth at least covering one vector iteration.
15697 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15699 * config/arm/arm.c (arm_test_cpu_arch_dat):
15702 2017-07-20 Nathan Sidwell <nathan@acm.org>
15704 Remove TYPE_METHODS.
15705 * tree.h (TYPE_METHODS): Delete.
15706 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15707 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15708 (dbxout_type_methods): Scan TYPE_FIELDS.
15709 (dbxout_type): Don't check TYPE_METHODS here.
15710 * function.c (use_register_for_decl): Always ignore register for
15711 class types when not optimizing.
15712 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15713 * tree.c (free_lang_data_in_type): Stitch out member functions and
15714 templates from TYPE_FIELDS.
15715 (build_distinct_type_copy, verify_type_variant,
15716 verify_type): Member fns are on TYPE_FIELDS.
15717 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15718 * tree-pretty-print.c (dump_generic_node): Likewise.
15720 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15723 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15724 V2TImode and V4TImode.
15725 (ix86_expand_vector_extract): Likewise.
15726 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15727 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15728 (ssescalarmode): Handle V4TImode and V2TImode.
15729 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15730 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15731 (VEXTRACTI128_MODE): New mode iterator.
15732 (splitter for *vec_extractv?ti first element): New.
15733 (VEC_INIT_MODE): New mode iterator.
15734 (vec_init<mode>): Consolidate 3 expanders into one using
15735 VEC_INIT_MODE mode iterator.
15737 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15739 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15740 non_spilled_static_chain_regno_p.
15742 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15744 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15746 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15748 * bb-reorder.c (connect_traces): Allow copying of blocks within
15751 2017-07-20 Richard Biener <rguenther@suse.de>
15753 * gimple.h (gimple_phi_result): Add gphi * overload.
15754 (gimple_phi_result_ptr): Likewise.
15755 (gimple_phi_arg): Likewise. Adjust index assert to only
15756 allow actual argument accesses rather than all slots available
15758 (gimple_phi_arg_def): Add gphi * overload.
15759 * tree-phinodes.c (make_phi_node): Initialize only actual
15761 (resize_phi_node): Clear memory not covered by old node,
15762 do not initialize excess argument slots.
15763 (reserve_phi_args_for_new_edge): Initialize new argument slot
15766 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15768 PR tree-optimization/81388
15770 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15772 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15773 by removing computation of may_be_zero.
15775 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15776 Tom de Vries <tom@codesourcery.com>
15778 PR middle-end/81030
15779 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15780 when gimple level profile disagrees with what RTL expander did.
15782 2017-07-20 Richard Biener <rguenther@suse.de>
15784 PR tree-optimization/61171
15785 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15786 (vect_analyze_stmt): Add slp instance parameter.
15787 (vectorizable_reduction): Likewise.
15788 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15789 (vect_is_simple_reduction): Deal with chains not detected
15790 as SLP reduction chain, specifically not properly associated
15791 chains containing a mix of plus/minus.
15792 (get_reduction_op): Remove.
15793 (get_initial_defs_for_reduction): Simplify, pass in whether
15794 this is a reduction chain, pass in the SLP node for the PHIs.
15795 (vect_create_epilog_for_reduction): Get the SLP instance as
15797 (vectorizable_reduction): Get the SLP instance as arg.
15798 During analysis remember the SLP node with the PHIs in the
15799 instance. Simplify getting at the vectorized reduction PHIs.
15800 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15801 through SLP instance.
15802 (vect_slp_analyze_operations): Likewise.
15803 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15804 (vect_transform_stmt): Likewise.
15806 2017-07-20 Tom de Vries <tom@codesourcery.com>
15808 PR tree-optimization/81489
15809 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15810 read of phi arg location to before loop that modifies phi.
15812 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15814 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15817 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15819 PR middle-end/81331
15820 * except.c (execute): Fix ordering issue.
15822 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15824 PR rtl-optimization/81423
15825 * combine.c (make_compound_operation_int): Don't try to optimize
15826 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15828 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15830 PR rtl-optimization/81423
15831 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15832 with a constant that is -1 in the truncated to mode.
15834 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15836 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15837 (determine_unlikely_bbs): ... here.
15838 * predict.h (propagate_unlikely_bbs_forward): Declare.
15839 * cfgexpand.c (pass_expand::execute): Use it.
15840 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15842 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15843 propagate_unlikely_bbs_forward.
15845 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15847 PR middle-end/81331
15848 * except.c (maybe_add_nop_after_section_switch): New function.
15851 2017-07-19 Tom de Vries <tom@codesourcery.com>
15853 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15855 2017-07-19 Tom de Vries <tom@codesourcery.com>
15857 * gimple.h (gimple_phi_arg): Make assert more strict.
15859 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15861 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15862 * config/rs6000/mmintrin.h: New file.
15863 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15865 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15867 PR tree-optimization/81346
15868 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15870 2017-07-19 Tom de Vries <tom@codesourcery.com>
15872 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15874 2017-07-19 Tom de Vries <tom@codesourcery.com>
15876 * config/nvptx/nvptx-modes.def: Add V2DImode.
15877 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15878 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15879 (nvptx_output_mov_insn): Handle lack of mov.b128.
15880 (nvptx_print_operand): Handle 'H' and 'L' codes.
15881 (nvptx_vector_mode_supported): Allow V2DImode.
15882 (nvptx_preferred_simd_mode): New function.
15883 (nvptx_data_alignment): New function.
15884 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15885 nvptx_preferred_simd_mode.
15886 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15888 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15890 2017-07-19 Tom de Vries <tom@codesourcery.com>
15892 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15893 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15894 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15895 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15896 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15897 (mov<VECIM>_insn): New define_insn.
15898 (define_expand "mov<VECIM>): New define_expand.
15900 2017-07-19 Tom de Vries <tom@codesourcery.com>
15902 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15904 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15906 PR tree-optimization/81346
15907 * fold-const.h (fold_div_compare, range_check_type): Declare.
15908 * fold-const.c (range_check_type): New function.
15909 (build_range_check): Use range_check_type.
15910 (fold_div_compare): No longer static, rewritten into
15911 a match.pd helper function.
15912 (fold_comparison): Don't call fold_div_compare here.
15913 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15914 as helper function.
15916 2017-07-19 Nathan Sidwell <nathan@acm.org>
15918 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15919 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15920 * tree.c (find_decls_types_r, verify_type): Use
15921 TYPE_{MIN,MAX}_VALUE_RAW.
15922 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15923 (hash_tree): Likewise.
15924 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15926 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15929 2017-07-18 Tom de Vries <tom@codesourcery.com>
15931 PR middle-end/81464
15932 * omp-expand.c (expand_omp_for_static_chunk): Handle
15933 equal-argument loop exit phi.
15935 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15938 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15939 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15940 operand 2 predicate.
15941 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15942 operand 2 predicate.
15943 (ror,rol -> rorx splitters): Use const_int_operand as
15944 operand 2 predicate.
15946 2017-06-18 Richard Biener <rguenther@suse.de>
15948 PR tree-optimization/81410
15949 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15950 the gap in the ! slp_perm SLP case after each group.
15952 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15954 PR middle-end/81463
15955 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15958 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15960 PR middle-end/81462
15961 * predict.c (set_even_probabilities): Cleanup; do not affect
15962 probabilities that are already known.
15963 (combine_predictions_for_bb): Call even when count is set.
15965 2017-07-18 Nathan Sidwell <nathan@acm.org>
15967 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15970 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15973 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15974 optimization for loop niter analysis.
15976 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15979 * config/avr/avr.c (avr_optimize_casesi): Don't use
15980 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15982 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15984 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15985 body_cost_vec from _vect_peel_extended_info.
15986 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15987 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15990 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15992 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15994 2017-07-18 Richard Biener <rguenther@suse.de>
15996 PR tree-optimization/80620
15997 PR tree-optimization/81403
15998 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15999 info when re-using a VN table entry.
16001 2017-07-18 Richard Biener <rguenther@suse.de>
16003 PR tree-optimization/81418
16004 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16005 vectype_in. Verify that with lane-reducing reduction operations
16006 we have a single def-use cycle.
16008 2017-07-17 Carl Love <cel@us.ibm.com>
16010 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16012 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16013 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16014 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16015 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16016 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16017 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16018 VMULOSW): New enum "unspec" values.
16019 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16020 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16021 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16022 altivec_vmulosw): New patterns.
16023 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16024 VMULOSW): Add definitions.
16025 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16027 * config/alpha/alpha.c: Include predict.h.
16029 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16031 * tree-vrp.c (compare_assert_loc): Fix comparison function
16032 to return predictable results.
16034 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16036 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16038 (subdi3): Likewise.
16039 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16040 * doc/invoke.texi (mexpand-adddi): Update text.
16042 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16044 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16045 that also clobbers the CC register. The old expand code is moved
16047 (*arc_clzsi2): ... here.
16048 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16049 the CC register. The old expand code is moved to ...
16050 (arc_ctzsi2): ... here.
16052 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16054 * config/arc/arc.opt (mindexed-loads): Use initial value
16055 TARGET_INDEXED_LOADS_DEFAULT.
16056 (mauto-modify-reg): Use initial value
16057 TARGET_AUTO_MODIFY_REG_DEFAULT.
16058 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16059 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16060 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16061 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16063 2017-07-17 Martin Liska <mliska@suse.cz>
16066 * opts.c (finish_options): Do not allow -fgnu-tm
16067 w/ -fsanitize={kernel-,}address. Say sorry.
16069 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16072 * tree-loop-distribution.c (classify_partition): Only assert on
16073 numer of iterations.
16074 (merge_dep_scc_partitions): Delete prameter. Update function call.
16075 (distribute_loop): Remove code handling loop with unknown niters.
16076 (pass_loop_distribution::execute): Skip loop with unknown niters.
16078 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16081 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16082 function sort_partitions_by_post_order.
16084 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16086 PR tree-optimization/81374
16087 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16088 the max index of basic blocks, rather than number of basic blocks.
16090 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16092 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16094 (arc_legitimate_pic_operand_p): Likewise.
16095 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16097 (arc_needs_pcl_p): Likewise.
16098 (arc_legitimate_pc_offset_p): Likewise.
16099 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16100 function is also used in constrains.md.
16101 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16102 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16103 PLUS. Only return true/false in known cases, otherwise assert.
16104 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16105 is already called in arc_legitimate_constant_p.
16106 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16108 (LEGITIMATE_PIC_OPERAND_P): Use
16109 arc_raw_symbolic_reference_mentioned_p function.
16110 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16115 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16116 Andrew Burgess <andrew.burgess@embecosm.com>
16118 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16119 (arc_return_address_register): New function.
16120 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16121 (arc_handle_fndecl_attribute): Add naked attribute.
16122 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16123 (TARGET_WARN_FUNC_RETURN): Likewise.
16124 (arc_allocate_stack_slots_for_args): New function.
16125 (arc_warn_func_return): Likewise.
16126 (machine_function): Change type fn_type.
16127 (arc_compute_function_type): Consider new naked function type,
16128 change function return type.
16129 (arc_must_save_register): Adapt to handle new
16130 arc_compute_function_type's return type.
16131 (arc_expand_prologue): Likewise.
16132 (arc_expand_epilogue): Likewise.
16133 (arc_return_address_regs): Delete.
16134 (arc_return_address_register): New function.
16135 (arc_epilogue_uses): Use above function.
16136 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16137 (arc_function_type): Change encoding, add naked type.
16138 (ARC_INTERRUPT_P): Change to handle the new encoding.
16139 (ARC_FAST_INTERRUPT_P): Likewise.
16140 (ARC_NORMAL_P): Define.
16141 (ARC_NAKED_P): Likewise.
16142 (arc_compute_function_type): Delete prototype.
16143 * config/arc/arc.md (in_ret_delay_slot): Use
16144 arc_return_address_register function.
16145 (simple_return): Likewise.
16146 (p_return_i): Likewise.
16148 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16150 PR tree-optimization/81428
16151 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16152 can't be built for those types.
16154 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16156 Remove stuff dead since r239246.
16158 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16159 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16160 (avr_inform_devices): Remove dead stuff.
16162 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16164 * config/arm/arm_neon.h: Fix softp typo.
16166 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16168 PR tree-optimization/81365
16169 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16170 aggregate moves onto bb predecessor edges, make sure there are no
16171 loads that could alias the lhs in between the start of bb and the
16174 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16177 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16178 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16179 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16181 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16183 PR tree-optimization/81396
16184 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16185 (init_symbolic_number): Initialize it to 1.
16186 (perform_symbolic_merge): Add n_ops from both operands into the new
16188 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16189 without base_addr as useless if they need more than one operation.
16190 (bswap_replace): Handle !bswap case for NULL base_addr.
16192 2017-07-17 Tom de Vries <tom@codesourcery.com>
16195 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16198 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16200 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16201 conditional builtin define __FIX_LEON3FT_B2BST.
16203 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16205 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16206 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16209 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16211 PR rtl-optimization/81424
16212 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16213 to remove potential trapping from operands if -fnon-call-exceptions.
16215 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16217 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16218 profile_proability for scalling.
16219 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16221 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16223 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16225 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16227 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16228 fixpoint arithmetics.
16230 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16232 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16233 fixpoint arithmetics.
16235 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16237 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16238 fixpoint arithmetics.
16240 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16242 * profile-count.h (profile_probability::from_reg_br_prob_note,
16243 profile_probability::to_reg_br_prob_note): New functions.
16244 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16245 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16246 * predict.c (probability_reliable_p): Update.
16247 (edge_probability_reliable_p): Update.
16248 (br_prob_note_reliable_p): Update.
16249 (invert_br_probabilities): Update.
16250 (add_reg_br_prob_note): New function.
16251 (combine_predictions_for_insn): Update.
16252 * asan.c (asan_clear_shadow): Update.
16253 * cfgbuild.c (compute_outgoing_frequencies): Update.
16254 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16255 (update_br_prob_note): Update.
16256 (rtl_verify_edges): Update.
16257 (purge_dead_edges): Update.
16258 (fixup_reorder_chain): Update.
16259 * emit-rtl.c (try_split): Update.
16260 * ifcvt.c (cond_exec_process_insns): Update.
16261 (cond_exec_process_if_block): Update.
16262 (dead_or_predicable): Update.
16263 * internal-fn.c (expand_addsub_overflow): Update.
16264 (expand_neg_overflow): Update.
16265 (expand_mul_overflow): Update.
16266 * loop-doloop.c (doloop_modify): Update.
16267 * loop-unroll.c (compare_and_jump_seq): Update.
16268 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16269 * predict.h: Update.
16270 * reorg.c (mostly_true_jump): Update.
16272 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16273 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16274 * config/arc/arc.c: (emit_unlikely_jump): Update.
16275 * config/arm/arm.c: (emit_unlikely_jump): Update.
16276 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16277 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16278 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16279 (ix86_print_operand): Update.
16280 (ix86_split_fp_branch): Update.
16281 (predict_jump): Update.
16282 * config/ia64/ia64.c (ia64_print_operand): Update.
16283 * config/mmix/mmix.c (mmix_print_operand): Update.
16284 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16285 (rs6000_expand_split_stack_prologue): Update.
16286 * config/rs6000/rs6000.c: Update.
16287 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16288 (s390_expand_vec_movstr): Update.
16289 (s390_expand_cs_tdsi): Update.
16290 (s390_expand_split_stack_prologue): Update.
16291 * config/sh/sh.c (sh_print_operand): Update.
16292 (expand_cbranchsi4): Update.
16293 (expand_cbranchdi4): Update.
16294 * config/sparc/sparc.c (output_v9branch): Update.
16295 * config/spu/spu.c (get_branch_target): Update.
16296 (ea_load_store_inline): Update.
16297 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16298 * config/tilepro/tilepro.c: Update.
16300 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16302 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16303 (gimplify_save_expr): Likewise.
16305 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16307 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16309 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16311 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16313 * timevar.def (TV_IPA_FNSUMMARY): Define.
16315 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16317 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16318 to back store errata sensitive sequence from being generated.
16319 (sqrtdf2_fix): Likewise.
16321 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16323 * tree-ssa-threadupdate.c (compute_path_counts,
16324 update_joiner_offpath_counts): Use profile_probability.
16326 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16329 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16331 * config/arm/arm-c.c (arm_cpu_builtins): Define
16332 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16334 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16336 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16337 array entries to represent __ieee128 versions of the
16338 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16339 scalar_extract_sig, and scalar_insert_exp built-in functions.
16340 (altivec_resolve_overloaded_builtin): Add special case handling
16341 for the __builtin_scalar_insert_exp function, as represented by
16342 the P9V_BUILTIN_VEC_VSIEDP constant.
16343 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16344 exponent support for __ieee128 argument.
16345 (VSESQP): Add scalar extract signature support for __ieee128
16347 (VSTDCNQP): Add scalar test negative support for __ieee128
16349 (VSIEQP): Add scalar insert exponent support for __int128 argument
16350 with __ieee128 result.
16351 (VSIEQPF): Add scalar insert exponent support for __ieee128
16352 argument with __ieee128 result.
16353 (VSTDCQP): Add scalar test data class support for __ieee128
16355 (VSTDCNQP): Add overload support for scalar test negative with
16356 __ieee128 argument.
16357 (VSTDCQP): Add overload support for scalar test data class
16358 __ieee128 argument.
16359 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16360 UNSPEC_VSX_SXSIGDP.
16361 (UNSPEC_VSX_SIEXPQP): New constant.
16362 (xsxexpqp): New insn for VSX scalar extract exponent quad
16364 (xsxsigqp): New insn for VSX scalar extract significand quad
16366 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16367 precision with floating point argument.
16368 (xststdcqp): New expand for VSX scalar test data class quad
16370 (xststdcnegqp): New expand for VSX scalar test negative quad
16372 (xststdcqp): New insn to match expansions for VSX scalar test data
16373 class quad precision and VSX scalar test negative quad precision.
16374 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16375 special case operand checking to enforce that second operand of
16376 VSX scalar test data class with quad precision argument is a 7-bit
16378 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16379 prototypes and descriptions of __ieee128 versions of
16380 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16381 scalar_test_data_class, and scalar_test_neg built-in functions.
16383 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16385 PR tree-optimization/81162
16386 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16387 replace a negate with an add.
16389 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16391 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16393 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16395 * config/arm/arm-c.c (arm_cpu_builtins): Define
16396 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16398 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16400 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16401 (armv8-r): Set ARM Cortex-R52 as default CPU.
16402 * config/arm/arm-tables.opt: Regenerate.
16403 * config/arm/arm-tune.md: Regenerate.
16404 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16406 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16407 extension for -mcpu=cortex-r52.
16409 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16411 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16412 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16413 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16414 (fp-armv8): Define it as FP_ARMv8 only.
16415 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16416 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16418 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16419 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16420 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16421 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16422 than TARGET_FPU_ARMV8.
16423 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16424 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16425 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16427 * config/arm/neon.md (neon_vrint): Likewise.
16428 (neon_vcvt): Likewise.
16429 (neon_<fmaxmin_op><mode>): Likewise.
16430 (<fmaxmin><mode>3): Likewise.
16431 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16432 * config/arm/predicates.md (arm_cond_move_operator): Check against
16433 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16435 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16437 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16438 to top of function.
16440 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16442 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16443 loop in comment with memset.
16445 2017-07-14 Martin Liska <mliska@suse.cz>
16447 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16448 * dwarf2out.c (is_java): Remove the function.
16449 (output_pubname): Remove usage of the function.
16450 (lower_bound_default): Remove usage of DW_LANG_Java.
16451 (gen_compile_unit_die): Likewise.
16452 * gcc.c: Remove compiler defaults for .java and .zip files.
16453 * gimple-expr.c (remove_suffix): Change as there's no longer
16454 extension than 4-letter one.
16455 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16456 (gimplify_save_expr): Likewise.
16457 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16458 as it's possible even for other languages than Java.
16459 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16460 * lto-opts.c (lto_write_options): Remove reference to Java.
16461 * opts.c (strip_off_ending): Update file extension handling.
16462 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16463 * tree-eh.c (lower_resx): Likewise.
16464 * tree.c (free_lang_data_in_type): Remove dead code.
16465 (find_decls_types_r): Likewise.
16466 (build_common_builtin_nodes): Remove Java from a comment.
16467 (verify_type): Remove dead code.
16468 * varasm.c (assemble_external): Remove Java from a comment.
16470 2017-07-14 Martin Liska <mliska@suse.cz>
16472 * opts.c (finish_options): Add quotes.
16473 (common_handle_option): Likewise.
16475 2017-07-14 Martin Liska <mliska@suse.cz>
16477 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16478 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16479 Remove N_SO_PASCAL.
16480 * dwarf2out.c (lower_bound_default): Do not handle
16482 (gen_compile_unit_die): Likewise.
16483 * gcc.c: Remove default extension binding for GNU Pascal.
16484 * stmt.c: Remove Pascal language from a comment.
16485 * xcoffout.c: Likewise.
16487 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16490 * diagnostic-show-locus.c (fixit_cmp): New function.
16491 (layout::layout): Sort m_fixit_hints.
16492 (column_range::column_range): Assert that the values are valid.
16493 (struct char_span): New struct.
16494 (correction::overwrite): New method.
16495 (struct source_line): New struct.
16496 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16497 calls in terms of classes source_line and char_span, and
16498 correction::overwrite.
16499 (selftest::test_overlapped_fixit_printing_2): New function.
16500 (selftest::diagnostic_show_locus_c_tests): Call it.
16502 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16504 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16505 early if there is no lhs.
16507 2017-07-13 Martin Liska <mliska@suse.cz>
16509 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16510 (gen_reference_type_die): Likewise.
16511 * stor-layout.c: Remove Pascal-related comment.
16513 2017-07-13 Martin Liska <mliska@suse.cz>
16515 * opts.c (finish_options): Add quotes to error messages.
16516 (parse_sanitizer_options): Likewise.
16518 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16520 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16522 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16524 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16526 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16528 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16530 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16532 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16535 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16536 provides the hardware capability bits, define the macro
16537 __BUILTIN_CPU_SUPPORTS__.
16538 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16539 if GLIBC does not provide the hardware capability bits. Add a
16540 gcc_unreachable call if the built-in cpu function is neither
16541 __builtin_cpu_is nor __builtin_cpu_supports.
16542 (rs6000_get_function_versions_dispatcher): Change the warning
16543 that an old GLIBC is used which does not export the capability
16544 bits to be an error.
16545 * doc/extend.texi (target_clones attribute): Document the
16546 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16547 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16548 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16549 the macros defined by GCC if the newer GLIBC is available.
16551 2017-07-12 Jeff Law <law@redhat.com>
16553 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16554 remaining includes slightly.
16555 * config/riscv/riscv-builtins.c: Include profile-count.h.
16557 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16560 * config/avr/avr.c (avr_set_current_function): In diagnostic
16561 messages: Quote keywords and (parts of) identifiers.
16562 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16565 2017-07-12 Carl Love <cel@us.ibm.com>
16567 * config/rs6000/rs6000-c.c: Add support for built-in functions
16568 vector bool char vec_revb (vector bool char);
16569 vector bool short vec_revb (vector short char);
16570 vector bool int vec_revb (vector bool int);
16571 vector bool long long vec_revb (vector bool long long);
16572 * doc/extend.texi: Update the built-in documentation file for the
16573 new built-in functions.
16575 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16577 * config/s390/s390.md: Remove movcc splitter.
16579 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16581 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16582 load/store on condition.
16584 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16587 * config/avr/avr.c (avr_encode_section_info)
16588 [progmem && !TREE_READONLY]: Error if progmem object needs
16591 2017-07-11 Michael Collison <michael.collison@arm.com>
16593 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16596 2017-07-11 Carl Love <cel@us.ibm.com>
16598 * config/rs6000/rs6000-c.c: Add support for builtins
16599 vector unsigned int vec_parity_lsbb (vector signed int);
16600 vector unsigned int vec_parity_lsbb (vector unsigned int);
16601 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16602 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16603 vector unsigned long long vec_parity_lsbb (vector signed long long);
16604 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16605 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16606 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16607 * doc/extend.texi: Update the built-in documentation file for the
16608 new built-in functions.
16610 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16612 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16613 (layout::m_primary_loc): New field.
16614 (layout::layout): Initialize new field. Move location filtering
16615 logic from here to...
16616 (layout::maybe_add_location_range): ...this new method. Add
16617 support for filtering to just the lines already specified by other
16619 (layout::will_show_line_p): New method.
16620 (gcc_rich_location::add_location_if_nearby): New method.
16621 (selftest::test_add_location_if_nearby): New test function.
16622 (selftest::diagnostic_show_locus_c_tests): Call it.
16623 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16626 2017-07-11 Tom de Vries <tom@codesourcery.com>
16628 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16629 (bb_first_real_insn): New function.
16630 (nvptx_single): Add extra initialization of broadcasted condition
16633 2017-07-11 Nathan Sidwell <nathan@acm.org>
16635 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16637 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16639 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16640 Binutils doc as TEXI will mess them up.
16641 * doc/invoke.texi (AVR Options): Same here.
16643 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16645 * config/sparc/sparc.opt (mfix-ut700): New option.
16646 (mfix-gr712rc): Likewise.
16647 (sparc_fix_b2bst): New variable.
16648 * doc/invoke.texi (SPARC options): Document them.
16649 (ARM options): Fix warnings.
16650 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16651 instructions to prevent sequences that can trigger the store-store
16652 errata for certain LEON3FT processors.
16653 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16654 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16655 * config/sparc/sparc.md (fix_b2bst): New attribute.
16656 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16658 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16661 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16663 (*rsqrtsf2_sse): Ditto.
16665 (div<mode>3): Macroize insn from divdf3 and divsf3
16666 using MODEF mode iterator.
16668 2017-07-10 Martin Sebor <msebor@redhat.com>
16670 PR tree-optimization/80397
16671 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16672 instead of testing for equality to INTEGER_TYPE.
16674 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16676 * config.gcc: Remove uclibc from arc target spec.
16678 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16680 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16682 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16685 * lto-wrapper.c (remove_option): New function.
16686 (merge_and_complain): Merge PIC/PIE options more realistically.
16688 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16690 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16694 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16695 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16696 * config.in: Regenerate.
16697 * configure: Regenerate.
16698 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16699 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16700 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16701 (TARGET_GASISR_PROLOGUES): ...target mask.
16702 * common/config/avr/avr-common.c
16703 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16704 Set -mgas-isr-prologues.
16705 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16706 INSERT_PASS_BEFORE for it.
16707 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16708 * config/avr/avr.c (avr_option_override)
16709 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16710 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16711 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16712 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16713 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16715 (make_avr_pass_pre_proep): New function.
16716 (emit_push_sfr) <treg>: Add argument to function and use it
16717 instead of TMP_REG.
16718 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16719 and set machine->gasisr.yes.
16720 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16721 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16722 __gcc_isr.n_pushed to .L__stack_usage.
16723 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16724 (avr_asm_final_postscan_insn): ...this new static function.
16725 * config/avr/avr.h (machine_function)
16726 <is_no_gccisr, use_L__stack_usage>: New fields.
16727 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16728 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16729 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16730 (gasisr, *gasisr): New expander and insn.
16731 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16732 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16733 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16735 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16737 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16740 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16742 Move jump-tables out of .text again.
16745 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16746 (ASM_OUTPUT_ADDR_VEC): New function.
16747 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16748 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16749 INSN_ADDRESSes as asm comment.
16750 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16751 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16752 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16753 * config/avr/avr.md (*tablejump): Adjust comment.
16754 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16755 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16757 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16758 (avr_output_addr_vec): New proto.
16759 (avr_log_t) <insn_addresses>: New field.
16761 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16764 * config/i386/i386.c (ix86_function_arg_advance): Set
16765 outgoing_args_on_stack to true if there are outgoing arguments
16767 (ix86_function_arg): Likewise.
16768 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16769 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16770 * config/i386/i386.h (machine_function): Add
16771 outgoing_args_on_stack.
16773 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16775 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16776 supporting pthreds.
16777 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16779 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16781 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16782 (REAL_H): Remove $(MACHMODE_H).
16783 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16785 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16786 $(MACHMODE_H) and double-int.h.
16787 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16789 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16792 2017-07-07 Andrew Pinski <apinski@cavium.com>
16794 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16795 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16797 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16799 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16800 Add warning if GCC was not configured to link against a GLIBC that
16801 exports the hardware capability bits.
16802 (make_resolver_func): Make resolver function private and not a
16803 COMDAT function. Create the name with clone_function_name instead
16804 of make_unique_name.
16807 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16808 correct operand in doing the split.
16810 2017-07-07 Carl Love <cel@us.ibm.com>
16812 * config/rs6000/rs6000-c: Add support for built-in function
16813 vector unsigned short vec_pack_to_short_fp32 (vector float,
16815 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16816 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16817 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16818 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16819 (convert_4f32_8i16): Add define_expand.
16820 * doc/extend.texi: Update the built-in documentation file for the
16821 new built-in function.
16823 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16825 * config/sparc/m8.md: New file.
16826 * config/sparc/sparc.md: Include m8.md.
16828 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16830 * config/sparc/sparc.opt: New option -mvis4b.
16831 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16832 (sparc_option_override): Handle VIS4B.
16833 (enum sparc_builtins): Define
16834 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16835 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16836 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16837 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16838 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16839 (check_constant_argument): New function.
16840 (sparc_vis_init_builtins): Define builtins
16841 __builtin_vis_dictunpack{8,16,32},
16842 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16843 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16844 __builtin_vis_fpcmpde{8,16,32}shl and
16845 __builtin_vis_fpcmpur{8,16,32}shl.
16846 (sparc_expand_builtin): Check that the constant operands to
16847 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16848 constant and in range.
16849 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16851 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16852 (SPARC_IMM5_P): Likewise.
16853 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16854 (enabled): Handle vis4b.
16855 (UNSPEC_DICTUNPACK): New unspec.
16856 (UNSPEC_FPCMPSHL): Likewise.
16857 (UNSPEC_FPUCMPSHL): Likewise.
16858 (UNSPEC_FPCMPDESHL): Likewise.
16859 (UNSPEC_FPCMPURSHL): Likewise.
16860 (cpu_feature): New CPU feature `vis4b'.
16861 (dictunpack{8,16,32}): New insns.
16862 (FPCSMODE): New mode iterator.
16863 (fpcscond): New code iterator.
16864 (fpcsucond): Likewise.
16865 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16866 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16867 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16868 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16869 * config/sparc/constraints.md: Define constraints `q' for unsigned
16870 2-bit integer constants and `t' for unsigned 5-bit integer
16872 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16874 (imm5_operand_dictunpack16): Likewise.
16875 (imm5_operand_dictunpack32): Likewise.
16876 (imm2_operand): Likewise.
16877 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16878 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16879 ditunpack* and fpcmp*shl builtins.
16881 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16883 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16884 * config.in: Add HAVE_AS_SPARC6 define.
16885 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16887 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16889 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16890 (CPP_CPU_SPEC): Handle m8.
16891 (ASM_CPU_SPEC): Likewise.
16892 * config/sparc/sparc-opts.h (enum processor_type): Add
16894 * config/sparc/sparc.c (m8_costs): New struct.
16895 (sparc_option_override): Handle TARGET_CPU_m8.
16896 (sparc32_initialize_trampoline): Likewise.
16897 (sparc64_initialize_trampoline): Likewise.
16898 (sparc_issue_rate): Likewise.
16899 (sparc_register_move_cost): Likewise.
16900 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16901 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16902 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16903 (CPP_CPU_SPEC): Handle M8.
16904 (ASM_CPU_SPEC): Likewise.
16905 (AS_M8_FLAG): Define.
16906 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16907 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16908 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16910 * configure: Regenerate.
16911 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16914 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16916 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16918 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16919 ("*movdi_insn_sp32"): Do not set v3pipe.
16920 ("*movsi_insn"): Likewise.
16921 ("*movdi_insn_sp64"): Likewise.
16922 ("*movsf_insn"): Likewise.
16923 ("*movdf_insn_sp32"): Likewise.
16924 ("*movdf_insn_sp64"): Likewise.
16925 ("*zero_extendsidi2_insn_sp64"): Likewise.
16926 ("*sign_extendsidi2_insn"): Likewise.
16927 ("*mov<VM32:mode>_insn"): Likewise.
16928 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16929 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16930 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16931 ("<vlop:code><VL:mode>3"): Likewise.
16932 ("*not_<vlop:code><VL:mode>3"): Likewise.
16933 ("*nand<VL:mode>_vis"): Likewise.
16934 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16935 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16936 ("one_cmpl<VL:mode>2"): Likewise.
16937 ("faligndata<VM64:mode>_vis"): Likewise.
16938 ("alignaddrsi_vis"): Likewise.
16939 ("alignaddrdi_vis"): Likweise.
16940 ("alignaddrlsi_vis"): Likewise.
16941 ("alignaddrldi_vis"): Likewise.
16942 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16943 ("bmaskdi_vis"): Likewise.
16944 ("bmasksi_vis"): Likewise.
16945 ("bshuffle<VM64:mode>_vis"): Likewise.
16946 ("cmask8<P:mode>_vis"): Likewise.
16947 ("cmask16<P:mode>_vis"): Likewise.
16948 ("cmask32<P:mode>_vis"): Likewise.
16949 ("pdistn<P:mode>_vis"): Likewise.
16950 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16952 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16954 * config/sparc/sparc.md ("subtype"): New insn attribute.
16955 ("*wrgsr_sp64"): Set insn subtype.
16956 ("*rdgsr_sp64"): Likewise.
16957 ("alignaddrsi_vis"): Likewise.
16958 ("alignaddrdi_vis"): Likewise.
16959 ("alignaddrlsi_vis"): Likewise.
16960 ("alignaddrldi_vis"): Likewise.
16961 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16962 ("fexpand_vis"): Likewise.
16963 ("fpmerge_vis"): Likewise.
16964 ("faligndata<VM64:mode>_vis"): Likewise.
16965 ("bshuffle<VM64:mode>_vis"): Likewise.
16966 ("cmask8<P:mode>_vis"): Likewise.
16967 ("cmask16<P:mode>_vis"): Likewise.
16968 ("cmask32<P:mode>_vis"): Likewise.
16969 ("fchksm16_vis"): Likewise.
16970 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16971 ("fmean16_vis"): Likewise.
16972 ("fp<plusminus_insn>64_vis"): Likewise.
16973 ("<plusminus_insn>v8qi3"): Likewise.
16974 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16975 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16976 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16977 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16978 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16979 ("*movqi_insn"): Likewise.
16980 ("*movhi_insn"): Likewise.
16981 ("*movsi_insn"): Likewise.
16982 ("movsi_pic_gotdata_op"): Likewise.
16983 ("*movdi_insn_sp32"): Likewise.
16984 ("*movdi_insn_sp64"): Likewise.
16985 ("movdi_pic_gotdata_op"): Likewise.
16986 ("*movsf_insn"): Likewise.
16987 ("*movdf_insn_sp32"): Likewise.
16988 ("*movdf_insn_sp64"): Likewise.
16989 ("*zero_extendhisi2_insn"): Likewise.
16990 ("*zero_extendqihi2_insn"): Likewise.
16991 ("*zero_extendqisi2_insn"): Likewise.
16992 ("*zero_extendqidi2_insn"): Likewise.
16993 ("*zero_extendhidi2_insn"): Likewise.
16994 ("*zero_extendsidi2_insn_sp64"): Likewise.
16995 ("ldfsr"): Likewise.
16996 ("prefetch_64"): Likewise.
16997 ("prefetch_32"): Likewise.
16998 ("tie_ld32"): Likewise.
16999 ("tie_ld64"): Likewise.
17000 ("*tldo_ldub_sp32"): Likewise.
17001 ("*tldo_ldub1_sp32"): Likewise.
17002 ("*tldo_ldub2_sp32"): Likewise.
17003 ("*tldo_ldub_sp64"): Likewise.
17004 ("*tldo_ldub1_sp64"): Likewise.
17005 ("*tldo_ldub2_sp64"): Likewise.
17006 ("*tldo_ldub3_sp64"): Likewise.
17007 ("*tldo_lduh_sp32"): Likewise.
17008 ("*tldo_lduh1_sp32"): Likewise.
17009 ("*tldo_lduh_sp64"): Likewise.
17010 ("*tldo_lduh1_sp64"): Likewise.
17011 ("*tldo_lduh2_sp64"): Likewise.
17012 ("*tldo_lduw_sp32"): Likewise.
17013 ("*tldo_lduw_sp64"): Likewise.
17014 ("*tldo_lduw1_sp64"): Likewise.
17015 ("*tldo_ldx_sp64"): Likewise.
17016 ("*mov<VM32:mode>_insn"): Likewise.
17017 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17018 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17020 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17022 * config/sparc/sparc.md ("type"): New insn type viscmp.
17023 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17025 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17026 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17027 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17028 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17030 ("n7_vis_logical_11cycle"): Likewise.
17031 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17032 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17033 * config/sparc/niagara.md ("niag_vis"): Likewise.
17034 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17035 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17037 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17039 * config/sparc/sparc.md: New instruction type `bmask'.
17040 (bmaskdi_vis): Use the `bmask' type.
17041 (bmasksi_vis): Likewise.
17042 * config/sparc/ultra3.md (us3_array): Likewise.
17043 * config/sparc/niagara7.md (n7_array): Likewise.
17044 * config/sparc/niagara4.md (n4_array): Likewise.
17045 * config/sparc/niagara2.md (niag2_vis): Likewise.
17046 (niag3_vis): Likewise.
17047 * config/sparc/niagara.md (niag_vis): Likewise.
17049 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17051 * ipa-comdats.c: Remove optimize check from gate.
17052 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17053 for functions not optimized.
17054 (ipa_fn_summary_read): Skip optimize check.
17055 (ipa_fn_summary_write): Likewise.
17056 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17058 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17060 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17061 (check_callers): Give up on caller which is not optimized.
17062 (inline_small_functions): Likewise.
17063 (ipa_inline): Do not give up when not optimizing.
17064 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17065 away unoptimizes cdtors.
17066 (whole_program_function_and_variable_visibility): Do
17067 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17068 * ipa.c (process_references): Do not check optimize.
17069 (symbol_table::remove_unreachable_nodes): Update optimize check.
17070 (set_writeonly_bit): Update optimize check.
17071 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17072 (pass_ipa_single_use::gate): Remove.
17074 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17076 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17077 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17078 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17079 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17080 permute_load, permute_store, adjust_extract, adjust_splat,
17081 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17082 replace_swap_with_copy, dump_swap_insn_table,
17083 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17084 recombine_lvx_pattern, recombine_stvx_pattern,
17085 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17086 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17087 to file rs6000-p8swap.c.
17088 * config/rs6000/rs6000-p8swap.c: New file.
17089 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17090 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17091 and rs6000*-*-* targets.
17093 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17095 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17097 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17099 * lto-wrapper.c (merge_and_complain): Do not merge
17100 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17101 fsigned_zeros, ftrapping_math, fwrapv.
17102 (append_compiler_options): Do not track these options.
17103 (append_linker_options): Likewie
17105 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17107 * cgraphunit.c (cgraph_node::finalize_function): When
17108 !flag_toplevel_reorde set no_reorder flag.
17109 (varpool_node::finalize_decl): Likewise.
17110 (symbol_table::compile): Drop no toplevel reorder path.
17112 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17114 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17115 edges; zero probability is not better than uninitialized.
17117 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17119 * asan.h (asan_sanitize_allocas_p): Declare.
17120 * asan.c (asan_sanitize_allocas_p): New function.
17121 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17122 (handle_builtin_alloca): Likewise.
17123 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17124 if !asan_sanitize_allocas_p.
17125 * params.def (asan-instrument-allocas): Add new option.
17126 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17127 * opts.c (common_handle_option): Disable allocas sanitization for
17130 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17132 * asan.c: Include gimple-fold.h.
17133 (get_last_alloca_addr): New function.
17134 (handle_builtin_stackrestore): Likewise.
17135 (handle_builtin_alloca): Likewise.
17136 (asan_emit_allocas_unpoison): Likewise.
17137 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17138 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17139 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17140 (instrument_builtin_call): Pass gimple iterator to
17141 get_mem_refs_of_builtin_call.
17142 (last_alloca_addr): New global.
17143 * asan.h (asan_emit_allocas_unpoison): Declare.
17144 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17145 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17146 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17147 if function calls alloca.
17148 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17149 * gimple-fold.h (replace_call_with_value): Declare.
17150 * internal-fn.c: Include asan.h.
17151 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17152 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17154 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17156 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17157 (C_SELFTEST_FLAGS): New.
17158 (CPP_SELFTEST_FLAGS): New.
17159 (SELFTEST_DEPS): New, from deps of s-selftest.
17160 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17161 (CPP_SELFTEST_DEPS): New.
17162 (selftest): Add dependency on s-selftest-c++.
17163 (s-selftest): Rename to...
17164 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17165 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17166 than SELFTEST_FLAGS.
17167 (selftest-gdb): Rename to...
17168 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17170 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17171 (selftest-valgrind): Rename to...
17172 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17174 (selftest-valgrind): Reintroduce as an alias for
17175 selftest-c-valgrind.
17176 (s-selftest-c++): New.
17177 (selftest-c++-gdb): New.
17178 (selftest-c++-valgrind): New.
17180 2017-07-06 Olivier Hainque <hainque@adacore.com>
17182 * gcc.c (process_command): When deciding if undefined variables
17183 should be ignored when processing specs, accept "gcc -v" as well.
17185 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17187 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17188 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17190 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17192 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17193 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17194 * config/arm/arm-tables.opt: Regenerate.
17195 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17197 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17199 2017-07-06 Carl Love <cel@us.ibm.com>
17201 * ChangeLog: Clean up from mid air collision
17203 2017-07-06 Carl Love <cel@us.ibm.com>
17205 * config/rs6000/rs6000-c.c: Add support for built-in functions
17206 vector signed int vec_subc (vector signed int, vector signed int);
17207 vector signed __int128 vec_subc (vector signed __int128,
17208 vector signed __int128);
17209 vector unsigned __int128 vec_subc (vector unsigned __int128,
17210 vector unsigned __int128);
17211 vector signed int vec_sube (vector signed int, vector signed int,
17212 vector signed int);
17213 vector unsigned int vec_sube (vector unsigned int,
17214 vector unsigned int,
17215 vector unsigned int);
17216 vector signed __int128 vec_sube (vector signed __int128,
17217 vector signed __int128,
17218 vector signed__int128);
17219 vector unsigned __int128 vec_sube (vector unsigned __int128,
17220 vector unsigned __int128,
17221 vector unsigned __int128);
17222 vector signed int vec_subec (vector signed int, vector signed int,
17223 vector signed int);
17224 vector unsigned int vec_subec (vector unsigned int,
17225 vector unsigned int,
17226 vector unsigned int);
17227 vector signed __int128 vec_subec (vector signed __int128,
17228 vector signed __int128,
17229 vector signed__int128);
17230 vector unsigned __int128 vec_subec (vector unsigned __int128,
17231 vector unsigned __int128,
17232 vector unsigned __int128);
17233 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17234 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17235 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17236 BU_ALTIVEC_OVERLOAD_X definitions.
17237 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17238 * doc/extend.texi: Update the built-in documentation file for the new
17239 built-in functions.
17241 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17244 * diagnostic-show-locus.c (layout::layout): Use start and finish
17245 spelling location for the start and finish of each range.
17246 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17247 Add unused aspect param.
17248 * input.c (expand_location_1): Add "aspect" param, and use it
17249 to access the correct part of the location.
17250 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17252 (expand_location_to_spelling_point): Likewise.
17253 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17254 param, and pass it to expand_location_1.
17256 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17258 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17259 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17260 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17261 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17262 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17263 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17264 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17265 _mm_maskz_getmant_ss): New intrinsics.
17266 (__builtin_ia32_getexpss128_mask): Changed to ...
17267 __builtin_ia32_getexpss128_round ... this.
17268 (__builtin_ia32_getexpsd128_mask): Changed to ...
17269 __builtin_ia32_getexpsd128_round ... this.
17270 * config/i386/i386-builtin-types.def
17271 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17272 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17273 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17274 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17275 __builtin_ia32_getmantss_mask_round): New builtins.
17276 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17277 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17278 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17279 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17280 * config/i386/sse.md
17281 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17282 avx512f_sgetexp<mode><mask_scalar_name>
17283 <round_saeonly_scalar_name> ... this.
17284 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17285 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17286 vgetexp<ssescalarmodesuffix>
17287 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17288 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17289 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17290 avx512f_vgetmant<mode><mask_scalar_name>
17291 <round_saeonly_scalar_name> ... this.
17292 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17293 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17294 vgetmant<ssescalarmodesuffix>
17295 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17296 %0<mask_scalar_operand4>, %1, %2
17297 <round_saeonly_scalar_mask_op4>, %3} ... this.
17298 * config/i386/subst.md (mask_scalar_operand4,
17299 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17300 round_saeonly_scalar_nimm_predicate): New subst attributes.
17302 2017-07-06 Julia Koval <julia.koval@intel.com>
17304 * config/i386/i386.c (ix86_erase_embedded_rounding):
17305 Remove code for old rounding pattern.
17307 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17309 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17311 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17313 * doc/sourcebuild.texi (Test Directives, Variants of
17314 dg-require-support): Add documentation for dg-require-stack-check.
17316 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17318 * config/i386/subst.md (mask_scalar, round_scalar,
17319 round_saeonly_scalar): New meta-templates.
17320 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17321 round_scalar_mask_operand3, round_scalar_mask_op3,
17322 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17323 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17324 round_saeonly_scalar_constraint,
17325 round_saeonly_scalar_prefix): New subst attribute.
17326 * config/i386/sse.md
17327 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17328 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17329 <round_scalar_name> ... this.
17330 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17331 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17332 <round_scalar_name> ... this.
17333 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17334 <sse>_vm<code><mode>3<mask_scalar_name>
17335 <round_saeonly_scalar_name> ... this.
17336 (v<plusminus_mnemonic><ssescalarmodesuffix>
17337 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17338 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17339 v<plusminus_mnemonic><ssescalarmodesuffix>
17340 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17341 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17342 (v<multdiv_mnemonic><ssescalarmodesuffix>
17343 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17344 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17345 v<multdiv_mnemonic><ssescalarmodesuffix>
17346 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17347 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17348 (v<maxmin_float><ssescalarmodesuffix>
17349 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17350 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17351 v<maxmin_float><ssescalarmodesuffix>
17352 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17353 %0<mask_scalar_operand3>, %1, %<iptr>2
17354 <round_saeonly_scalar_mask_op3>} ... this.
17356 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17358 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17359 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17361 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17362 Alan Hayward <alan.hayward@arm.com>
17363 David Sherwood <david.sherwood@arm.com>
17365 * combine.c (simplify_if_then_else): Remove "enum" before
17367 * compare-elim.c (can_eliminate_compare): Likewise.
17368 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17370 (aarch64_lookup_simd_builtin_type): Likewise.
17371 (aarch64_simd_builtin_type): Likewise.
17372 (aarch64_init_simd_builtin_types): Likewise.
17373 (aarch64_simd_expand_args): Likewise.
17374 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17376 (aarch64_reverse_mask): Likewise.
17377 (aarch64_simd_emit_reg_reg_move): Likewise.
17378 (aarch64_gen_adjusted_ldpstp): Likewise.
17379 (aarch64_ccmp_mode_to_code): Likewise.
17380 (aarch64_operands_ok_for_ldpstp): Likewise.
17381 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17382 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17384 (aarch64_min_divisions_for_recip_mul): Likewise.
17385 (aarch64_reassociation_width): Likewise.
17386 (aarch64_get_condition_code_1): Likewise.
17387 (aarch64_simd_emit_reg_reg_move): Likewise.
17388 (aarch64_simd_attr_length_rglist): Likewise.
17389 (aarch64_reverse_mask): Likewise.
17390 (aarch64_operands_ok_for_ldpstp): Likewise.
17391 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17392 (aarch64_gen_adjusted_ldpstp): Likewise.
17393 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17395 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17396 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17397 (arm_lookup_simd_builtin_type): Likewise.
17398 (arm_simd_builtin_type): Likewise.
17399 (arm_init_simd_builtin_types): Likewise.
17400 (arm_expand_builtin_args): Likewise.
17401 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17402 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17403 (ft32_setup_incoming_varargs): Likewise.
17404 (ft32_function_arg): Likewise.
17405 (ft32_function_arg_advance): Likewise.
17406 (ft32_pass_by_reference): Likewise.
17407 (ft32_arg_partial_bytes): Likewise.
17408 (ft32_valid_pointer_mode): Likewise.
17409 (ft32_addr_space_pointer_mode): Likewise.
17410 (ft32_addr_space_legitimate_address_p): Likewise.
17411 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17413 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17414 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17415 (iamcu_alignment): Likewise.
17416 (canonicalize_vector_int_perm): Likewise.
17417 (ix86_noce_conversion_profitable_p): Likewise.
17418 (ix86_mpx_bound_mode): Likewise.
17419 (ix86_operands_ok_for_move_multiple): Likewise.
17420 * config/microblaze/microblaze-protos.h
17421 (microblaze_expand_conditional_branch_reg): Likewise.
17422 * config/microblaze/microblaze.c
17423 (microblaze_expand_conditional_branch_reg): Likewise.
17424 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17426 (rs6000_reassociation_width): Likewise.
17427 (rs6000_invalid_binary_op): Likewise.
17428 (fusion_p9_p): Likewise.
17429 (emit_fusion_p9_load): Likewise.
17430 (emit_fusion_p9_store): Likewise.
17431 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17433 (riscv_hard_regno_mode_ok_p): Likewise.
17434 (riscv_address_insns): Likewise.
17435 (riscv_split_symbol): Likewise.
17436 (riscv_legitimize_move): Likewise.
17437 (riscv_function_value): Likewise.
17438 (riscv_hard_regno_nregs): Likewise.
17439 (riscv_expand_builtin): Likewise.
17440 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17441 (riscv_build_integer): Likewise.
17442 (riscv_split_integer): Likewise.
17443 (riscv_legitimate_constant_p): Likewise.
17444 (riscv_cannot_force_const_mem): Likewise.
17445 (riscv_regno_mode_ok_for_base_p): Likewise.
17446 (riscv_valid_base_register_p): Likewise.
17447 (riscv_valid_offset_p): Likewise.
17448 (riscv_valid_lo_sum_p): Likewise.
17449 (riscv_classify_address): Likewise.
17450 (riscv_legitimate_address_p): Likewise.
17451 (riscv_address_insns): Likewise.
17452 (riscv_load_store_insns): Likewise.
17453 (riscv_force_binary): Likewise.
17454 (riscv_split_symbol): Likewise.
17455 (riscv_force_address): Likewise.
17456 (riscv_legitimize_address): Likewise.
17457 (riscv_move_integer): Likewise.
17458 (riscv_legitimize_const_move): Likewise.
17459 (riscv_legitimize_move): Likewise.
17460 (riscv_address_cost): Likewise.
17461 (riscv_subword): Likewise.
17462 (riscv_output_move): Likewise.
17463 (riscv_canonicalize_int_order_test): Likewise.
17464 (riscv_emit_int_order_test): Likewise.
17465 (riscv_function_arg_boundary): Likewise.
17466 (riscv_pass_mode_in_fpr_p): Likewise.
17467 (riscv_pass_fpr_single): Likewise.
17468 (riscv_pass_fpr_pair): Likewise.
17469 (riscv_get_arg_info): Likewise.
17470 (riscv_function_arg): Likewise.
17471 (riscv_function_arg_advance): Likewise.
17472 (riscv_arg_partial_bytes): Likewise.
17473 (riscv_function_value): Likewise.
17474 (riscv_pass_by_reference): Likewise.
17475 (riscv_setup_incoming_varargs): Likewise.
17476 (riscv_print_operand): Likewise.
17477 (riscv_elf_select_rtx_section): Likewise.
17478 (riscv_save_restore_reg): Likewise.
17479 (riscv_for_each_saved_reg): Likewise.
17480 (riscv_register_move_cost): Likewise.
17481 (riscv_hard_regno_mode_ok_p): Likewise.
17482 (riscv_hard_regno_nregs): Likewise.
17483 (riscv_class_max_nregs): Likewise.
17484 (riscv_memory_move_cost): Likewise.
17485 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17486 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17487 (rl78_addr_space_address_mode): Likewise.
17488 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17490 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17491 (rs6000_reassociation_width): Likewise.
17492 (rs6000_invalid_binary_op): Likewise.
17493 (fusion_p9_p): Likewise.
17494 (emit_fusion_p9_load): Likewise.
17495 (emit_fusion_p9_store): Likewise.
17496 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17497 (ok_for_simple_move_operands): Likewise.
17498 (ok_for_simple_move_strict_operands): Likewise.
17499 (ok_for_simple_arith_logic_operands): Likewise.
17500 (visium_legitimize_reload_address): Likewise.
17501 (visium_select_cc_mode): Likewise.
17502 (output_cbranch): Likewise.
17503 (visium_split_double_move): Likewise.
17504 (visium_expand_copysign): Likewise.
17505 (visium_expand_int_cstore): Likewise.
17506 (visium_expand_fp_cstore): Likewise.
17507 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17508 (visium_function_arg): Likewise.
17509 (visium_function_arg_advance): Likewise.
17510 (visium_libcall_value): Likewise.
17511 (visium_setup_incoming_varargs): Likewise.
17512 (visium_legitimate_constant_p): Likewise.
17513 (visium_legitimate_address_p): Likewise.
17514 (visium_legitimize_address): Likewise.
17515 (visium_secondary_reload): Likewise.
17516 (visium_register_move_cost): Likewise.
17517 (visium_memory_move_cost): Likewise.
17518 (prepare_move_operands): Likewise.
17519 (ok_for_simple_move_operands): Likewise.
17520 (ok_for_simple_move_strict_operands): Likewise.
17521 (ok_for_simple_arith_logic_operands): Likewise.
17522 (visium_function_value_1): Likewise.
17523 (rtx_ok_for_offset_p): Likewise.
17524 (visium_legitimize_reload_address): Likewise.
17525 (visium_split_double_move): Likewise.
17526 (visium_expand_copysign): Likewise.
17527 (visium_expand_int_cstore): Likewise.
17528 (visium_expand_fp_cstore): Likewise.
17529 (visium_split_cstore): Likewise.
17530 (visium_select_cc_mode): Likewise.
17531 (visium_split_cbranch): Likewise.
17532 (output_cbranch): Likewise.
17533 (visium_print_operand_address): Likewise.
17534 * expmed.c (flip_storage_order): Likewise.
17535 * expmed.h (emit_cstore): Likewise.
17536 (flip_storage_order): Likewise.
17537 * genrecog.c (validate_pattern): Likewise.
17538 * hsa-gen.c (gen_hsa_addr): Likewise.
17539 * internal-fn.c (expand_arith_overflow): Likewise.
17540 * ira-color.c (allocno_copy_cost_saving): Likewise.
17541 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17542 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17543 (process_invariant_for_inheritance): Likewise.
17544 * lra-eliminations.c (move_plus_up): Likewise.
17545 * omp-low.c (lower_oacc_reductions): Likewise.
17546 * simplify-rtx.c (simplify_subreg): Likewise.
17547 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17548 (TARGET_CHKP_BOUND_MODE): Likewise..
17549 * targhooks.c (default_chkp_bound_mode): Likewise.
17550 (default_setup_incoming_vararg_bounds): Likewise.
17551 * targhooks.h (default_chkp_bound_mode): Likewise.
17552 (default_setup_incoming_vararg_bounds): Likewise.
17553 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17554 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17555 (have_whole_vector_shift): Likewise.
17556 * tree-vect-stmts.c (vectorizable_load): Likewise.
17557 * doc/tm.texi: Regenerate.
17559 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17561 Graceful degrade if Binutils PR21472 is not available.
17564 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17565 .rodata in flash test fails.
17566 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17567 * confgure: Regenerate.
17568 * config.in: Regenerate.
17569 * config/avr/avr.c (avr_asm_named_section)
17570 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17571 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17572 (avr_asm_init_sections): Same.
17574 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17576 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17577 (fma<VH:mode>4_intrinsic): Likewise.
17578 (*fmsub<VCVTF:mode>4): Likewise.
17579 (*fmsub<VH:mode>4_intrinsic): Likewise.
17581 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17584 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17585 Don't depend on "optimize > 0".
17586 (out_movhi_r_mr, out_movqi_mr_r): Same.
17587 (out_movhi_mr_r, out_movqi_r_mr): Same.
17588 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17589 io_address_operand on "optimize > 0".
17591 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17593 * tree-loop-distribution.c: Add general explanantion on the pass.
17594 (generate_loops_for_partition): Mark distributed loop.
17595 (pg_add_dependence_edges): New parameter. Handle alias data
17596 dependence specially and record it in the parameter if asked.
17597 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17598 (init_partition_graph_vertices, add_partition_graph_edge): New.
17599 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17600 (free_partition_graph_vdata, build_partition_graph): New.
17601 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17602 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17603 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17604 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17605 (version_for_distribution_p, finalize_partitions): New.
17606 (distribute_loop): Handle alias data dependence specially. Factor
17607 out loop fusion code as functions and call these functions.
17609 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17611 * tree-loop-distribution.c (classify_partition): New parameter and
17612 better handle reduction statement.
17613 (rdg_build_partitions): Revise comment.
17614 (distribute_loop): Compute statements in all partitions and pass it
17615 to classify_partition.
17617 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17619 * tree-loop-distribution.c (enum partition_type): New.
17620 (struct partition): New field type.
17621 (partition_merge_into): Add parameter. Update partition type.
17622 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17623 (build_rdg_partition_for_vertex): Compute partition type.
17624 (rdg_build_partitions): Dump partition type.
17625 (distribute_loop): Update calls to partition_merge_into.
17627 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17629 * tree-loop-distribution.c (struct ddr_hasher): New.
17630 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17632 (classify_partition): Call get_data_dependence.
17633 (pg_add_dependence_edges): Ditto.
17634 (distribute_loop): Release data dependence hash table.
17636 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17638 * tree-loop-distribution.c (ref_base_address): Delete.
17639 (similar_memory_accesses): Rename ...
17640 (share_memory_accesses): ... to this. Check if partitions access
17641 the same memory reference.
17642 (distribute_loop): Call share_memory_accesses.
17644 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17646 * tree-loop-distribution.c (struct partition): New field recording
17647 its data reference.
17648 (partition_alloc, partition_free): Init and release data refs.
17649 (partition_merge_into): Merge data refs.
17650 (build_rdg_partition_for_vertex): Collect data refs for partition.
17651 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17653 (distribute_loop): Remve data refs from vertice data of partition
17656 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17658 * tree-loop-distribution.c (params.h): Include header file.
17659 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17660 (datarefs_vec): New global var.
17661 (create_rdg_vertices): Use datarefs_vec directly.
17662 (free_rdg): Don't free data references.
17663 (build_rdg): Update use. Don't free data references.
17664 (distribute_loop): Compute global variable for data references.
17665 Bail out if there are too many data references.
17667 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17669 * tree-loop-distribution.c (loop_nest): New global var.
17670 (build_rdg): Use loop directly, rather than loop nest.
17671 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17673 (distribute_loop): Compute global variable loop nest. Update use.
17675 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17677 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17678 (partition_merge_into): New parameter. Dump reason for fusion.
17679 (distribute_loop): Update use of partition_merge_into.
17681 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17683 * tree-loop-distribution.c (bb_top_order_index): New.
17684 (bb_top_order_index_size, bb_top_order_cmp): New.
17685 (stmts_from_loop): Use topological order.
17686 (pass_loop_distribution::execute): Compute and release topological
17687 order for basic blocks.
17689 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17691 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17694 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17696 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17697 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17698 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17699 * internal-fn.def (LOOP_DIST_ALIAS): New.
17700 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17701 (fold_loop_internal_call): ... this.
17702 (vect_loop_dist_alias_call): New function.
17703 (set_uid_loop_bbs): Call fold_loop_internal_call.
17704 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17707 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17710 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17711 Require dead FLAGS_REG at the beginning of a peephole.
17713 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17716 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17717 arguments in the call to __builtin_ia32_sbb_u32.
17718 (_subborrow_u64): Swap _X and _Y arguments in the call to
17719 __builtin_ia32_sbb_u64.
17721 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17724 * tree-vrp.c (compare_assert_loc): Turn into a function template
17725 with stable template parameter. Only test if a->e is NULL,
17726 !a->e == !b->e has been verified already. Use e == NULL or
17727 e != NULL instead of e or ! e tests. If stable is true, don't use
17728 iterative_hash_expr, on the other side allow a or b or both NULL
17729 and sort the NULLs last.
17730 (process_assert_insertions): Sort using compare_assert_loc<false>
17731 instead of compare_assert_loc, later sort using
17732 compare_assert_loc<true> before calling process_assert_insertions_for
17733 in a loop. Use break instead of continue once seen NULL pointer.
17735 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17737 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17738 Cortex-R7 and Cortex-R8 processors.
17740 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17742 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17743 uninitialized while src is not.
17745 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17747 * common/config/arm/arm-common.c: Adjust include path for
17749 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17750 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17751 (arm-cpu-data.h): Likewise.
17752 (arm-cpu-cdata.h): Likewise.
17753 * config/arm/arm-cpu.h: Delete.
17754 * config/arm/arm-cpu-cdata.h: Delete.
17755 * config/arm/arm-cpu-data.h: Delete.
17757 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17759 * config/arm/arm-cpus.in (cortex-a55): New.
17760 (cortex-a75): Likewise.
17761 (cortex-a75.cortex-a55): Likewise.
17762 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17764 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17765 * config/arm/arm-cpu-cdata.h: Regenerate.
17766 * config/arm/arm-cpu-data.h: Regenerate.
17767 * config/arm/arm-cpu.h: Regenerate.
17768 * config/arm/arm-tables.opt: Regenerate.
17769 * config/arm/arm-tune.md: Regenerate.
17771 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17773 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17775 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17777 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17780 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17782 PR tree-optimization/81292
17783 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17784 full_string_p, also call adjust_related_strinfos if the adjustment
17785 is simple, otherwise invalidate related strinfos.
17787 2017-07-04 Martin Liska <mliska@suse.cz>
17790 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17791 newly created variable as DECL_IGNORED_P.
17793 2017-07-04 Martin Liska <mliska@suse.cz>
17796 * ipa-inline.c (inline_small_functions):
17797 Use xstrdup_for_dump.
17799 2017-07-04 Tom de Vries <tom@codesourcery.com>
17801 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17803 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17806 * config/darwin.c (darwin_function_switched_text_sections):
17809 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17811 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17813 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17815 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17817 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17819 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17820 min_profitable_iters, and th as inclusive lower bounds.
17821 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17822 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17823 for min_profitable_iters and min_profitable_estimate.
17824 (vect_transform_loop): Treat th as an inclusive lower bound.
17825 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17827 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17830 * config/darwin.c (darwin_function_switched_text_sections):
17831 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17832 in two pieces, and suppress the use of buf.
17834 2017-07-03 Nathan Sidwell <nathan@acm.org>
17836 * hash-table.h (hash_table_mod1): Fix indentation.
17838 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17840 PR middle-end/81290
17841 * predict.c (force_edge_cold): Be more careful about propagation
17843 * profile-count.h (profile_probability::guessed,
17844 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17846 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17848 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17850 * doc/invoke.texi (rcpc architecture extension): Document it.
17852 2017-07-03 Richard Biener <rguenther@suse.de>
17854 PR tree-optimization/60510
17855 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17856 the scalar reduction PHI and use it.
17857 (vectorizable_reduction): Properly guard the single_defuse_cycle
17858 path for non-SLP reduction chains where we cannot use it.
17859 Rework reduc_def/index and vector type deduction. Rework
17860 vector operand gathering during reduction op code-gen.
17861 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17862 chains dissolve the chain and leave it to non-SLP reduction
17865 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17867 * tree-data-ref.h (dr_alignment): Declare.
17868 * tree-data-ref.c (dr_alignment): New function.
17869 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17870 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17872 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17874 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17876 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17877 and base_misalignment fields.
17878 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17879 * tree-data-ref.c: Include builtins.h.
17880 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17881 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17882 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17883 * tree-vect-data-refs.c: Include tree-cfg.h.
17884 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17885 fields instead of calculating an alignment here.
17886 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17887 innermost_loop_behavior fields.
17889 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17891 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17893 (DR_STEP_ALIGNMENT): New macro.
17894 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17895 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17896 (create_data_ref): Print it.
17897 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17898 to tell whether the step preserves vector (mis)alignment.
17899 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17900 Move the check for an integer step and generalise to all INTEGER_CST.
17901 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17902 Print the outer step alignment.
17904 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17906 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17907 with offset_alignment.
17908 (DR_ALIGNED_TO): Delete.
17909 (DR_OFFSET_ALIGNMENT): New macro.
17910 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17911 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17912 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17913 (create_data_ref): Likewise.
17914 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17915 (vect_analyze_data_refs): Likewise.
17916 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17917 creating dummy innermost behavior.
17919 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17921 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17922 with a "innermost_loop_behavior *" and refeence tree.
17923 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17924 (create_data_ref): Update call accordingly.
17925 * tree-predcom.c (find_looparound_phi): Likewise.
17927 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17929 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17930 fields with dr_wrt_vec_loop.
17931 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17932 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17933 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17934 (vect_dr_behavior): New function.
17935 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17936 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17937 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17938 track whether the step preserves the misalignment.
17939 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17940 Use vect_dr_behavior.
17941 (vect_setup_realignment): Update call accordingly.
17942 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17943 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17944 call to vect_create_addr_base_for_vector_ref.
17945 (vect_create_cond_for_align_checks): Likewise.
17946 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17947 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17948 (vect_recog_mask_conversion_pattern): Likewise.
17949 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17950 (new_stmt_vec_info): Remove redundant zeroing.
17952 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17954 * common/config/arm/arm-common.c (arm_be8_option): New function.
17955 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17956 (ISA_ARMv6): Add isa_bit_be8.
17957 * config/arm/arm.h (arm_be8_option): Add prototype.
17958 (BE8_SPEC_FUNCTION): New define.
17959 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17960 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17961 (mlittle-endian): Similarly.
17962 (mbe8, mbe32): New options.
17963 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17964 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17966 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17968 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17970 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17972 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17973 (cleanup_tree_cfg_bb): Use it.
17974 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17976 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17978 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17981 * loop-doloop.c (add_test): Update profile.
17983 2017-07-03 Martin Liska <mliska@suse.cz>
17986 * sanopt.c (rewrite_usage_of_param): New function.
17987 (sanitize_rewrite_addressable_params): Likewise.
17988 (pass_sanopt::execute): Call rewrite_usage_of_param.
17990 2017-07-03 Richard Biener <rguenther@suse.de>
17992 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17993 back to using VIEW_CONVERT_EXPR.
17995 2017-07-03 Martin Liska <mliska@suse.cz>
17998 * doc/extend.texi: Document when a resolver function is
17999 generated for target_clones.
18001 2017-07-03 Martin Liska <mliska@suse.cz>
18003 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18004 and shadow memory of auto variables which are subject of
18005 use-after-scope sanitization.
18006 (asan_expand_mark_ifn): Add do set only when is_poison.
18008 2016-07-03 Richard Biener <rguenther@suse.de>
18010 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18012 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18013 (vectorizable_reduction): Perform reduction PHI creation when
18014 visiting a reduction PHI and adjust and simplify code generation
18015 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18016 (vect_transform_loop): Visit reduction PHIs.
18017 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18018 defs into the SLP tree.
18019 (vect_build_slp_tree): Reduction defs terminate the recursion.
18020 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18022 (vect_get_vec_defs_for_stmt_copy): Export.
18023 (vect_get_vec_defs): Likewise.
18024 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18026 (vect_get_vec_defs_for_stmt_copy): Declare.
18027 (vect_get_vec_defs): Likewise.
18029 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18031 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18032 parameter with a "loop" parameter and use it instead of the
18033 loop containing DR_STMT. Don't check simple_iv when doing
18034 BB analysis. Describe the two analysis modes in the comment.
18036 2017-07-03 Tom de Vries <tom@codesourcery.com>
18038 PR tree-optimization/69468
18039 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18040 (find_same_succ_bb): Handle ignore_edge_flags.
18042 2017-07-03 Tom de Vries <tom@codesourcery.com>
18044 PR tree-optimization/81192
18045 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18047 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18049 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18051 2017-07-03 Tom de Vries <tom@codesourcery.com>
18053 PR tree-optimization/81192
18054 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18055 BB_SAME_SUCC (bb) == NULL.
18057 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18059 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18062 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18064 * dumpfile.c: Include profile-count.h
18065 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18067 (insert_cond_bb): Update profile.
18068 * tree-cfg.h (insert_cond_bb): Update prototype.
18069 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18070 * tree-dump.c: Do not include tree-cfg.
18072 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18074 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18076 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18078 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18081 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18083 * tree-complex.c (expand_complex_div_wide): update profile.
18085 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18086 Alan Hayward <alan.hayward@arm.com>
18087 David Sherwood <david.sherwood@arm.com>
18089 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18090 (CORETYPES_H): New define.
18091 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18092 (insn-modes-inline.h, s-modes-inline-h): New rules.
18093 (generated_files): Add insn-modes-inline.h.
18094 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18095 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18096 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18097 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18098 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18099 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18100 (build/gencodes.o, build/genconditions.o): Likewise.
18101 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18102 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18103 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18104 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18105 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18106 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18107 * coretypes.h: Include everything up to real.h for generators.
18108 Include insn-modes.h first. Include wide-int-print.h after
18109 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18110 * machmode.h: Don't include insn-modes.h here.
18111 * function-tests.c: Remove includes of signop.h, machmode.h,
18112 double-int.h and wide-int.h.
18114 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18116 * optc-save-gen.awk: Likewise.
18117 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18118 * godump.c: Remove include of wide-int-print.h.
18119 * pretty-print.h: Likewise.
18120 * wide-int-print.cc: Likewise.
18121 * wide-int.cc: Likewise.
18122 * hash-map-tests.c: Remove include of signop.h.
18123 * hash-set-tests.c: Likewise.
18124 * rtl-tests.c: Likewise.
18125 * mkconfig.sh: Remove include of machmode.h.
18126 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18128 (emit_insn_modes_inline_h): ...this new function. Emit the code
18129 into an insn-modes-inline.h header file, adding appropriate
18130 include guards and end comments.
18131 (emit_insn_modes_c_header): Remove include of machmode.h.
18132 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18134 (main): Handle -i flag and call emit_insn_modes_inline_h when
18137 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18139 * tree-ssa-strlen.c (strinfo): Rename the length field to
18140 nonzero_chars. Add a full_string_p field.
18141 (compare_nonzero_chars, zero_length_string_p): New functions.
18142 (get_addr_stridx): Add an offset_out parameter.
18143 Use compare_nonzero_chars.
18144 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18145 (new_strinfo): Update after above changes to strinfo.
18146 (set_endptr_and_length): Set full_string_p.
18147 (get_string_length): Update after above changes to strinfo.
18148 (unshare_strinfo): Update call to new_strinfo.
18149 (maybe_invalidate): Likewise.
18150 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18151 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18152 as a uhwi instead of an shwi. Update after above changes to
18153 strinfo and new_strinfo.
18154 (zero_length_string): Assert that chainsi contains full strings.
18155 Use zero_length_string_p. Update call to new_strinfo.
18156 (adjust_related_strinfos): Update after above changes to strinfo.
18157 Copy full_string_p from origsi.
18158 (adjust_last_stmt): Use zero_length_string_p.
18159 (handle_builtin_strlen): Update after above changes to strinfo and
18160 new_strinfo. Install the lhs as the string length if the previous
18161 entry didn't describe a full string.
18162 (handle_builtin_strchr): Update after above changes to strinfo
18164 (handle_builtin_strcpy): Likewise.
18165 (handle_builtin_strcat): Likewise.
18166 (handle_builtin_malloc): Likewise.
18167 (handle_pointer_plus): Likewise.
18168 (handle_builtin_memcpy): Likewise. Track nonzero characters
18169 that aren't necessarily followed by a nul terminator.
18170 (handle_char_store): Likewise.
18172 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18174 PR tree-optimization/80769
18175 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18176 for malloc and calloc. Document the new invariant that all related
18177 strinfos have delayed lengths or none do.
18178 (verify_related_strinfos): Move earlier in file.
18179 (set_endptr_and_length): New function, split out from...
18180 (get_string_length): ...here. Also set the lengths of related
18182 (zero_length_string): Assert that chainsi has known (rather than
18184 (adjust_related_strinfos): Likewise.
18186 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18188 PR tree-optimization/81136
18189 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18190 assert that two references with the same misalignment have the same
18191 compile-time misalignment if those compile-time misalignments
18194 2017-07-01 Andi Kleen <ak@linux.intel.com>
18196 * print-tree.c (print_node): Print all attributes.
18198 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18200 * cfg.c (scale_bbs_frequencies): New function.
18201 * cfg.h (scale_bbs_frequencies): Declare it.
18202 * cfgloopanal.c (single_likely_exit): Cleanup.
18203 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18205 (scale_loop_profile): Likewise.
18206 (loop_version): Likewise.
18207 (create_empty_loop_on_edge): Update.
18208 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18209 scale_loop_frequencies, scale_loop_profile, loopify,
18210 loop_version): Update prototypes.
18211 * modulo-sched.c (sms_schedule): Update.
18212 * predict.c (unlikely_executed_edge_p): Also check probability.
18213 (probably_never_executed_edge_p): Fix typo.
18214 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18215 * tree-parloops.c (gen_parallel_loop): Update.
18216 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18217 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18218 * tree-ssa-loop-split.c (split_loop): Update.
18219 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18220 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18221 (vect_loop_versioning): Update.
18222 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18224 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18226 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18228 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18230 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18231 to keep profile consistent.
18233 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18235 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18236 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18237 * profile-count.h (max_safe_multiplier): Make unsigned.
18238 (profile_count::guessed_zero): New.
18240 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18242 * bb-reorder.c (fix_up_crossing_landing_pad,
18243 fix_crossing_conditional_branches): Use make_single_succ_edge
18244 to keep profile consistent.
18246 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18248 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18251 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18254 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18255 the right scopes, make sure cond_jump isn't preserved between multiple
18256 iterations. Search for fallthru edge whenever there are 3+ edges and
18257 use find_fallthru_edge for it.
18259 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18261 Patch by Alexander Monakov <amonakov@ispras.ru>
18262 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18263 probabilities consistently.
18265 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18267 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18268 emit_cmp_and_jump_insns.
18270 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18273 * tree-inline.c (expand_call_inline): Combine profile statuses.
18275 2017-06-30 Andrew Pinski <apinski@cavium.com>
18277 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18278 fold_stmt returned true.
18280 2017-06-30 Nathan Sidwell <nathan@acm.org>
18282 * ggc.h (empty_string): Delete.
18283 * cfgexpand.c (expand_asm_stmt): Use plain "".
18284 * optabs.c (expand_asm_memory_barrier): Likewise.
18285 * stringpool.c (empty_string): Delete.
18286 (digit_vector, digit_string): Delete.
18287 (ggc_alloc_string): Use plain "", don't optimize single digit
18288 strings. Use ggc_alloc_atomic.
18290 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18292 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18293 comparison set and one other set, use the cost of the non-comparison
18296 2017-06-30 Nathan Sidwell <nathan@acm.org>
18298 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18301 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18303 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18304 loops. Remove now unneeded calls to gimple_switch_set_label() that
18305 just set removed labels to NULL_TREE.
18307 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18309 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18310 (set_range_info): ...here. Only call set_range_info_raw if domain
18312 (set_nonzero_bits): Call set_range_info_raw.
18313 * tree-ssanames.h (set_range_info_raw): New.
18315 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18318 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18319 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18320 of nonimmediate_operand and <store_mask_constraint> instead of m
18321 for the input operand. For V8FI iterator, always split if input
18322 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18323 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18324 <store_mask_predicate> instead of register_operand and
18325 <store_mask_constraint> instead of v for the input operand. Make
18326 sure both operands aren't MEMs for if not <mask_applied>.
18328 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18330 * lto-wrapper.c (copy_file) Close both file descriptors before
18333 2017-06-30 Martin Liska <mliska@suse.cz>
18336 * multiple_target.c (create_dispatcher_calls): Make ifunc
18337 also for function that don't have calls or are not referenced.
18339 2017-06-30 Richard Biener <rguenther@suse.de>
18341 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18342 analyze the first scalar stmt. Move vector type computation
18343 for the BB case here from ...
18344 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18345 live operation processing in the SLP case properly.
18347 2017-06-30 Richard Biener <rguenther@suse.de>
18349 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18351 2017-06-30 Martin Liska <mliska@suse.cz>
18354 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18355 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18357 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18359 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18362 2017-06-30 Richard Biener <rguenther@suse.de>
18364 PR tree-optimization/81249
18365 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18366 condition reduction result to original scalar type.
18368 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18370 * profile-count.h (enum profile_quality): Fix typos and whitespace
18373 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18375 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18376 type for branch probabilities.
18378 2017-06-29 Julian Brown <julian@codesourcery.com>
18379 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18381 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18382 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18383 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18384 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18386 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18388 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18389 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18390 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18391 CC usage from generic code to here.
18392 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18393 CC usage into the target macros.
18395 2017-06-29 Maya Rashish <coypu@sdf.org>
18397 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18400 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18402 * arm/arm-builtins.c: Include profile-count.h
18403 * except.c (sjlj_emit_function_enter): Use
18404 profile_probability::unlikely.
18406 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18408 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18409 and tocrel_offset be pointer args rather than implicitly using
18411 (legitimate_constant_pool_address_p, rs6000_emit_move,
18412 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18413 tocrel_offset and use in toc_relative_expr_p call.
18414 (print_operand, print_operand_address): Use static tocrel_base_oac
18415 and tocrel_offset_oac.
18416 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18419 2017-06-29 Maya Rashish <coypu@sdf.org>
18421 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18423 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18425 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18426 objects, take into account only the alignment of 'op0' and 'mode1' if
18429 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18431 * ccmp.c (ccmp_tree_comparison_p): New function.
18432 (ccmp_candidate_p): Update to use above function.
18433 (get_compare_parts): New function.
18434 (expand_ccmp_next): Update to use new functions.
18435 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18437 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18438 take mode as argument.
18439 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18440 * expr.c (expand_expr_real_1): Pass mode as argument.
18442 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18444 * combine.c (combine_instructions): Print insns to dump_file, together
18447 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18449 * asan.c (asan_emit_stack_protection): Update.
18450 (create_cond_insert_point): Update.
18451 * auto-profile.c (afdo_propagate_circuit): Update.
18452 * basic-block.h (struct edge_def): Turn probability to
18453 profile_probability.
18454 (EDGE_FREQUENCY): Update.
18455 * bb-reorder.c (find_traces_1_round): Update.
18456 (better_edge_p): Update.
18457 (sanitize_hot_paths): Update.
18458 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18459 (make_single_succ_edge): Update.
18460 (check_bb_profile): Update.
18461 (dump_edge_info): Update.
18462 (update_bb_profile_for_threading): Update.
18463 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18464 probabilitycount to 0.
18465 * cfgbuild.c (compute_outgoing_frequencies): Update.
18466 * cfgcleanup.c (try_forward_edges): Update.
18467 (outgoing_edges_match): Update.
18468 (try_crossjump_to_edge): Update.
18469 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18470 (expand_gimple_tailcall): Update.
18471 (construct_init_block): Use make_single_succ_edge.
18472 (construct_exit_block): Use make_single_succ_edge.
18473 * cfghooks.c (verify_flow_info): Update.
18474 (redirect_edge_succ_nodup): Update.
18475 (split_edge): Update.
18476 (account_profile_record): Update.
18477 * cfgloopanal.c (single_likely_exit): Update.
18478 * cfgloopmanip.c (scale_loop_profile): Update.
18479 (set_zero_probability): Remove.
18480 (duplicate_loop_to_header_edge): Update.
18481 * cfgloopmanip.h (loop_version): Update prototype.
18482 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18483 (force_nonfallthru_and_redirect): Update.
18484 (update_br_prob_note): Update.
18485 (rtl_verify_edges): Update.
18486 (purge_dead_edges): Update.
18487 (rtl_lv_add_condition_to_bb): Update.
18488 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18489 * cgraphunit.c (init_lowered_empty_function): Update.
18490 (cgraph_node::expand_thunk): Update.
18491 * cilk-common.c: Include profile-count.h
18492 * dojump.c (inv): Remove.
18493 (jumpifnot): Update.
18494 (jumpifnot_1): Update.
18495 (do_jump_1): Update.
18497 (do_jump_by_parts_greater_rtx): Update.
18498 (do_compare_rtx_and_jump): Update.
18499 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18500 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18501 * dwarf2cfi.c: Include profile-count.h
18502 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18503 (sjlj_emit_dispatch_table): Likewise.
18504 * explow.c: Include profile-count.h
18505 * expmed.c (emit_store_flag_force): Update.
18506 (do_cmp_and_jump): Update.
18507 * expr.c (compare_by_pieces_d::generate): Update.
18508 (compare_by_pieces_d::finish_mode): Update.
18509 (emit_block_move_via_loop): Update.
18510 (store_expr_with_bounds): Update.
18511 (store_constructor): Update.
18512 (expand_expr_real_2): Update.
18513 (expand_expr_real_1): Update.
18514 * expr.h (try_casesi, try_tablejump): Update prototypes.
18515 * gimple-pretty-print.c (dump_probability): Update.
18516 (dump_profile): New.
18517 (dump_gimple_label): Update.
18518 (dump_gimple_bb_header): Update.
18519 * graph.c (draw_cfg_node_succ_edges): Update.
18520 * hsa-gen.c (convert_switch_statements): Update.
18521 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18522 (find_if_case_1): Update.
18523 (find_if_case_2): Update.
18524 * internal-fn.c (expand_arith_overflow_result_store): Update.
18525 (expand_addsub_overflow): Update.
18526 (expand_neg_overflow): Update.
18527 (expand_mul_overflow): Update.
18528 (expand_vector_ubsan_overflow): Update.
18529 * ipa-cp.c (good_cloning_opportunity_p): Update.
18530 * ipa-split.c (split_function): Use make_single_succ_edge.
18531 * ipa-utils.c (ipa_merge_profiles): Update.
18532 * loop-doloop.c (add_test): Update.
18533 (doloop_modify): Update.
18534 * loop-unroll.c (compare_and_jump_seq): Update.
18535 (unroll_loop_runtime_iterations): Update.
18536 * lra-constraints.c (lra_inheritance): Update.
18537 * lto-streamer-in.c (input_cfg): Update.
18538 * lto-streamer-out.c (output_cfg): Update.
18539 * mcf.c (adjust_cfg_counts): Update.
18540 * modulo-sched.c (sms_schedule): Update.
18541 * omp-expand.c (expand_omp_for_init_counts): Update.
18542 (extract_omp_for_update_vars): Update.
18543 (expand_omp_ordered_sink): Update.
18544 (expand_omp_for_ordered_loops): Update.
18545 (expand_omp_for_generic): Update.
18546 (expand_omp_for_static_nochunk): Update.
18547 (expand_omp_for_static_chunk): Update.
18548 (expand_cilk_for): Update.
18549 (expand_omp_simd): Update.
18550 (expand_omp_taskloop_for_outer): Update.
18551 (expand_omp_taskloop_for_inner): Update.
18552 * omp-simd-clone.c (simd_clone_adjust): Update.
18553 * optabs.c (expand_doubleword_shift): Update.
18554 (expand_abs): Update.
18555 (emit_cmp_and_jump_insn_1): Update.
18556 (expand_compare_and_swap_loop): Update.
18557 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18558 * predict.c (predictable_edge_p): Update.
18559 (edge_probability_reliable_p): Update.
18560 (set_even_probabilities): Update.
18561 (combine_predictions_for_insn): Update.
18562 (combine_predictions_for_bb): Update.
18563 (propagate_freq): Update.
18564 (estimate_bb_frequencies): Update.
18565 (force_edge_cold): Update.
18566 * profile-count.c (profile_count::dump): Add missing space into dump.
18567 (profile_count::debug): Add newline.
18568 (profile_count::differs_from_p): Explicitly convert to unsigned.
18569 (profile_count::stream_in): Update.
18570 (profile_probability::dump): New member function.
18571 (profile_probability::debug): New member function.
18572 (profile_probability::differs_from_p): New member function.
18573 (profile_probability::differs_lot_from_p): New member function.
18574 (profile_probability::stream_in): New member function.
18575 (profile_probability::stream_out): New member function.
18576 * profile-count.h (profile_count_quality): Rename to ...
18577 (profile_quality): ... this one.
18578 (profile_probability): New.
18579 (profile_count): Update.
18580 * profile.c (compute_branch_probabilities): Update.
18581 * recog.c (peep2_attempt): Update.
18582 * sched-ebb.c (schedule_ebbs): Update.
18583 * sched-rgn.c (find_single_block_region): Update.
18584 (compute_dom_prob_ps): Update.
18585 (schedule_region): Update.
18586 * sel-sched-ir.c (compute_succs_info): Update.
18587 * stmt.c (struct case_node): Update.
18588 (do_jump_if_equal): Update.
18589 (get_outgoing_edge_probs): Update.
18590 (conditional_probability): Update.
18591 (emit_case_dispatch_table): Update.
18592 (expand_case): Update.
18593 (expand_sjlj_dispatch_table): Update.
18594 (emit_case_nodes): Update.
18595 * targhooks.c: Update.
18596 * tracer.c (better_p): Update.
18597 (find_best_successor): Update.
18598 * trans-mem.c (expand_transaction): Update.
18599 * tree-call-cdce.c: Update.
18600 * tree-cfg.c (gimple_split_edge): Upate.
18601 (move_sese_region_to_fn): Upate.
18602 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18603 * tree-eh.c (lower_resx): Upate.
18604 (cleanup_empty_eh_move_lp): Upate.
18605 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18606 * tree-inline.c (copy_edges_for_bb): Update.
18607 (copy_cfg_body): Update.
18608 * tree-parloops.c (gen_parallel_loop): Update.
18609 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18610 (gimple_gen_time_profiler): Update.
18611 * tree-ssa-dce.c (remove_dead_stmt): Update.
18612 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18613 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18614 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18615 (unloop_loops): Update.
18616 (try_peel_loop): Update.
18617 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18618 * tree-ssa-loop-split.c (connect_loops): Update.
18619 (split_loop): Update.
18620 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18621 (hoist_guard): Update.
18622 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18623 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18624 (value_replacement): Update.
18625 * tree-ssa-reassoc.c (branch_fixup): Update.
18626 * tree-ssa-tail-merge.c (replace_block_by): Update.
18627 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18628 (create_edge_and_update_destination_phis): Update.
18629 (compute_path_counts): Update.
18630 (recompute_probabilities): Update.
18631 (update_joiner_offpath_counts): Update.
18632 (freqs_to_counts_path): Update.
18633 (duplicate_thread_path): Update.
18634 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18635 (struct switch_conv_info): Update.
18636 (gen_inbound_check): Update.
18637 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18638 (vect_do_peeling): Update.
18639 (vect_loop_versioning): Update.
18640 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18641 (optimize_mask_stores): Update.
18642 * ubsan.c (ubsan_expand_null_ifn): Update.
18643 * value-prof.c (gimple_divmod_fixed_value): Update.
18644 (gimple_divmod_fixed_value_transform): Update.
18645 (gimple_mod_pow2): Update.
18646 (gimple_mod_pow2_value_transform): Update.
18647 (gimple_mod_subtract): Update.
18648 (gimple_mod_subtract_transform): Update.
18649 (gimple_ic): Update.
18650 (gimple_stringop_fixed_value): Update.
18651 (gimple_stringops_transform): Update.
18652 * value-prof.h: Update.
18654 2017-06-29 Carl Love <cel@us.ibm.com>
18656 * config/rs6000/rs6000-c.c: Add support for built-in functions
18657 vector signed int vec_signed (vector float);
18658 vector signed long long vec_signed (vector double);
18659 vector signed int vec_signed2 (vector double, vector double);
18660 vector signed int vec_signede (vector double);
18661 vector signed int vec_signedo (vector double);
18662 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18663 instruction generator.
18664 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18665 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18666 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18668 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18669 vunsignede_v2df): Add define_expands.
18670 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18671 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18672 VEC_UNSIGNEDO): Add definitions.
18673 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18674 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18675 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18676 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18677 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18678 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18679 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18680 vec_unsignede, vec_unsignedo): Add builtin defines.
18681 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18683 * doc/extend.texi: Update the built-in documentation file for the
18684 new built-in functions.
18686 2017-06-29 Richard Biener <rguenther@suse.de>
18688 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18689 reduction chains to LOOP_VINFO_REDUCTIONS.
18690 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18691 SLP reductions after processing reduction chains.
18693 2017-06-29 Nathan Sidwell <nathan@acm.org>
18695 * builtins.c (fold_builtin_FUNCTION): Use
18696 lang_hooks.decl_printable_name.
18698 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18700 PR middle-end/81194
18701 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18702 with only one label.
18703 * stmt.c (expand_case): Assert NCASES is greater than one.
18705 2017-06-29 Richard Biener <rguenther@suse.de>
18707 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18709 (group_case_labels): Likewise.
18710 (find_taken_edge): Push sanity checking on val to workers...
18711 (find_taken_edge_cond_expr): ... here
18712 (find_taken_edge_switch_expr): ... and here, handle cases
18713 with just a default label.
18714 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18715 (group_case_labels): Likewise.
18716 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18717 group_case_labels does anything cleanup the CFG again.
18719 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18721 PR tree-optimization/81196
18722 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18723 exit condition comparing two IVs.
18725 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18727 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18728 profile to the dummy entry at the end of the list of architectures.
18729 * config/arm/arm-cpu-cdata.h: Regenerated.
18731 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18732 Michael Collison <michael.collison@arm.com>
18735 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18737 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18738 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18739 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18740 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18741 of shift when the shift amount is masked with constant equal to
18742 the size of the mode.
18743 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18746 2017-06-29 Martin Liska <mliska@suse.cz>
18748 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18750 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18753 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18756 (edge_growth_cache_entry::edge_growth_cache_entry): New
18758 (reset_edge_growth_cache): Update to use constructor.
18760 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18762 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18763 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18764 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18766 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18768 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18769 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18771 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18773 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18774 (*-linux-uclibc*): Add t-uclibc tmake_file.
18775 * config/t-musl: New.
18776 * config/t-uclibc: New.
18778 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18780 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18782 (gen_comm_data): Emit architectural setting of arch_prof.
18783 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18785 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18786 (armv8-m.base, armv8-m.main): Likewise.
18787 * arm-protos.h (arm_build_target): Add profile field.
18788 (arch_option): Likewise.
18789 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18791 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18792 arm_active_target.profile.
18794 2017-06-28 Richard Biener <rguenther@suse.de>
18796 PR middle-end/81227
18797 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18798 TYPE_OVERFLOW_WRAPS.
18799 * match.pd (negate_expr_p): Likewise.
18800 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18801 fold_build2, not fold_binary.
18803 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18805 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18806 Convert memory address to Pmode.
18807 (aarch64_print_operand): Assert MEM operands are always Pmode.
18809 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18812 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18813 Remove redundant if.
18814 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18815 * config/arm/aarch-common-protos.h
18816 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18817 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18819 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18822 * multiple_target.c (create_dispatcher_calls): Set the default
18823 clone to be static, not public.
18825 2017-06-28 Richard Biener <rguenther@suse.de>
18827 * tree-vect-loop.c (vectorizable_reduction): Move special
18828 cond reduction IV var creation ...
18829 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18830 parameter. Use STMT_VINFO_VECTYPE.
18831 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18834 2017-06-28 Martin Liska <mliska@suse.cz>
18837 * ipa-visibility.c (non_local_p): Handle visibility.
18839 2017-06-28 Martin Liska <mliska@suse.cz>
18842 * common.opt: Add IntegerRange to various options.
18843 * opt-functions.awk (integer_range_info): New function.
18844 * optc-gen.awk: Add integer_range_info to cl_options struct.
18845 * opts-common.c (decode_cmdline_option): Handle
18846 CL_ERR_INT_RANGE_ARG.
18847 (cmdline_handle_error): Likewise.
18848 * opts.c (print_filtered_help): Show valid interval in
18849 when --help is provided.
18850 * opts.h (struct cl_option): Add range_min and range_max fields.
18851 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18853 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18855 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18856 (x * C EQ/NE y * C): New transformation.
18858 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18860 * genmultilib (combination_space): Accept '+' in option names.
18862 2017-06-28 Martin Liska <mliska@suse.cz>
18865 * asan.c (instrument_derefs): Bail out inner references
18866 that are hard register variables.
18868 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18871 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18872 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18874 2017-06-28 Richard Biener <rguenther@suse.de>
18876 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18877 (vect_get_slp_defs): Adjust.
18878 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18880 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18882 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18883 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18884 (vectorizable_reduction): Adjust.
18885 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18887 (vect_get_slp_defs): Likewise.
18888 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18889 (vectorizable_bswap): Adjust.
18890 (vectorizable_call): Likewise.
18891 (vectorizable_conversion): Likewise.
18892 (vectorizable_assignment): Likewise.
18893 (vectorizable_shift): Likewise.
18894 (vectorizable_operation): Likewise.
18895 (vectorizable_store): Likewise.
18896 (vectorizable_condition): Likewise.
18897 (vectorizable_comparison): Likewise.
18899 2017-06-28 Michael Collison <michael.collison@arm.com>
18902 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18904 (arm_gen_movmemqi): Removed unused variable 'i'.
18905 Convert 'for' loop into 'while' loop.
18906 (arm_expand_prologue): Remove last unnecessary set of insn.
18907 (thumb_pop): Remove unused variable 'pushed_words'.
18908 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18910 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18912 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18913 * config/s390/s390-protos.h: Add prototype of
18914 s390_rel_address_ok_p.
18915 * config/s390/s390.c (s390_got_symbol): New function.
18916 (s390_rel_address_ok_p): New function.
18917 (legitimize_pic_address): Use s390_rel_address_ok_p.
18918 (s390_load_got): Use s390_got_symbol.
18919 (s390_option_override): Issue error if
18920 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18921 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18923 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18925 2017-06-27 Andrew Pinski <apinski@cavium.com>
18927 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18928 (X * copysign (1.0, X)): New pattern.
18929 (X * copysign (1.0, -X)): New pattern.
18930 (copysign (-1.0, CST)): New pattern.
18932 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18934 * genmultilib (combination_space): Remove variable.
18935 Validate reuse rules against regular expression for any sequence
18936 of multilib options in any order.
18938 2017-06-27 Michael Collison <michael.collison@arm.com>
18940 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18941 call aarch64_split_simd_combine.
18942 * (aarch64_combine_internal<mode>): Delete pattern.
18943 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18944 Allow register and subreg operands.
18946 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18948 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18949 specific need, just fallback on defaults.
18950 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18952 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18953 Olivier Hainque <hainque@adacore.com>
18955 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18957 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18958 targets. Pick a default if no particular attempt applied.
18959 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18962 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18964 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18965 (x86_64-wrs-vxworks7): Likewise.
18967 2017-06-27 Marek Polacek <polacek@redhat.com>
18970 * ubsan.c (instrument_null): Check get_base_address's result for null.
18972 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18974 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18976 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18978 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18979 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18980 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18981 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18982 New function types.
18983 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18984 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18985 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18986 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18987 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18988 BUILT_IN_FEUPDATEENV): New builtins.
18989 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18990 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18991 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18992 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18994 (builtin_structptr_types): Adjust size.
18995 * tree.c (builtin_structptr_types): Add four entries.
18997 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18998 Olivier Hainque <hainque@adacore.com>
19000 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19001 (TLS_SYM): New local macro, forcing reference to __tls__ on
19002 link command lines for VxWorks 7 RTPs, triggering initialization
19004 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19005 OS features TLS support, true for RTPs on VxWorks 7.
19006 * config/vxworks.c (vxworks_override_options): Setup emutls
19009 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19011 * predict.c (test_prediction_value_range): Use -1U instead of -1
19012 to avoid narrowing conversion warning.
19013 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19014 to avoid narrowing conversion warning.
19015 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19017 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19019 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19021 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19022 64bit configurations.
19023 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19024 (SIZE_TYPE): Likewise.
19025 * config/vxworks.c (vxworks_emutls_var_fields): Use
19026 long_unsigned_type_node instead of unsigned_type_node as the offset
19027 field type, which is "pointer" mode in emutls.c.
19029 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19032 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19034 PR middle-end/81207
19035 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19036 gimple_vuse copying separately from gimple_vdef copying.
19038 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19040 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19043 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19044 Olivier Hainque <hainque@adacore.com>
19046 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19047 for all vxworks7 targets.
19048 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19049 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19050 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19051 variations for VX6/VX7 and 32/64bits later on in ...
19052 (VXWORKS_LIB_SPEC): Leverage new macros.
19053 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19054 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19056 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19058 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19059 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19061 2017-06-26 Carl Love <cel@us.ibm.com>
19063 * config/rs6000/rs6000-c.c: Add support for built-in functions
19064 vector bool char vec_reve (vector bool char);
19065 vector signed char vec_reve (vector signed char);
19066 vector unsigned char vec_reve (vector unsigned char);
19067 vector bool int vec_reve (vector bool int);
19068 vector signed int vec_reve (vector signed int);
19069 vector unsigned int vec_reve (vector unsigned int);
19070 vector bool long long vec_reve (vector bool long long);
19071 vector signed long long vec_reve (vector signed long long);
19072 vector unsigned long long vec_reve (vector unsigned long long);
19073 vector bool short vec_reve (vector bool short);
19074 vector signed short vec_reve (vector signed short);
19075 vector double vec_reve (vector double);
19076 vector float vec_reve (vector float);
19077 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19078 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19079 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19080 (altivec_vreve): New pattern.
19081 * config/rs6000/altivec.h (vec_reve): New define.
19082 * doc/extend.texi (vec_rev): Update the built-in documentation file
19083 for the new built-in functions.
19085 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19087 PR tree-optimization/71815
19088 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19090 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19092 (slsr_process_phi): Likewise.
19093 (replace_uncond_cands_and_profitable_phis): Don't replace a
19094 multiply candidate with a stride of 1 (copy or cast).
19095 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19097 (lowest_cost_path): Likewise.
19098 (total_savings): Likewise.
19100 2017-06-26 Richard Biener <rguenther@suse.de>
19103 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19104 Use def_builtin_pure for all gather builtins.
19106 2017-06-26 Richard Biener <rguenther@suse.de>
19108 PR tree-optimization/81203
19109 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19110 non-dominating BBs.
19112 2017-06-26 Marek Polacek <polacek@redhat.com>
19115 * doc/invoke.texi: Document -Wmultistatement-macros.
19117 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19119 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19120 arm_neon_ok_no_float_abi effective target.
19122 2017-06-26 Richard Biener <rguenther@suse.de>
19124 PR tree-optimization/80928
19125 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19126 (copy_bbs): Set BB_DUPLICATED flag early.
19127 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19129 (execute_on_shrinking_pred): Likewise.
19130 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19131 BB_DUPLICATED blocks.
19132 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19133 iterate over all PHIs considering removal of *gsi.
19135 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19137 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19140 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19142 * config/rs6000/rs6000-string.c: (expand_block_clear,
19143 do_load_for_compare, select_block_compare_mode,
19144 compute_current_alignment, expand_block_compare,
19145 expand_strncmp_align_check, expand_strn_compare,
19146 expand_block_move, rs6000_output_load_multiple)
19147 Move functions related to string/block move/compare
19148 to a separate file.
19149 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19150 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19151 for this function which is now used in two files.
19152 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19153 * config.gcc: Add rs6000-string.o to extra_objs for
19154 targets powerpc*-*-* and rs6000*-*-*.
19156 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19159 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19160 32-bit, since indexed is not valid for DImode.
19161 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19162 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19163 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19164 (define_peephole2 for Altivec d-form store): Likewise.
19167 * multiple_target.c (create_dispatcher_calls): Only create the
19168 dispatcher call if the function is the default clone of a
19169 versioned function.
19171 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19173 PR middle-end/80902
19174 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19175 a call, force the call to not be a tail call.
19177 2017-06-23 Jeff Law <law@redhat.com>
19179 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19182 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19184 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19185 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19186 handling for early expansion of vector shifts (sl,sr,sra,rl).
19187 (builtin_function_type): Add vector shift right instructions
19188 to the unsigned argument list.
19190 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19192 rtl-optimizatoin/79286
19193 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19194 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19195 trap. PIC register plus a const unspec without offset can never trap.
19197 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19199 * tree.h (builtin_structptr_type): New type.
19200 (builtin_structptr_types): Declare new array.
19201 * tree.c (builtin_structptr_types): New array.
19202 (free_lang_data, build_common_tree_nodes): Use it.
19204 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19207 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19210 2017-06-22 Matt Turner <mattst88@gmail.com>
19212 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19213 Lake models to skylake case. Assume skylake for unknown
19214 models with clflushopt.
19216 2017-06-22 Jeff Law <law@redhat.com>
19218 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19219 frame sizes that do not satisfy aarch64_uimm12_shift.
19221 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19223 * profile-count.h (apply_probability,
19224 apply_scale, probability_in): Fix checks for zero.
19226 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19228 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19229 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19231 2016-06-22 Richard Biener <rguenther@suse.de>
19233 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19234 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19235 REDUC_MAX_EXPR support.
19236 (vectorizable_reduction): Likewise.
19237 (vect_create_epilog_for_reduction): Likewise.
19239 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19241 * match.pd (A / (1 << B) -> A >> B): New.
19242 * generic-match-head.c: Include optabs-tree.h.
19243 * gimple-match-head.c: Likewise.
19244 * optabs-tree.h (target_supports_op_p): New.
19245 * optabs-tree.c (target_supports_op_p): New.
19247 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19249 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19250 $gcc_cv_ld --help output.
19251 (gcc_cv_ld_demangle): Likewise.
19252 (gcc_cv_ld_eh_frame_hdr): Likewise.
19253 (gcc_cv_ld_pie): Likewise.
19254 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19255 (gcc_cv_ld_buildid): Likewise.
19256 (gcc_cv_ld_sysroot): Likewise.
19257 (ld_bndplt_support): Likewise.
19258 (ld_pushpopstate_support): Likewise.
19259 * configure: Regenerate.
19260 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19262 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19265 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19266 operands indexes to avoid gap between operands and match_dups.
19268 2017-06-21 Andrew Pinski <apinski@cavium.com>
19270 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19271 Increment Arith_shift and Arith_shift_reg by 1.
19272 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19274 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19275 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19276 (aarch64_strip_extend): Add new argument and test for it.
19277 (aarch64_cheap_mult_shift_p): New function.
19278 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19279 add a cost if it is true.
19280 Update calls to aarch64_strip_extend.
19281 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19283 2017-06-21 Andrew Pinski <apinski@cavium.com>
19285 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19287 (thunderxt88): Likewise.
19288 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19289 (thunderx_prefetch_tune): New variable.
19290 (thunderx2t99_prefetch_tune): Update for the correct values.
19291 (thunderxt88_tunings): New variable.
19292 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19293 generic_prefetch_tune.
19294 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19296 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19298 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19299 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19300 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19301 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19302 (aarch64_atomic_cas<mode>, GPI): Likewise.
19304 2017-06-21 Martin Liska <mliska@suse.cz>
19306 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19307 statements on cold and hot labels.
19308 * predict.c (tree_estimate_probability_bb): Remove the
19309 prediction from this place.
19311 2017-06-21 Martin Liska <mliska@suse.cz>
19313 PR tree-optimization/79489
19314 * gimplify.c (maybe_add_early_return_predict_stmt): New
19316 (gimplify_return_expr): Call the function.
19317 * predict.c (tree_estimate_probability_bb): Remove handling
19319 * predict.def: Update comment about early return predictor.
19320 * gimple-predict.h (is_gimple_predict): New function.
19321 * predict.def: Change default value of early return to 66.
19322 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19324 * passes.def: Put pass_strip_predict_hints to the beginning of
19327 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19329 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19330 FUNCTION_DECL declarations.
19331 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19333 (dwaf2out_decl): Likewise.
19334 * godump.c (go_early_global_decl): Skip call to the real debug hook
19335 for FUNCTION_DECL declarations.
19336 * passes.c (rest_of_decl_compilation): Skip call to the
19337 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19338 -fdump-go-spec is passed.
19340 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19342 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19343 Reorder for compactness.
19344 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19345 (def_builtin_pure, def_builtin_pure2): New functions.
19346 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19348 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19350 * match.pd (nop_convert): New predicate.
19351 ((A +- CST1) +- CST2): Allow some NOP conversions.
19353 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19356 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19357 with ctors/dtors if GOVD_SHARED is set.
19359 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19361 * config/aarch64/aarch64.md (movti_aarch64):
19362 Emit mov rather than orr.
19363 (movtf_aarch64): Likewise.
19364 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19365 Emit mov rather than orr.
19367 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19369 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19370 Swap alternatives, make integer dup more expensive.
19372 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19374 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19375 Return true for non-tls symbols.
19377 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19379 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19380 (cortex-a75): Likewise.
19381 (cortex-a75.cortex-a55): Likewise.
19382 * config/aarch64/aarch64-tune.md: Regenerate.
19383 * doc/invoke.texi (-mtune): Document new values for -mtune.
19385 2017-06-21 Tom de Vries <tom@codesourcery.com>
19387 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19388 stack_size feature.
19389 (Effective-Target Keywords, Other attributes): Suggest using
19390 dg-add-options stack_size feature to get stack limit in stack_size
19391 effective target documentation.
19393 2017-06-21 Julian Brown <julian@codesourcery.com>
19394 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19396 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19397 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19398 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19400 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19401 attribute type list for neon_multiply.
19402 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19403 attribute type list for neon_multiply.
19404 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19405 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19406 attribute type list for neon_multiply.
19407 * config/arm/types.md (crypto_pmull): Add.
19408 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19409 attribute type list.
19411 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19413 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19416 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19418 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19419 to make sure not to dereference a NULL cost_classes_ptr pointer.
19421 2017-06-20 Carl Love <cel@us.ibm.com>
19423 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19424 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19425 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19426 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19427 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19428 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19429 VMULOSW): New enum "unspec" values.
19430 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19431 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19432 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19433 altivec_vmulosw): New patterns.
19434 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19435 VMULOSW): Add definitions.
19437 2017-06-20 Julia Koval <julia.koval@intel.com>
19439 * config/i386/i386.c: Fix rounding expand for new pattern.
19440 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19442 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19444 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19445 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19447 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19449 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19452 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19454 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19455 architecture, then by implementer ID.
19456 * config/aarch64/aarch64-tune.md: Regenerate.
19458 2017-06-20 Richard Biener <rguenther@suse.de>
19460 PR middle-end/81097
19461 * fold-const.c (split_tree): Fold to type before negating.
19463 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19465 * diagnostic-show-locus.c
19466 (selftest::test_fixit_deletion_affecting_newline): New function.
19467 (selftest::diagnostic_show_locus_c_tests): Call it.
19469 2017-06-20 Andreas Schwab <schwab@suse.de>
19472 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19475 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19477 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19478 __ARM_FEATURE_COPROC according to support.
19480 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19482 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19483 Rewritten to avoid overflow for > 32-bit pointers.
19486 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19487 by removing enum keyword.
19488 (ubsan_type_descriptor): Likewise. Formatting fix.
19491 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19492 splitter): Require TARGET_SSE2 in the condition.
19494 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19497 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19498 for doing vector set of SFmode on ISA 3.0.
19499 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19500 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19502 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19503 SFmode value into a V4SF variable that was extracted from another
19504 V4SF variable without converting the element to double precision
19505 and back to single precision vector format.
19506 (vsx_insert_extract_v4sf_p9_2): Likewise.
19508 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19510 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19511 in UWHI to avoid undefined overflow.
19514 * ubsan.h (enum ubsan_encode_value_phase): New.
19515 (ubsan_encode_value): Change second argument to
19516 enum ubsan_encode_value_phase with default value of
19517 UBSAN_ENCODE_VALUE_GENERIC.
19518 * ubsan.c (ubsan_encode_value): Change second argument to
19519 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19520 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19521 create_tmp_var_raw instead of create_tmp_var and use a
19523 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19524 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19525 ubsan_encode_value callers.
19528 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19529 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19530 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19532 2017-06-19 Richard Biener <rguenther@suse.de>
19534 PR middle-end/81118
19535 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19536 estimates if we changed anything.
19538 2017-06-19 Richard Biener <rguenther@suse.de>
19540 PR tree-optimization/80887
19541 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19542 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19543 simplified lookups, then reset mprts_hook.
19544 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19546 (try_to_simplify): Likewise.
19548 2017-06-19 Martin Liska <mliska@suse.cz>
19551 * gimplify.c (gimplify_switch_expr):
19552 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19554 2017-06-19 Martin Liska <mliska@suse.cz>
19556 * doc/install.texi: Document that PGO runs in 4 stages.
19558 2017-06-19 Martin Liska <mliska@suse.cz>
19561 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19562 to dispatcher function name.
19563 * multiple_target.c (replace_function_decl): New function.
19564 (create_dispatcher_calls): Redirect both edges and references.
19566 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19568 * profile-count.c (profile_count::dump): Dump quality.
19569 (profile_count::differs_from_p): Update for unsigned val.
19570 * profile-count.h (profile_count_quality): New enum.
19571 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19573 2017-06-19 Richard Biener <rguenther@suse.de>
19575 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19576 struct function as arg.
19577 (estimate_numbers_of_iterations): Export overload with loop arg.
19578 (free_numbers_of_iterations_estimates_loop): Use an overload of
19579 free_numbers_of_iterations_estimates instead.
19580 * tree-cfg.c (remove_bb): Adjust.
19581 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19582 * tree-parloops.c (gen_parallel_loop): Likewise.
19583 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19585 (tree_unroll_loops_completely): Likewise.
19586 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19587 Use an overload instead and export.
19588 (estimated_loop_iterations): Adjust.
19589 (max_loop_iterations): Likewise.
19590 (likely_max_loop_iterations): Likewise.
19591 (estimate_numbers_of_iterations): Take struct function as arg
19593 (loop_exits_before_overflow): Adjust.
19594 (free_numbers_of_iterations_estimates_loop): Use an overload.
19595 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19596 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19598 2017-06-19 Richard Biener <rguenther@suse.de>
19601 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19602 RANGE_EXPR conservatively.
19604 2017-06-16 Carl Love <cel@us.ibm.com>
19606 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19607 definitions for vec_float, vec_float2, vec_floato,
19608 vec_floate built-ins.
19609 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19610 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19612 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19613 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19614 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19615 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19616 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19617 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19618 vec_floato): Add builtin defines.
19619 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19620 Update the built-in documentation file for the new built-in
19623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19625 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19626 (mthumb): Mark as the negative of -marm.
19628 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19630 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19632 (ARM Options, -mtune): Document that this accepts the same
19633 extension options as -mcpu.
19634 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19636 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19638 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19639 permitted extensions.
19641 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19643 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19644 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19645 (armv8-m.main): Add option +nodsp.
19646 * config/arm/arm-cpu-cdata.h: Regenerated.
19648 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19650 * config/arm/t-fuchsia: New file.
19651 * config.gcc (arm*-*-fuchsia*): Use it.
19653 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19655 * config/arm/t-symbian: Rewrite for new option infrastructure.
19657 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19659 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19660 (MULTILIB_REQUIRED): Likewise.
19662 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19664 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19665 (MULTILIB_RESUE): Likewise.
19666 (MULTILIB_MATCHES): Likewise.
19667 (MULTLIB_REQUIRED): Likewise.
19669 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19671 * config/arm/t-rtems: Rewrite for new option framework.
19673 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19675 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19676 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19677 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19678 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19679 * config/arm/t-multilib: ... here.
19680 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19681 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19682 armv7-a and armv8*-a when A-profile libraries have not been built.
19683 * config/arm/t-rmprofile: Rewrite.
19685 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19687 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19688 with a backslash. Remove the backslash after substituting unescaped
19690 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19694 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19695 the driver to pass through the default setting of -mfloat-abi.
19696 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19698 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19699 (all_feat_combs): New rule.
19700 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19702 * config/arm/t-aprofile: Rewrite.
19704 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19706 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19707 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19708 fpu is not specified by the user/command-line.
19709 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19710 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19711 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19712 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19713 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19714 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19716 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19718 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19719 * config/arm/t-arm-elf: Rewritten.
19721 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19723 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19724 have some floating-point instructions.
19725 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19726 (TARGET_MAYBE_HARD_FLOAT): New macro.
19727 * config/arm/arm-builtins.c (arm_init_builtins): Use
19728 TARGET_MAYBE_HARD_FLOAT.
19729 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19731 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19733 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19734 (configargs.h): Include it.
19735 (arm_print_hint_for_fpu_option): New function.
19736 (arm_parse_fpu_option): New function.
19737 (candidate_extension): New class.
19738 (arm_canon_for_multilib): New function.
19739 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19740 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19741 (ARCH_CANONICAL_SPECS): New macro.
19742 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19744 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19746 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19747 are set after handling multilib fragments. Set target_cpu_default2
19750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19752 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19754 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19755 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19757 (check_arch): Likewise.
19758 * config/arm/arm.c (arm_configure_build_target): Handle
19759 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19760 options in the default.
19762 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19764 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19765 when an option is an alias of another.
19766 * config/arm/parsecpu.awk (optalias): New parser token.
19767 (gen_comm_data): Mark non-alias options as such. Emit entries
19768 for extension aliases.
19769 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19770 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19771 (armv6kz, armv6zk, armv6t2): Likewise.
19772 (armv7): Make vfpv3-d16 an alias.
19773 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19775 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19776 Sort in canonical order.
19777 (armv8-a): Sort in canonical order.
19778 (armv8.1-a, armv8.2-a): Likewise.
19779 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19781 (cortex-a9): Sort in canonical order.
19782 * config/arm/arm.c (selftests.h): Include it.
19783 (arm_test_cpu_arch_data): New function.
19784 (arm_run_self_tests): New function.
19785 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19786 (targetm): Move declaration to the end of the file.
19787 * arm-cpu-cdata.h: Regenerated.
19789 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19791 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19792 call to target_mode_check describing the type of option passed.
19793 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19794 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19795 arm_parse_cpu_option_name to match parameters against list of
19797 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19798 arm_arch_core_flags data structure.
19799 * config/arm/arm-cpu_cdata.h: Regenerated.
19801 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19803 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19805 (arm_print_hint_for_cpu_option): Likewise.
19806 (arm_print_hint_for_arch_option): Likewise.
19807 (arm_parse_cpu_option_name): Likewise.
19808 (arm_parse_arch_option_name): Likewise.
19809 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19810 of entries in the all_fpus list.
19811 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19812 (arm_parse_cpu_option_name): Declare.
19813 (arm_parse_arch_option_name): Declare.
19814 (arm_parse_option_features): Declare.
19815 (arm_intialize_isa): Declare.
19816 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19818 (gen_comm_data): ... here. Make definitions non-static.
19819 * config/arm/arm-cpu-data.h: Regenerated.
19820 * config/arm/arm-cpu-cdata.h: Regenerated.
19822 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19824 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19825 (cpu_arch_extension): New structure.
19826 (cpu_arch_option, arch_option, cpu_option): New structures.
19827 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19828 architecture types.
19829 (gen_data): Generate new format data tables.
19830 * config/arm/arm.c (cpu_tune): New structure.
19831 (cpu_option, processors): Delete.
19832 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19833 (arm_print_hint_for_cpu_option): ... this and ...
19834 (arm_print_hint_for_arch_option): ... this.
19835 (arm_parse_arch_cpu_name): Delete. Replace with ...
19836 (arm_parse_cpu_option_name): ... this and ...
19837 (arm_parse_arch_option_name): ... this.
19838 (arm_unrecognized_feature): Change type of target parameter to
19840 (arm_parse_arch_cpu_features): Delete. Replace with ...
19841 (arm_parse_option_features): ... this.
19842 (arm_configure_build_target): Rework to use new configuration data
19844 (arm_print_tune_info): Rework for new configuration data tables.
19845 * config/arm/arm-cpu-data.h: Regenerated.
19846 * config/arm/arm-cpu.h: Regenerated.
19848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19850 * Makefile.in (OBJS): Move sbitmap.o from here ...
19851 (OBJS-libcommon): ... to here.
19853 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19855 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19856 (ISA_ALL_CRYPTO): New macro.
19857 (ISA_ALL_SIMD): New macro
19858 (ISA_ALL_FP): New macro.
19859 * config/arm/arm.c (fpu_bitlist): Update initializer.
19860 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19862 (arm9e): Add fpu. Add option for nofp
19863 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19864 (arm926ej-s, arm1026ej-s): Likewise.
19865 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19866 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19867 neon-fp16, neon-vfpv4, nofp and nosimd.
19868 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19869 (cortex-a8): Add fpu. Add option for nofp.
19870 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19871 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19872 (cortex-r4f): Add fpu.
19873 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19874 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19876 (cortex-r8): Likewise.
19877 (cortex-m4): Add fpu. Add option for nofp.
19878 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19879 (cortex-a17.cortex-a7): Likewise.
19880 (cortex-a32): Add fpu. Add options for crypto and nofp.
19881 (cortex-a35, cortex-a53): Likewise.
19882 (cortex-a57): Add fpu. Add option for crypto.
19883 (cortex-a72, cortex-a73): Likewise.
19884 (exynos-m1): Likewise.
19885 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19886 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19887 (cortex-m33): Add fpu. Add option for nofp.
19888 * config/arm/arm-cpu-cdata.h: Regenerated
19889 * config/arm/arm-cpu-data.h: Regenerated.
19891 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19893 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19894 (armv5te, armv5tej): Likewise.
19895 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19896 (armv7): Add options fp and vfpv3-d16.
19897 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19898 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19900 (armv7ve): Likewise.
19901 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19902 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19903 (armv8-a): Add nocrypto option.
19904 (armv8.1-a, armv8.2-a): Likewise.
19905 (armv8-m.main): add options fp, fp.dp and nofp.
19907 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19909 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19911 (armv8-a+crc): Delete.
19912 (armv8.1-a): Add options simd, crypto and nofp.
19913 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19914 (armv8.2-a+fp16): Delete.
19915 (armv8-m.main): Add option dsp.
19916 (armv8-m.main+dsp): Delete.
19917 (cortex-a8): Add fpu. Add nofp option.
19918 (cortex-a9): Add fpu. Add nofp and nosimd options.
19919 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19920 link to main cpu and architecture data structures.
19921 (gen_comm_data): Only put isa attributes from the main architecture
19923 (option): New statement for architecture and CPU entries.
19924 * arm.c (struct cpu_option): New structure.
19925 (struct processors): Add entry for options.
19926 (arm_unrecognized_feature): New function.
19927 (arm_parse_arch_cpu_name): Ignore any characters after the first
19929 (arm_parse_arch_cpu_feature): New function.
19930 (arm_configure_build_target): Separate out any CPU and architecture
19931 features and parse separately. Don't error out if -mfpu=auto is
19932 used with only an architecture string.
19933 (arm_print_asm_arch_directives): New function.
19934 (arm_file_start): Call it.
19935 * config/arm/arm-cpu-cdata.h: Regenerated.
19936 * config/arm/arm-cpu-data.h: Likewise.
19937 * config/arm/arm-tables.opt: Likewise.
19939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19941 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19942 assembler when it is not -mfpu=auto.
19944 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19946 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19947 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19948 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19949 (ASM_CPU_SPEC): Rewrite.
19950 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19951 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19952 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19953 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19954 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19955 copied string is NUL-terminated. Also strip any characters prefixed
19957 (arm_rewrite_selected_arch): New function.
19958 (arm_rewrite_march): New function.
19960 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19962 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19963 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19964 (march, mcpu, mtune): Convert to string-based options.
19965 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19966 (arm_parse_arch_cpu_name): New function.
19967 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19968 identify selected architecture or CPU.
19969 (arm_option_save): New function.
19970 (TARGET_OPTION_SAVE): Redefine.
19971 (arm_option_restore): Restore string options.
19972 (arm_option_print): Print string options.
19974 2017-06-16 Martin Sebor <msebor@redhat.com>
19976 PR tree-optimization/80933
19977 PR tree-optimization/80934
19978 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19979 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19980 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19981 (gimple_fold_builtin): Call them.
19983 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19985 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19986 as unlikely; update profile.
19988 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19990 * predict.c (force_edge_cold): Handle declaring edges impossible
19993 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19995 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19997 (try_unroll_loop_completely): Fix reporting.
19999 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20001 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20003 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20006 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20007 if given a non-constant argument for an intrinsic which requires a
20010 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20012 * profile.c (compare_freqs): New function.
20013 (branch_prob): Sort edge list.
20014 (find_spanning_tree): Assume that the list is priority sorted.
20016 2017-06-16 Richard Biener <rguenther@suse.de>
20018 PR tree-optimization/81090
20019 * passes.def (pass_record_bounds): Remove.
20020 * tree-pass.h (make_pass_record_bounds): Likewise.
20021 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20022 make_pass_record_bounds): Likewise.
20023 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20024 not free niter estimates at the beginning but at the end.
20025 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20027 2017-06-16 Richard Biener <rguenther@suse.de>
20029 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20030 initializer to workaround ICE in host GCC 4.8.
20032 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20034 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20036 (clone_inlined_nodes): Update.
20038 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20040 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20041 prefetch settings, and enable prefetching by default at -O3.
20043 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20045 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20046 Set flag_prefetch_loop_arrays according to tuning data.
20048 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20050 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20051 New tune structure.
20052 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20053 [Unrelated to main purpose of the patch] Place the pointer field last
20054 to enable type checking errors when tune structure are wrongly merged.
20055 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20056 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20057 (thunderx2t99_prefetch_tune): New tune constants.
20058 (tune_params *_tunings): Update all tunings (no functional change).
20059 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20060 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20061 from tunings structures.
20063 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20066 * ubsan.c (instrument_null): Add T argument, use it instead
20067 of computing it based on IS_LHS.
20068 (instrument_object_size): Likewise.
20069 (pass_ubsan::execute): Adjust instrument_null and
20070 instrument_object_size callers to pass gimple_get_lhs or
20071 gimple_assign_rhs1 result to it. Use instrument_null instead of
20072 calling get_base_address and instrument_mem_ref. Handle
20073 aggregate call arguments for object-size sanitization.
20075 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20077 PR tree-optimization/81089
20078 * tree-vrp.c (is_masked_range_test): Validate operands of
20081 2017-06-15 Martin Sebor <msebor@redhat.com>
20084 * dumpfile.c (dump_register): Avoid calling memset to initialize
20085 a class with a default ctor.
20086 * gcc.c (struct compiler): Remove const qualification.
20087 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20088 * hash-table.h: Ditto.
20089 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20091 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20092 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20094 * params.h (struct param_info): Make struct members non-const.
20095 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20096 with default initialization.
20097 * vec.h (vec_copy_construct, vec_default_construct): New helper
20099 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20100 with vec_copy_construct.
20101 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20102 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20103 * doc/invoke.texi (-Wclass-memaccess): Document.
20105 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20107 * emit-rtl.h (is_leaf): Update comment about local
20108 register allocator.
20110 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20113 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20114 for a variable to have a section before checking if the section has a
20116 Set section to.persistent if persistent attribute is set.
20117 Warn if .persistent attribute is used on an automatic variable.
20119 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20121 PR rtl-optimization/80474
20122 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20124 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20126 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20129 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20131 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20132 (casesi): Emit bounds checking as RTL.
20133 (casesi_internal_mips16_<mode>): Remove bounds checking.
20135 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20137 * config/xtensa/xtensa.c (xtensa_option_override): Append
20138 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20139 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20140 xtensa_doloop_hooks): Define unconditionally.
20141 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20143 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20144 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20145 for it in xtensa_option_override.
20146 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20147 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20149 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20151 * doc/cppopts.texi: Document '-' special value to -MF.
20153 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20155 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20156 (cortex_a53_fconst): Likewise.
20157 (cortex_a53_fpmul): Likewise.
20158 (cortex_a53_f_load_64): Likewise.
20159 (cortex_a53_f_load_many): Likewise.
20160 (cortex_a53_advsimd_alu): Likewise.
20161 (cortex_a53_advsimd_alu_q): Likewise.
20162 (cortex_a53_advsimd_mul): Likewise.
20163 (cortex_a53_advsimd_mul_q): Likewise.
20164 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20165 Add missing fmul, r2f_cvt and fconst cases.
20167 2017-06-14 Richard Biener <rguenther@suse.de>
20169 PR middle-end/81088
20170 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20172 (fold_binary_loc): When associating do not treat pre-existing
20173 TREE_OVERFLOW on literal constants as a reason to allow
20174 TREE_OVERFLOW on associated literal constants.
20176 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20178 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20179 (MASK_FEATURES): New macro.
20180 * config/sparc/sparc.c (sparc_option_override): Remove the special
20181 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20183 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20185 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20186 a division of 0 if non-call exceptions are enabled.
20188 2017-06-14 Andrew Pinski <apinski@cavium.com>
20189 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20192 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20193 Improve vector initialization code gen for only variable case.
20195 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20197 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20199 2017-06-14 Richard Biener <rguenther@suse.de>
20201 PR tree-optimization/81083
20202 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20205 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20207 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20208 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20209 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20210 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20211 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20212 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20214 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20216 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20217 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20219 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20221 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20223 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20225 * config/rs6000/t-rtems: Don't handle SPE.
20227 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20229 * config/rs6000/t-linux: Don't handle SPE.
20231 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20233 * config/rs6000/eabispe.h: Delete file.
20235 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20237 * config/rs6000/t-spe: Delete file.
20239 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20241 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20242 (rs6000_legitimate_offset_address_p): Return false for anything in
20243 V2SImode or V2SFmode.
20245 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20247 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20248 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20249 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20251 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20252 (rs6000_legitimate_offset_address_p): Ditto.
20253 (rs6000_emit_move): Ditto.
20254 (rs6000_init_builtins): Remove V4HI_type_node.
20256 2017-06-13 Martin Liska <mliska@suse.cz>
20259 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20260 (gate_asan): Likewise.
20261 * asan.h (asan_no_sanitize_address_p): Remove the function.
20262 (sanitize_flags_p): New function.
20263 * builtins.def: Fix coding style.
20264 * common.opt: Use renamed enum value.
20265 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20266 * doc/extend.texi: Document no_sanitize attribute.
20267 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20268 to SANITIZE_UNDEFINED_NONDEFAULT.
20269 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20270 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20271 Use sanitize_flags_p.
20272 * gimplify.c (gimplify_function_tree): Likewise.
20273 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20274 * opts.c (parse_no_sanitize_attribute): New function.
20275 (common_handle_option): Use renamed enum value.
20276 * opts.h (parse_no_sanitize_attribute): Declare.
20277 * tree.c (sanitize_flags_p): New function.
20278 * tree.h: Declared here.
20279 * tsan.c: Use sanitize_flags_p.
20280 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20281 (instrument_mem_ref): Likewise.
20282 (instrument_bool_enum_load): Likewise.
20283 (do_ubsan_in_current_function): Remove the function.
20284 (pass_ubsan::execute): Use sanitize_flags_p.
20285 * ubsan.h: Remove do_ubsan_in_current_function
20286 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20287 (dump_function_to_file): Use it here.
20289 2017-06-13 Martin Jambor <mjambor@suse.cz>
20291 PR tree-optimization/80803
20292 PR tree-optimization/81063
20293 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20294 (propagate_subaccesses_across_link): Enqueue subtree whenever
20295 necessary instead of relying on the caller.
20297 2017-06-13 Martin Jambor <mjambor@suse.cz>
20299 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20300 that have a first_link.
20301 (sort_and_splice_var_accesses): Do not check first_link before
20303 (subtree_mark_written_and_enqueue): Likewise.
20304 (propagate_all_subaccesses): Likewise and do not stop at first
20305 parent with a first_link.
20307 2017-06-13 Martin Jambor <mjambor@suse.cz>
20309 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20312 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20314 * match.pd: New pattern.
20316 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20318 * tree-vrp.c (is_masked_range_test): New function.
20319 (register_edge_assert_for): Determine ranges for
20322 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20324 PR tree-optimization/67328
20325 * fold-const.c (maskable_range_p): New function.
20326 (build_range_check): Generate bittests if possible.
20328 2017-06-13 Martin Liska <mliska@suse.cz>
20330 * gimple-pretty-print.c (dump_probability): Add new argument.
20331 (dump_edge_probability): Dump both probability and count.
20332 (dump_gimple_label): Likewise.
20333 (dump_gimple_bb_header): Likewise.
20335 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20338 * config/avr/avr-devices.c: Fix indentation.
20339 * config/avr/gen-avr-mmcu-specs.c: Dito.
20341 2017-06-13 Richard Biener <rguenther@suse.de>
20343 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20344 instead get vector type from stmt_info.
20345 (vectorizable_reduction): Adjust. Remove dead code.
20347 2017-06-13 Richard Biener <rguenther@suse.de>
20349 PR middle-end/81065
20350 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20351 case of C * (x * C2 + C3).
20352 (fold_addr_of_array_ref_difference): Properly fold index difference.
20354 2017-06-12 David S. Miller <davem@davemloft.net>
20357 * config/sparc/sparc.md (return expander): Emit frame blockage if
20358 function uses alloca.
20360 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20362 * combine.c (make_field_assignment): Check len rather than the mode
20363 precision when calling force_to_mode.
20365 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20367 Support multilibs and devices that see flash in RAM address range.
20370 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20371 (avr_mcu_t) <flash_pm_offset>: New field.
20372 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20373 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20374 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20375 (AVR_TINY_PM_OFFSET): Remove macro.
20376 * config/avr/avr.opt (-mshort-calls): New option.
20377 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20378 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20379 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20380 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20381 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20382 instead of avr_arch->have_jmp_call.
20383 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20384 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20385 avr_arch->flash_pm_offset to define.
20386 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20387 new field flash_pm_offset. Add entry for avrxmega3.
20388 (avr_texinfo): Add entry for avrxmega3.
20389 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20390 attiny212, attiny214,
20391 attiny412, attiny414, attiny416, attiny417,
20392 attiny814, attiny816, attiny817,
20393 attiny1614, attiny1616, attiny1617,
20394 attiny3214, attiny3216, attiny3217.
20395 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20396 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20397 (avr_print_operand_address) [AVR_TINY]: Same.
20398 (avr_asm_init_sections) <readonly_data_section>: Only patch
20399 callback if avr_arch->flash_pm_offset = 0.
20400 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20401 for rodata if avr_arch->flash_pm_offset != 0.
20402 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20403 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20404 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20405 (m_options): Append opt_rcall.
20406 (m_dirnames): Append dir_rcall.
20407 * config/avr/t-multilib: Regenerate.
20409 * configure.ac [target=avr]: Check whether avrxmega3 default
20410 linker description file works as needed.
20411 * configure: Regenerate.
20412 * doc/avr-mmcu.texi: Regenerate.
20413 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20414 <__AVR_ARCH__>: Document avrxmega3 and 103.
20415 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20416 <__AVR_SHORT_CALLS__>: Document it.
20417 <__AVR_PM_BASE_ADDRESS__>: Document it.
20418 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20419 (AVR Variable Attributes) <progmem>: Document this is
20420 not needed for avrxmega3.
20421 (AVR Named Address Spaces) <__flash>: Dito.
20423 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20425 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20427 2017-06-12 Doug Rupp <rupp@adacore.com>
20429 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20430 Append vxworks-stdint.h to the tm_file list.
20431 * config/vxworks-stdint.h: New file.
20433 2017-06-12 Martin Liska <mliska@suse.cz>
20435 PR tree-optimization/81041
20436 * tree-profile.c (gimple_gen_ic_func_profiler):
20437 Create an extra BB in profile-generate
20438 (gimple_gen_time_profiler): Likewise.
20440 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20442 PR tree-optimization/81003
20443 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20444 (update_range_test): Use it instead of force_gimple_operand_gsi.
20446 2017-06-12 Richard Biener <rguenther@suse.de>
20448 PR tree-optimization/81053
20449 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20450 with backedge value not defined in loop. Simplify def stmt
20453 2017-06-11 Tom de Vries <tom@codesourcery.com>
20456 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20458 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20459 nvptx_cannot_force_const_mem.
20461 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20463 * opts.c (finish_options): Move test for flag_split_stack after
20464 it has been initialized.
20466 2017-06-11 Jason Merrill <jason@redhat.com>
20468 * tree.h (id_equal): New.
20469 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20470 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20471 instead of strcmp of IDENTIFIER_POINTER.
20473 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20475 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20476 (mark_all_inlined_calls_cdtor): Fix formating.
20477 (inline_transform): Rescale profile before inlining.
20479 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20481 * cgraph.h (cgraph_edge::clone): Update prototype.
20482 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20483 (cgraph_node::create_clone): Update.
20484 (cgraph_node::create_version_clone): Update.
20485 * tree-inline.c (copy_bb): Update.
20486 (expand_call_inline): Update.
20488 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20490 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20491 factored out from ...
20492 (rs6000_emit_prologue): ... here.
20494 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20496 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20497 factored out from ...
20498 (rs6000_emit_prologue): ... here.
20500 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20502 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20504 (handle_missing_profiles): Fix computation of tp_first_run.
20505 (counts_to_freqs): Do not touch freqs when count is 0.
20507 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20509 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20512 2017-06-10 Tom de Vries <tom@codesourcery.com>
20514 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20515 attributes): Document signal effective target.
20517 2017-06-10 Tom de Vries <tom@codesourcery.com>
20519 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20520 Document effective target stack_size.
20522 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20524 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20525 to the edit_context if they can be auto-applied.
20527 2017-06-9 Ian Lance Taylor <iant@golang.org>
20529 * opts.c (finish_options): If -fsplit-stack, disable implicit
20530 -forder-blocks-and-partition.
20531 * doc/invoke.texi (Optimize Options): Document that when using
20532 -fsplit-stack -forder-blocks-and-partition is not implicitly
20535 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20537 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20538 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20539 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20540 * builtins.def (abort, trap, unreachable): Declare cold.
20541 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20542 * tree-core.h (ECF_COLD): New.
20543 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20544 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20546 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20548 * predict.c (unlikely_executed_stmt_p): Cleanup.
20550 2017-06-09 Richard Biener <rguenther@suse.de>
20552 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20553 model if the ref is always written to.
20555 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20557 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20559 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20561 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20564 2017-06-09 Tom de Vries <tom@codesourcery.com>
20567 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20568 "target cannot support label values" when encountering LABEL_REF.
20570 2017-06-09 Martin Liska <mliska@suse.cz>
20572 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20573 (gimple_gen_ic_func_profiler): Emit direct comparison
20574 of __gcov_indirect_call_callee with NULL.
20575 (gimple_gen_time_profiler): Change probability from
20576 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20578 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20580 * profile.c (edge_gcov_counts): Turn to pointer.
20581 (compute_branch_probabilities, compute_branch_probabilities): Update.
20582 (branch_prob): Do not clear edge_gcov_count.
20583 * profile.h (edge_gcov_counts): Turn to pointer.
20584 (edge_gcov_count): Update.
20586 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20588 * gimple.h (gimple_check_failed): Mark cold.
20590 2017-06-09 Richard Biener <rguenther@suse.de>
20592 PR tree-optimization/66623
20593 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20594 refactor check_reduction into two parts, properly computing
20595 whether we have to check reduction validity for outer loop
20598 2017-06-09 Richard Biener <rguenther@suse.de>
20600 PR tree-optimization/79483
20601 * graphite-scop-detection.c (order): New global.
20602 (get_order): Compute bb to order mapping that satisfies code
20603 generation constraints.
20604 (cmp_pbbs): New helper.
20605 (build_scops): Start domwalk at entry block, sort generated
20608 2017-06-09 Richard Biener <rguenther@suse.de>
20610 PR middle-end/81007
20611 * ipa-polymorphic-call.c
20612 (ipa_polymorphic_call_context::restrict_to_inner_class):
20613 Skip FIELD_DECLs with error_mark_node type.
20614 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20617 2017-06-09 Martin Liska <mliska@suse.cz>
20619 * predict.c (struct branch_predictor): New struct.
20620 (test_prediction_value_range): New test.
20621 (predict_c_tests): New function.
20622 * selftest-run-tests.c (selftest::run_tests): Run the function.
20623 * selftest.h: Declare new tests.
20625 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20628 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20629 gen_add3_insn did not fail.
20630 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20631 r0, construct that number in a temporary reg and add that reg to r0.
20632 If asked to put the result in r0 as well, fail.
20634 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20636 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20637 for early expansion of vec_eqv.
20639 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20641 PR middle-end/81005
20642 * ubsan.c (instrument_null): Avoid pointless code temporary.
20643 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20645 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20649 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20652 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20655 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20657 * predict.c (unlikely_executed_edge_p): Move ahead.
20658 (probably_never_executed_edge_p): Use it.
20660 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20662 PR middle-end/79988
20663 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20664 gimple_call_builtin_p call.
20666 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20668 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20669 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20670 rtl_check_failed_type2, rtl_check_failed_code1,
20671 rtl_check_failed_code2, rtl_check_failed_code_mode,
20672 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20673 rtvec_check_failed_bounds, rtl_check_failed_flag,
20674 _fatal_insn_not_found, _fatal_insn): Likewise.
20675 * tree.h (tree_contains_struct_check_failed,
20676 tree_check_failed, tree_not_check_failed,
20677 tree_class_check_failed, tree_range_check_failed,
20678 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20679 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20680 tree_operand_check_failed, omp_clause_check_failed,
20681 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20684 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20686 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20687 flag_branch_probabilities.
20688 * ipa-inline.c (edge_badness): Likewise.
20689 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20690 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20691 * predict.c (maybe_hot_frequency_p): Likewise.
20692 (probably_never_executed): Likewise.
20693 * sched-ebb.c (schedule_ebbs): Likewise.
20694 * sched-rgn.c (find_single_block_region): Likewise.
20695 * tracer.c (tail_duplicate): Likewise.
20697 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20699 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20700 longer requires x_flag_profile_use.
20702 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20704 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20705 instead of flag_reorder_blocks_and_partition.
20706 * dbxout.c (dbxout_function_end): Likewise.
20707 * dwarf2out.c (gen_subprogram_die): Likewise.
20708 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20709 * hw-doloop.c (reorg_loops): Likewise.
20710 * varasm.c (assemble_start_function,
20711 assemble_end_function): Likewise.
20712 (decide_function_section): Do not check for
20713 flag_reorder_blocks_and_partition.
20715 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20717 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20719 (chkp_get_hard_register_fake_addr_expr): Ditto.
20720 (chkp_build_addr_expr): Add check for hard reg case.
20721 (chkp_parse_array_and_component_ref): Ditto.
20722 (chkp_find_bounds_1): Ditto.
20723 (chkp_process_stmt): Don't generate bounds store for
20726 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20728 * predict.c (maybe_hot_bb_p): Do not check profile status.
20729 (maybe_hot_edge_p): Likewise.
20730 (probably_never_executed): Check for zero counts even if profile
20732 (unlikely_executed_edge_p): New function.
20733 (unlikely_executed_stmt_p): New function.
20734 (unlikely_executed_bb_p): New function.
20735 (set_even_probabilities): Use unlikely predicates.
20736 (combine_predictions_for_bb): Likewise.
20737 (predict_paths_for_bb): Likewise.
20738 (predict_paths_leading_to_edge): Likewise.
20739 (determine_unlikely_bbs): New function.
20740 (estimate_bb_frequencies): Use it.
20741 (compute_function_frequency): Use zero counts even if profile is
20743 * profile-count.h: Fix typo.
20745 2017-08-08 Julia Koval <julia.koval@intel.com>
20747 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20748 _mm512_mask_cvtsepi16_storeu_epi8,
20749 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20750 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20751 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20752 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20753 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20754 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20755 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20756 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20757 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20758 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20759 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20760 __builtin_ia32_pmovuswb256mem_mask,
20761 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20762 __builtin_ia32_pmovwb512mem_mask): New builtins.
20764 2017-08-08 Julia Koval <julia.koval@intel.com>
20766 PR target/73350,80862
20767 * config/i386/subst.md (round): Fix round pattern.
20768 * config/i386/i386.c (ix86_erase_embedded_rounding):
20769 Fix erasing rounding for the fixed pattern.
20771 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20773 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20775 2017-06-08 Martin Liska <mliska@suse.cz>
20777 PR gcov-profile/80911
20778 * gcov.c (block_info::block_info): New constructor.
20780 2017-06-07 Carl Love <cel@us.ibm.com>
20782 * config/rs6000/rs6000-c: The return type of the following
20783 built-in functions was implemented as int not long long. Fix sign
20784 of return value for the unsigned version of vec_mulo and vec_mule.
20785 vector unsigned long long vec_bperm (vector unsigned long long,
20786 vector unsigned char)
20787 vector signed long long vec_mule (vector signed int,
20789 vector unsigned long long vec_mule (vector unsigned int,
20790 vector unsigned int)
20791 vector signed long long vec_mulo (vector signed int,
20793 vector unsigned long long vec_mulo (vector unsigned int,
20794 vector unsigned int)
20795 * doc/extend.texi: Fix the documentation for the built-in
20798 2017-06-07 Carl Love <cel@us.ibm.com>
20801 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20804 2017-06-07 Carl Love <cel@us.ibm.com>
20806 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20807 support, Generate doublehv for signed int/float for BE case only.
20809 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20811 * doc/invoke.texi (mcx16): Rewrite.
20813 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20815 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20816 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20817 *mov<mode>_softfloat, and an anonymous splitter): Use
20818 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20820 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20822 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20824 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20825 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20826 (rs6000_debug_reg_global): Adjust.
20827 (rs6000_init_hard_regno_mode_ok): Adjust.
20828 (rs6000_dbx_register_number): Adjust.
20829 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20830 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20831 Remove SPE_ACC and SPEFSCR.
20832 (REG_ALLOC_ORDER): Ditto.
20833 (FRAME_POINTER_REGNUM): Change to 111.
20834 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20835 (REG_CLASS_NAMES): Ditto.
20836 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20837 (REGISTER_NAMES): Ditto.
20838 (ADDITIONAL_REG_NAMES): Ditto.
20839 (rs6000_reg_names): Ditto.
20840 * config/rs6000/rs6000.md: Renumber some register number
20843 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20845 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20847 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20848 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20850 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20851 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20852 Delete the SPE high registers.
20853 (REG_ALLOC_ORDER): Ditto.
20854 (enum reg_class): Remove SPE_HIGH_REGS.
20855 (REG_CLASS_NAMES): Ditto.
20856 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20857 (REGISTER_NAMES): Ditto.
20858 (rs6000_reg_names): Ditto.
20859 * doc/tm.texi.in: Remove SPE as example.
20860 * doc/tm.texi: Regenerate.
20862 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20864 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20865 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20866 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20867 * config/rs6000/rs6000.md (type): Remove "brinc".
20869 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20871 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20872 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20873 * config/rs6000/linuxspe.h: Delete file.
20874 * config/rs6000/rs6000.md: Don't include spe.md.
20875 * config/rs6000/spe.h: Delete file.
20876 * config/rs6000/spe.md: Delete file.
20877 * config/rs6000/t-rs6000: Remove spe.md.
20879 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20881 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20882 (reg_or_none500mem_operand): Delete.
20883 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20884 instead of reg_or_none500mem_operand.
20886 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20888 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20889 handling of SPE flags.
20890 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20892 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20894 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20896 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20897 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20898 paired_divv2sf3): Similar.
20899 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20900 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20901 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20903 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20904 Rename the paired_* instruction patterns.
20905 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20907 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20908 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20909 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20910 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20911 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20912 PAIRED_VECTOR_MODE.
20913 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20914 (spe_func_has_64bit_regs_p): Delete.
20915 (spe_expand_predicate_builtin): Delete.
20916 (spe_expand_evsel_builtin): Delete.
20917 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20918 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20919 (invalid_e500_subreg): Delete.
20920 (rs6000_legitimize_address): Always force_reg op2 as well, for
20921 paired single memory accesses.
20922 (rs6000_member_type_forces_blk): Delete.
20923 (rs6000_spe_function_arg): Delete.
20924 (rs6000_expand_unop_builtin): Delete SPE handling.
20925 (rs6000_expand_binop_builtin): Ditto.
20926 (spe_expand_stv_builtin): Delete.
20927 (bdesc_2arg_spe): Delete.
20928 (spe_expand_builtin): Delete.
20929 (spe_expand_predicate_builtin): Delete.
20930 (spe_expand_evsel_builtin): Delete.
20931 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20932 (spe_init_builtins): Delete.
20933 (spe_func_has_64bit_regs_p): Delete.
20934 (savres_routine_name): Delete "info" parameter. Adjust callers.
20935 (rs6000_emit_stack_reset): Ditto.
20936 (rs6000_dwarf_register_span): Delete.
20937 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20938 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20939 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20941 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20943 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20944 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20945 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20946 mulv2sf3, divv2sf3): Delete expanders.
20948 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20950 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20952 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20954 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20955 * config/rs6000/rs6000.c: Ditto.
20957 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20959 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20960 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20961 comparison_operator.
20963 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20965 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20966 * config/rs6000/rs6000.opt: Ditto.
20967 * config/rs6000/t-rtems: Ditto.
20969 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20971 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20972 TARGET_E500_SINGLE by 0, simplify.
20973 * config/rs6000/rs6000.c: Ditto.
20974 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20975 (spe_build_register_parallel): Delete.
20976 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20977 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20978 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20979 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20980 (E500_CONVERT): Delete.
20981 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20983 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20985 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20986 * config/rs6000/dfp.md: Ditto.
20987 (negdd2, *negdd2_fpr): Merge.
20988 (absdd2, *absdd2_fpr): Merge.
20989 (negtd2, *negtd2_fpr): Merge.
20990 (abstd2, *abstd2_fpr): Merge.
20991 * config/rs6000/e500.h: Delete file.
20992 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20993 TARGET_FPRS by 1 and simplify.
20994 * config/rs6000/rs6000-c.c: Ditto.
20995 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20996 TARGET_DF_SPE by 0.
20997 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20999 * config/rs6000/rs6000.md: Ditto.
21000 (floatdidf2, *floatdidf2_fpr): Merge.
21001 (move_from_CR_gt_bit): Delete.
21002 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21003 (E500_CR_IOR_COMPARE): Delete.
21004 (All patterns that require !TARGET_FPRS): Delete.
21005 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21007 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21009 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21011 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21013 * graphds.c (add_edge): Intitialize edge's attached data.
21014 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21015 pointer parameter. Call pointed function on each edge during
21016 graph traversing. Skip traversing the edge when the function
21018 (graphds_dfs, graphds_scc): Ditto.
21019 (for_each_edge): New parameter. Pass the new parameter to callback
21021 * graphds.h (skip_edge_callback): New function pointer type.
21022 (graphds_dfs, graphds_scc): New function pointer parameter.
21023 (graphds_edge_callback, for_each_edge): New parameter.
21025 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21027 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21028 out code checking if runtime alias check is possible to below ...
21029 Call the new function.
21030 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21031 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21033 2017-06-07 Marek Polacek <polacek@redhat.com>
21036 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21037 TYPE_OVERFLOW_WRAPS check.
21039 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21041 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21042 if versioning is required.
21043 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21044 peeling with the check for versioning.
21046 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21048 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21049 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21050 Set true to new parameter if new ssa variable is defined.
21051 (vect_gen_vector_loop_niters): Refactor. Set range information
21052 for the new vector loop bound variable.
21053 (vect_do_peeling): Ditto.
21055 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21057 * tree-affine.c (ssa.h): Include header file.
21058 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21059 has wrapping overflow behavior.
21061 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21063 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21065 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21067 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21068 (tree_to_aff_combination): ... here.
21070 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21072 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21073 reg_pressure model function.
21074 (ivopts_global_cost_for_size): Delete.
21075 (determine_set_costs, iv_ca_recount_cost): Call new model function
21076 ivopts_estimate_reg_pressure.
21078 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21080 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21081 expensive than udiv. Remove floating point cases from mod.
21083 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21085 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21086 Increase idiv cost.
21088 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21090 * config/aarch64/aarch64.md
21091 (copysignsf3): Fix mask generation.
21093 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21095 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21097 (class dump_manager): Add register_dumps method.
21098 * dumpfile.c: Include langhooks.h.
21099 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21100 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21101 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21102 (dump_manager::dump_register): Start with 512 entries instead of 32.
21103 (dump_manager::register_dumps): New method.
21104 * toplev.c (general_init): Instead of invoking register_dumps
21105 langhook, invoke register_dumps method on the dump manager.
21106 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21109 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21111 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21112 Say that # requires an associated define_split to exist, and that
21113 the define_split must be suitable for use after register allocation.
21115 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21117 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21118 (compute_outgoing_frequencies): Also initialize zero counts.
21119 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21120 around loops; preserve more of profile when nothing changes.
21122 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21124 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21126 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21127 * config/arm/arm-cpu-cdata.h: Regenerate.
21128 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21129 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21130 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21131 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21132 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21134 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21136 * config/arm/t-rmprofile: Likewise.
21137 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21139 2017-06-06 David S. Miller <davem@davemloft.net>
21142 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21143 blockage if function uses alloca.
21145 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21147 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21148 New "uid" fields to hold pretty-print IDs of group and ref.
21149 Memory references are now identified as <group_id>:<ref_id>
21150 instead of using [random] addresses.
21151 (dump_mem_details): Simplify, no functional change.
21152 (dump_mem_ref): Simplify and make output more concise.
21153 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21154 (find_or_create_group): Initialize group uid.
21155 (record_ref): Initialize ref uid. Improve debug output.
21156 (prune_group_by_reuse, should_issue_prefetch_p,)
21157 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21158 (mark_nontemporal_store, determine_loop_nest_reuse):
21159 Improve debug output.
21161 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21163 * dbgcnt.def (prefetch): New debug counter.
21164 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21165 (schedule_prefetches): Stop issueing prefetches if debug counter
21168 2017-06-06 Tom de Vries <tom@codesourcery.com>
21170 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21171 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21173 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21175 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21176 Use aarch64_reg_or_zero predicate for operand 4.
21177 (aarch64_compare_and_swap<mode> define_insn_and_split):
21178 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21179 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21181 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21183 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21184 (arm_compute_save_core_reg_mask): This.
21185 (thumb1_compute_save_reg_mask): Rename into ...
21186 (thumb1_compute_save_core_reg_mask): This.
21187 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21188 (arm_compute_frame_layout): Likewise.
21190 2017-06-06 Richard Biener <rguenther@suse.de>
21192 PR tree-optimization/80974
21193 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21194 keep or clear leaders SSA info.
21196 2017-06-06 Tom de Vries <tom@codesourcery.com>
21198 * config/nvptx/nvptx.c (split_mode_p): New function.
21199 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21201 2017-06-06 Tom de Vries <tom@codesourcery.com>
21203 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21205 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21208 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21211 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21213 * shrink-wrap.c (handle_simple_exit): Update profile.
21214 (try_shrink_wrapping): Upate profile.
21216 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21218 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21219 (tree_guess_outgoing_edge_probabilities): New.
21220 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21221 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21223 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21225 * ipa-split.c (split_function): Initialize return bb profile.
21227 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21229 * profile.c (compute_branch_probabilities): Also initialize
21230 EXIT_BLOCK profile.
21232 2017-06-06 Richard Biener <rguenther@suse.de>
21234 PR tree-optimization/80928
21235 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21236 (vect_analyze_loop_operations): Properly guard analysis for
21238 (vect_transform_loop): Likewise.
21239 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21240 (vect_model_induction_cost): Do not cost for pure SLP.
21241 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21242 of induction in inner loop vectorization.
21243 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21244 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21245 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21247 (vect_analyze_slp_cost_1): Cost induction.
21248 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21249 (vect_get_slp_vect_defs): Likewise.
21250 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21251 (vect_transform_stmt): Handle SLP reductions.
21252 * tree-vectorizer.h (vectorizable_induction): Adjust.
21254 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21256 * config/rs6000/rs6000.c (make_resolver_func): Update
21257 init_lowered_empty_function call.
21259 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21261 * doc/invoke.texi: Document the -fprofile-abs-path option.
21262 * common.opt (fprofile-abs-path): New option.
21263 * gcov-io.h (gcov_write_filename): Declare.
21264 * gcov-io.c (gcov_write_filename): New function.
21265 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21266 * profile.c (output_location): Likewise.
21268 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21270 * shring-wrap.c: Revert accidental commit.
21272 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21274 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21276 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21278 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21280 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21281 profile in callgraph edge.
21282 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21283 (apply_scale): Likewise.
21284 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21285 Also scale profile when inlining function with zero profile.
21286 (initialize_cfun): Update exit block profile even when it is zero.
21287 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21288 when profile is read.
21290 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21292 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21293 (CLONE_*): New constants to define the processors we can generate
21294 code for with the target_clone attribute.
21295 (rs6000_clone_map): New array to identify which clone processors
21296 the current program is running on.
21297 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21298 target_clone attribute.
21299 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21300 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21301 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21302 (cpu_expand_builtin): Add support for target_clone attribute.
21303 (rs6000_valid_attribute_p): Allow "default" attribute.
21304 (get_decl_name): New debug function to simplify printing the
21305 current function name in debugging statements.
21306 (rs6000_clone_priority): New functions to support the target_clone
21307 attribute, and be able to generate code to switch between ISA 2.05
21308 through ISA 3.0 (power6 through power9).
21309 (rs6000_compare_version_priority): Likewise.
21310 (rs6000_get_function_versions_dispatcher): Likewise.
21311 (make_resolver_func): Likewise.
21312 (add_condition_to_bb): Likewise.
21313 (dispatch_function_versions): Likewise.
21314 (rs6000_generate_version_dispatcher_body): Likewise.
21315 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21316 (fusion_gpr_load_p): Fix a spacing issue.
21317 * doc/extend.texi (Common Function Attributes): Document that the
21318 PowerPC supports the target_clone attribute.
21320 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21322 * config/arm/arm.h: explain F symbol found in description of ARM
21323 register allocation in its legend.
21325 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21327 * config/mips/frame-header-opt.c: Include profile-count.h.
21328 * config/riscv/riscv.c: Include profile-count.h
21330 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21332 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21334 (sm_set_flag_if_changed): Add bbs field.
21335 (execute_sm_if_changed_flag_set): Pass BBS.
21336 (execute_sm): Update.
21338 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21340 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21343 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21345 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21346 (peephole2): New peephole2 to emit the above.
21347 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21349 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21351 * config/aarch64/aarch64.c (define_peephole2 above
21352 *sub_<shift>_<mode>): New peephole.
21354 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21356 * config/i386/i386.c (make_resolver_func): Update.
21357 * Makefile.in: Add profile-count.h and profile-count.o
21358 * auto-profile.c (afdo_indirect_call): Update to new API.
21359 (afdo_set_bb_count): Update.
21360 (afdo_propagate_edge): Update.
21361 (afdo_propagate_circuit): Update.
21362 (afdo_calculate_branch_prob): Update.
21363 (afdo_annotate_cfg): Update.
21364 * basic-block.h: Include profile-count.h
21365 (struct edge_def): Turn count to profile_count.
21366 (struct basic_block_def): Likewie.
21367 (REG_BR_PROB_BASE): Move to profile-count.h
21368 (RDIV): Move to profile-count.h
21369 * bb-reorder.c (max_entry_count): Turn to profile_count.
21370 (find_traces): Update.
21371 (rotate_loop):Update.
21372 (connect_traces):Update.
21373 (sanitize_hot_paths):Update.
21374 * bt-load.c (migrate_btr_defs): Update.
21375 * cfg.c (RDIV): Remove.
21376 (init_flow): Use alloc_block.
21377 (alloc_block): Uninitialize count.
21378 (unchecked_make_edge): Uninitialize count.
21379 (check_bb_profile): Update.
21380 (dump_edge_info): Update.
21381 (dump_bb_info): Update.
21382 (update_bb_profile_for_threading): Update.
21383 (scale_bbs_frequencies_int): Update.
21384 (scale_bbs_frequencies_gcov_type): Update.
21385 (scale_bbs_frequencies_profile_count): New.
21386 * cfg.h (update_bb_profile_for_threading): Update.
21387 (scale_bbs_frequencies_profile_count): Declare.
21388 * cfgbuild.c (compute_outgoing_frequencies): Update.
21389 (find_many_sub_basic_blocks): Update.
21390 * cfgcleanup.c (try_forward_edges): Update.
21391 (try_crossjump_to_edge): Update.
21392 * cfgexpand.c (expand_gimple_tailcall): Update.
21393 (construct_exit_block): Update.
21394 * cfghooks.c (verify_flow_info): Update.
21395 (dump_bb_for_graph): Update.
21396 (split_edge): Update.
21397 (make_forwarder_block): Update.
21398 (duplicate_block): Update.
21399 (account_profile_record): Update.
21400 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21401 (get_estimated_loop_iterations): Update.
21402 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21403 (single_likely_exit): Update.
21404 * cfgloopmanip.c (scale_loop_profile): Update.
21406 (set_zero_probability): Update.
21407 (lv_adjust_loop_entry_edge): Update.
21408 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21409 (purge_dead_edges): Update.
21410 (rtl_account_profile_record): Update.
21411 * cgraph.c (cgraph_node::create): Uninitialize count.
21412 (symbol_table::create_edge): Uninitialize count.
21413 (cgraph_update_edges_for_call_stmt_node): Update.
21414 (cgraph_edge::dump_edge_flags): Update.
21415 (cgraph_node::dump): Update.
21416 (cgraph_edge::maybe_hot_p): Update.
21417 * cgraph.h: Include profile-count.h
21418 (create_clone), create_edge, create_indirect_edge): Update.
21419 (cgraph_node): Turn count to profile_count.
21420 (cgraph_edge0: Likewise.
21421 (make_speculative, clone): Update.
21422 (create_edge): Update.
21423 (init_lowered_empty_function): Update.
21424 * cgraphclones.c (cgraph_edge::clone): Update.
21425 (duplicate_thunk_for_node): Update.
21426 (cgraph_node::create_clone): Update.
21427 * cgraphunit.c (cgraph_node::analyze): Update.
21428 (cgraph_node::expand_thunk): Update.
21429 * final.c (dump_basic_block_info): Update.
21430 * gimple-streamer-in.c (input_bb): Update.
21431 * gimple-streamer-out.c (output_bb): Update.
21432 * graphite.c (print_global_statistics): Update.
21433 (print_graphite_scop_statistics): Update.
21434 * hsa-brig.c: Include basic-block.h.
21435 * hsa-dump.c: Include basic-block.h.
21436 * hsa-gen.c (T sum_slice): Update.
21437 (convert_switch_statements):Update.
21438 * hsa-regalloc.c: Include basic-block.h.
21439 * ipa-chkp.c (chkp_produce_thunks): Update.
21440 * ipa-cp.c (struct caller_statistics): Update.
21441 (init_caller_stats): Update.
21442 (gather_caller_stats): Update.
21443 (ipcp_cloning_candidate_p): Update.
21444 (good_cloning_opportunity_p): Update.
21445 (get_info_about_necessary_edges): Update.
21446 (dump_profile_updates): Update.
21447 (update_profiling_info): Update.
21448 (update_specialized_profile): Update.
21449 (perhaps_add_new_callers): Update.
21450 (decide_about_value): Update.
21451 (ipa_cp_c_finalize): Update.
21452 * ipa-devirt.c (struct odr_type_warn_count): Update.
21453 (struct decl_warn_count): Update.
21454 (struct final_warning_record): Update.
21455 (possible_polymorphic_call_targets): Update.
21456 (ipa_devirt): Update.
21457 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21458 * ipa-icf.c (sem_function::merge): Update.
21459 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21460 * ipa-inline.c (compute_uninlined_call_time): Update.
21461 (compute_inlined_call_time): Update.
21462 (want_inline_small_function_p): Update.
21463 (want_inline_self_recursive_call_p): Update.
21464 (edge_badness): Update.
21465 (lookup_recursive_calls): Update.
21466 (recursive_inlining): Update.
21467 (inline_small_functions): Update.
21468 (dump_overall_stats): Update.
21469 (dump_inline_stats): Update.
21470 * ipa-profile.c (ipa_profile_generate_summary): Update.
21471 (ipa_propagate_frequency): Update.
21472 (ipa_profile): Update.
21473 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21474 * ipa-utils.c (ipa_merge_profiles): Update.
21475 * loop-doloop.c (doloop_modify): Update.
21476 * loop-unroll.c (report_unroll): Update.
21477 (unroll_loop_runtime_iterations): Update.
21478 * lto-cgraph.c (lto_output_edge): Update.
21479 (lto_output_node): Update.
21480 (input_node): Update.
21481 (input_edge): Update.
21482 (merge_profile_summaries): Update.
21483 * lto-streamer-in.c (input_cfg): Update.
21484 * lto-streamer-out.c (output_cfg): Update.
21485 * mcf.c (create_fixup_graph): Update.
21486 (adjust_cfg_counts): Update.
21487 (sum_edge_counts): Update.
21488 * modulo-sched.c (sms_schedule): Update.
21489 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21490 * predict.c (maybe_hot_count_p): Update.
21491 (probably_never_executed): Update.
21492 (dump_prediction): Update.
21493 (combine_predictions_for_bb): Update.
21494 (propagate_freq): Update.
21495 (handle_missing_profiles): Update.
21496 (counts_to_freqs): Update.
21497 (rebuild_frequencies): Update.
21498 (force_edge_cold): Update.
21499 * predict.h: Include profile-count.h
21500 (maybe_hot_count_p, counts_to_freqs): UPdate.
21501 * print-rtl-function.c: Do not include cfg.h
21502 * print-rtl.c: Include basic-block.h
21503 * profile-count.c: New file.
21504 * profile-count.h: New file.
21505 * profile.c (is_edge_inconsistent): Update.
21506 (correct_negative_edge_counts): Update.
21507 (is_inconsistent): Update.
21508 (set_bb_counts): Update.
21509 (read_profile_edge_counts): Update.
21510 (compute_frequency_overlap): Update.
21511 (compute_branch_probabilities): Update; Initialize and deinitialize
21513 (branch_prob): Update.
21514 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21515 (edge_gcov_count): New.
21516 (bb_gcov_count): New.
21517 * shrink-wrap.c (try_shrink_wrapping): Update.
21518 * tracer.c (better_p): Update.
21519 * trans-mem.c (expand_transaction): Update.
21520 (ipa_tm_insert_irr_call): Update.
21521 (ipa_tm_insert_gettmclone_call): Update.
21522 * tree-call-cdce.c: Update.
21523 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21524 (gimple_duplicate_sese_tail): Update.
21525 (gimple_account_profile_record): Update.
21526 (execute_fixup_cfg): Update.
21527 * tree-inline.c (copy_bb): Update.
21528 (copy_edges_for_bb): Update.
21529 (initialize_cfun): Update.
21530 (freqs_to_counts): Update.
21531 (copy_cfg_body): Update.
21532 (expand_call_inline): Update.
21533 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21534 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21535 (try_unroll_loop_completely): Update.
21536 (try_peel_loop): Update.
21537 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21538 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21539 * tree-ssa-loop-split.c (connect_loops): Update.
21540 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21541 * tree-ssa-reassoc.c (branch_fixup): Update.
21542 * tree-ssa-tail-merge.c (replace_block_by): Update.
21543 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21544 (compute_path_counts): Update.
21545 (update_profile): Update.
21546 (recompute_probabilities): Update.
21547 (update_joiner_offpath_counts): Update.
21548 (estimated_freqs_path): Update.
21549 (freqs_to_counts_path): Update.
21550 (clear_counts_path): Update.
21551 (ssa_fix_duplicate_block_edges): Update.
21552 (duplicate_thread_path): Update.
21553 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21554 (struct switch_conv_info): Update.
21555 * tree-tailcall.c (decrease_profile): Update.
21556 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21557 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21558 * value-prof.c (check_counter): Update.
21559 (gimple_divmod_fixed_value): Update.
21560 (gimple_mod_pow2): Update.
21561 (gimple_mod_subtract): Update.
21562 (gimple_ic_transform): Update.
21563 (gimple_stringop_fixed_value): Update.
21564 * value-prof.h (gimple_ic): Update.
21566 2017-06-02 Carl Love <cel@us.ibm.com>
21568 * config/rs6000/rs6000-c: Add support for built-in functions
21569 vector double vec_doublee (vector signed int);
21570 vector double vec_doublee (vector unsigned int);
21571 vector double vec_doublee (vector float);
21572 vector double vec_doubleh (vector signed int);
21573 vector double vec_doubleh (vector unsigned int);
21574 vector double vec_doubleh (vector float);
21575 vector double vec_doublel (vector signed int);
21576 vector double vec_doublel (vector unsigned int);
21577 vector double vec_doublel (vector float);
21578 vector double vec_doubleo (vector signed int);
21579 vector double vec_doubleo (vector unsigned int);
21580 vector double vec_doubleo (vector float);.
21581 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21582 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21584 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21585 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21586 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21588 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21589 vec_doublel, vec_doubleh.
21590 * doc/extend.texi: Update the built-in documentation file for the
21591 new built-in functions.
21593 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21596 * ipa-inline-analysis.c (free_growth_caches): Set
21597 edge_removal_hook_holder to NULL after removing it.
21599 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21601 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21602 comparision with zero.
21604 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21605 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21606 for early expansion of vec_min and vec_max builtins.
21607 (builtin_function_type): Add min/max unsigned variants to those
21608 identified as having unsigned arguments.
21610 2017-06-02 Olivier Hainque <hainque@adacore.com>
21612 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21614 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21616 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21617 Use VALL_F16 iterator rather than VALL.
21619 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21621 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21622 Emit CBNZ inside loop when doing a strong exchange and comparing
21623 against zero. Generate the CC flags after the loop.
21625 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21627 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21628 (dl_section_ref): New.
21629 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21630 On AIX, append an expression to subtract the size of the
21631 section length to dl_section_ref.
21633 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21635 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21636 for early expansion of vector absolute builtins.
21638 2017-06-02 Richard Biener <rguenther@suse.de>
21640 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21641 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21643 2017-06-02 Richard Biener <rguenther@suse.de>
21645 PR tree-optimization/80948
21646 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21647 stmt order as well.
21649 2017-06-02 Richard Biener <rguenther@suse.de>
21651 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21653 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21654 for inductions as relevant.
21656 2017-06-02 Richard Biener <rguenther@suse.de>
21658 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21659 (vectorizable_induction): ... this. Remove dead code.
21661 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21663 * builtins. (expand_builtin_alloca): Remove second parameter and
21664 infer its value from the first parameter instead.
21665 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21667 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21669 PR rtl-optimization/80903
21670 * loop-doloop.c (add_test): Unshare sequence.
21672 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21674 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21676 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21678 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21680 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21681 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21682 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21683 (xlogue_layout::get_stub_rtx): Make static.
21684 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21685 (xlogue_layout::compute_stub_managed_regs): Rename to...
21686 (xlogue_layout::count_stub_managed_regs): ...this.
21687 (xlogue_layout::is_stub_managed_reg): New function.
21688 (xlogue_layout::m_stub_names): Rename to...
21689 (xlogue_layout::s_stub_names): ...this, make static.
21690 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21691 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21692 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21693 xlogue_layout::s_stub_names): Instantiate statics.
21694 (stub_managed_regs): Remove.
21695 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21696 (disable_call_ms2sysv_xlogues): Rename to...
21697 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21698 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21700 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21701 change after reload_completed.
21702 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21704 (ix86_expand_prologue): Likewise.
21705 (ix86_expand_epilogue): Likewise.
21706 (ix86_expand_split_stack_prologue): Likewise.
21707 (ix86_compute_frame_layout): Remove frame parameter ...
21708 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21709 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21711 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21712 (ix86_frame): Move from here ...
21713 * config/i386/i386.h (ix86_frame): ... to here.
21714 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21715 complete ix86_frame data structure instead. Remove some_ld_name.
21717 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21719 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21720 symbols that hold a DECL_VALUE_EXPR.
21722 2017-06-01 Martin Jambor <mjambor@suse.cz>
21724 PR tree-optimization/80898
21725 * tree-sra.c (process_subtree_disqualification): Removed.
21726 (disqualify_candidate): Do not acll
21727 process_subtree_disqualification.
21728 (subtree_mark_written_and_enqueue): New function.
21729 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21730 RHS has been disqualified and re-queue LHS if necessary. Apart
21731 from that, ignore disqualified RHS.
21733 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21735 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21736 address fetch for z10 or later.
21738 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21740 * config/arc/arc.md (tst_movb): Add guard when splitting.
21742 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21744 * config/arc/arc.c (arc_can_eliminate): Test against
21745 arc_frame_pointer_needed.
21747 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21749 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21750 to prevent store reordering.
21751 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21752 (type): Add block type.
21753 (stack_tie): Define special instruction to be used in
21756 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21758 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21759 constraint. It is not valid for the pattern.
21760 (noncommutative_binary_comparison): Likewise.
21762 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21764 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21767 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21769 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21770 be used by the reg-alloc.
21772 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21774 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21775 reg-alloc when having mul64 or mul32x16 instructions.
21776 (mulsidi3): Likewise.
21777 (umulsidi3): Likewise.
21778 (mulsi32x16): New pattern.
21779 (mulsi64): Likewise.
21780 (mulsidi64): Likewise.
21781 (umulsidi64): Likewise.
21782 (MUL32x16_REG): Define.
21783 (mul64_600): Use MUL32x16_REG.
21784 (mac64_600): Likewise.
21785 (umul64_600): Likewise.
21786 (umac64_600): Likewise.
21788 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21790 * config/arc/arc.md (mulsi3_700): Make it commutative.
21792 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21794 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21796 (*sign_extendsidi2_insn): Likewise for movstosw.
21798 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21800 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21801 the type of the input discriminant value. Convert the
21802 discriminant value of signedness vary.
21804 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21806 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21807 Add to -Wall section.
21809 2017-06-01 Richard Biener <rguenther@suse.de>
21811 PR middle-end/66313
21812 * fold-const.c (fold_plusminus_mult_expr): If the factored
21813 factor may be zero use a wrapping type for the inner operation.
21814 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21815 and handle moved defs.
21816 (process_assignment): Properly guard the unary op case. Return a
21817 tri-state indicating that moving the stmt before the call may allow
21818 to continue. Pass through to_move.
21819 (find_tail_calls): Handle moving unrelated defs before
21822 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21825 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21826 splitter result in the canonical way.
21828 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21830 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21831 also for 32bit target. Update insn attributes.
21832 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21834 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21836 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21837 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21838 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21839 (_mm_maskz_min_ss): New intrinsics.
21841 2017-05-31 Martin Liska <mliska@suse.cz>
21843 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21844 Change comment style to one we normally use.
21845 (vectorizable_reduction): Likewise.
21846 (vectorizable_induction): Likewise.
21847 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21848 (vectorizable_call): Likewise.
21849 (vectorizable_simd_clone_call): Likewise.
21850 (vectorizable_conversion): Likewise.
21851 (vectorizable_assignment): Likewise.
21852 (vectorizable_shift): Likewise.
21853 (vectorizable_operation): Likewise.
21854 (vectorizable_store): Likewise.
21855 (vectorizable_load): Likewise.
21856 * tree-vectorizer.h: Likewise.
21858 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21860 * passes.c (emergency_dump_function): New.
21861 * tree-pass.h (emergency_dump_function): Declare.
21862 * plugin.c (plugins_internal_error_function): Remove.
21863 * plugin.h (plugins_internal_error_function): Remove declaration.
21864 * toplev.c (internal_error_function): New static function. Use it...
21865 (general_init): ...here.
21867 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21869 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21870 (arc_rtx_costs): Add costs for new patterns.
21871 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21872 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21874 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21876 * tree-ssa-strlen.c (get_next_strinfo): New function.
21877 (get_stridx_plus_constant): Use it.
21878 (zero_length_string): Likewise.
21879 (adjust_related_strinfos): Likewise.
21880 (adjust_last_stmt): Likewise.
21882 2017-05-31 Richard Biener <rguenther@suse.de>
21885 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21886 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21888 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21890 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21891 loop_vinfo argument and use of dependence distance vectors.
21892 Check instead whether the two references differ only in their
21893 initial value and assume that they have the same alignment if the
21894 difference is a multiple of the vector alignment.
21895 (vect_analyze_data_refs_alignment): Update call accordingly.
21897 2017-05-31 Martin Liska <mliska@suse.cz>
21900 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21902 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21904 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21905 (create_intersect_range_checks): Move from ...
21906 * tree-data-ref.c (create_intersect_range_checks_index)
21907 (create_intersect_range_checks): ... to here.
21908 (create_runtime_alias_checks): New function factored from ...
21909 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21910 here. Call above function.
21911 * tree-data-ref.h (create_runtime_alias_checks): New function.
21913 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21915 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21916 segment length for dr_b and compute it in wide_int.
21918 2017-05-31 Richard Biener <rguenther@suse.de>
21920 PR tree-optimization/80906
21921 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21922 and pass through iv_map.
21923 (copy_bb_and_scalar_dependences): Adjust.
21924 (translate_pending_phi_nodes): Likewise.
21925 (copy_loop_close_phi_args): Handle code-generating IVs instead
21928 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21930 * diagnostic-color.c (color_dict): Add "type-diff".
21931 (parse_gcc_colors): Update comment.
21932 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21933 -fdiagnostics-show-template-tree and -fno-elide-type.
21934 (GCC_COLORS): Add type-diff to example.
21936 (-fdiagnostics-show-template-tree): New.
21937 (-fno-elide-type): New.
21938 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21939 the pp_format_decoder callback. Call any m_format_postprocessor's
21941 (pretty_printer::pretty_printer): Initialize
21942 m_format_postprocessor.
21943 (pretty_printer::~pretty_printer): Delete any
21944 m_format_postprocessor.
21945 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21946 (class format_postprocessor): New class.
21947 (struct pretty_printer::format_decoder): Document the new parameters.
21948 (struct pretty_printer::m_format_postprocessor): New field.
21949 * tree-diagnostic.c (default_tree_printer): Update for new
21950 bool and const char ** params.
21951 * tree-diagnostic.h (default_tree_printer): Likewise.
21953 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21955 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21956 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21957 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21958 rs6000_gen_cell_microcode code.
21959 (rs6000_final_prescan_insn): Delete.
21960 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21961 "warn-cell-microcode" entries.
21962 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21963 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21964 throughout. Change cc_reg_not_micro_cr0_operand to
21965 cc_reg_not_cr0_operand throughout.
21966 (*extendhi<mode>2_noload): Delete.
21967 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21968 (mwarn-cell-microcode): Delete.
21969 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21970 -mgen-cell-microcode and -mwarn-cell-microcode.
21972 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21975 * config/i386/constraints.md (Yd): New constraint.
21977 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21978 and (?Yd, r) alternatives. Update insn attributes.
21979 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21980 and (?*Yd, r) alternatives. Update insn attributes.
21981 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21983 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21985 * gimplify.c (gimplify_modify_expr): Don't create a
21986 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21989 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21991 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21993 2017-05-30 Richard Biener <rguenther@suse.de>
21995 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21996 and reduc_def fields.
21997 (STMT_VINFO_REDUC_TYPE): New define.
21998 (STMT_VINFO_REDUC_DEF): Likewise.
21999 (vect_force_simple_reduction): Adjust prototype.
22000 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22001 (vect_is_simple_reduction): Remove check_reduction argument.
22002 (vect_force_simple_reduction): Adjust and set
22003 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22004 (vectorizable_reduction): Do not re-do reduction analysis
22005 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22006 * tree-parloops.c (gather_scalar_reductions): Adjust.
22008 2017-05-30 Richard Biener <rguenther@suse.de>
22010 PR middle-end/80901
22011 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22014 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22016 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22017 Introduce unknown_misalignment parameter and remove vf.
22018 (vect_peeling_hash_get_lowest_cost):
22019 Pass unknown_misalignment parameter.
22020 (vect_enhance_data_refs_alignment):
22021 Fix unsupportable data ref treatment.
22023 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22025 * tree-vect-data-refs.c (vect_get_data_access_cost):
22026 Workaround for SLP handling.
22027 (vect_enhance_data_refs_alignment):
22028 Compute costs for doing no peeling at all, compare to the best
22029 peeling costs so far and avoid peeling if cheaper.
22031 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22033 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22034 Return peeling info and set costs to zero for unlimited cost
22036 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22037 with unknown misalignment. Compute and costs for unknown
22038 misalignment, compare them to the costs for known misalignment
22039 and choose the cheapest for peeling.
22041 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22043 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22044 (vect_get_peeling_costs_all_drs): Create function.
22045 (vect_peeling_hash_get_lowest_cost):
22046 Use vect_get_peeling_costs_all_drs.
22047 (vect_peeling_supportable): Create function.
22048 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22050 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22052 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22053 DR_HAS_NEGATIVE_STEP.
22054 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22055 (vect_enhance_data_refs_alignment): Use.
22056 (vect_duplicate_ssa_name_ptr_info): Use.
22057 * tree-vectorizer.h (dr_misalignment): Use.
22058 (known_alignment_for_access_p): Use.
22060 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22063 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22065 (has_section_name): New function.
22067 2017-05-30 Martin Liska <mliska@suse.cz>
22070 * auto-profile.c (get_function_decl_from_block): Fix
22073 2017-05-30 Richard Biener <rguenther@suse.de>
22075 PR middle-end/80876
22076 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22078 2017-05-30 Martin Liska <mliska@suse.cz>
22080 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22081 * dumpfile.h (struct dump_file_info): Remove ctors.
22083 2017-05-30 Martin Liska <mliska@suse.cz>
22085 * predict.def: Fix GNU coding style.
22087 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22089 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22090 Mark 'to' argument with ATTRIBUTE_UNUSED.
22092 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22094 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22095 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22096 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22099 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22101 * doc/install.texi (Options specification): Restore entry of
22102 --enable-sjlj-exceptions.
22104 2017-05-27 Michael Eager <eager@eagercon.com>
22107 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22109 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22111 * config/microblaze/microblaze.h
22112 (FIXED_REGISTERS): Update in macro.
22113 (CALL_USED_REGISTERS): Update in macro.
22115 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22117 * doc/install.texi: Add links to macOS binary distributions.
22119 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22123 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22125 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22127 2017-05-26 Martin Liska <mliska@suse.cz>
22129 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22131 2017-05-26 Martin Liska <mliska@suse.cz>
22133 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22134 always leading ';; '.
22135 (dump_bb_info): Likewise.
22136 (brief_dump_cfg): Likewise.
22137 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22138 * dumpfile.c: Remove usage of TDF_VERBOSE.
22139 * dumpfile.h (enum dump_kind): Likewise.
22140 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22141 * print-tree.c (debug_verbose): Remove.
22142 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22143 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22144 * tree-diagnostic.c (default_tree_printer): Replace
22145 TDF_DIAGNOSTIC with TDF_SLIM.
22147 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22149 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22150 in parameter loop, rather than loop_vinfo.
22151 (create_intersect_range_checks): Ditto.
22152 (vect_create_cond_for_alias_checks): Update call to above functions.
22154 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22156 PR tree-optimization/80815
22157 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22158 for merging runtime alias checks. Handle negative DR_STEPs.
22160 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22162 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22164 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22165 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22166 out code pruning runtime alias checks.
22167 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22168 factored out from above.
22169 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22171 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22173 (prune_runtime_alias_test_list): New decalaration.
22175 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22177 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22178 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22179 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22180 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22181 (operator==, comp_dr_with_seg_len_pair): Ditto.
22182 (vect_prune_runtime_alias_test_list): Ditto.
22184 2017-05-26 Martin Liska <mliska@suse.cz>
22187 * params.def: Bound partial-inlining-entry-probability param.
22189 2017-05-26 Marek Polacek <polacek@redhat.com>
22192 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22195 2017-05-26 Richard Biener <rguenther@suse.de>
22197 PR tree-optimization/80842
22198 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22201 2017-05-26 Richard Biener <rguenther@suse.de>
22203 PR tree-optimization/80844
22204 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22206 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22208 * doc/md.texi (Machine Constraints): Update x86 family
22209 machine constraints section to match 'config/i386/constraints.md'.
22211 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22213 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22215 2017-05-25 Nathan Sidwell <nathan@acm.org>
22217 * doc/invoke.texi (--enable-languages): Update documentation.
22219 2017-05-25 Martin Liska <mliska@suse.cz>
22221 * dumpfile.c: Add TDF_FOLDING.
22222 * dumpfile.h (enum dump_kind): Likewise.
22223 * genmatch.c (dt_simplify::gen_1): Use it.
22225 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22227 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22229 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22231 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22232 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22234 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22236 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22237 * match.pd (X == C): Rewrite it here.
22238 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22239 with_certain_nonzero_bits2): New predicates.
22240 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22242 2017-05-24 Nathan Sidwell <nathan@acm.org>
22244 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22247 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22250 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22252 * config/powerpcspe: New port. Files are copied from the rs6000
22253 port, with "rs6000" in filenames replaced by "powerpcspe".
22255 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22257 PR rtl-optimization/80754
22258 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22260 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22262 * config/sparc/sparc.md (length): Return the correct value for -mflat
22263 sibcalls to match output_sibcall.
22265 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22269 * config/rs6000/rs6000.c (struct machine_function): Add new field
22271 (rs6000_get_separate_components): Init that field, use it.
22272 (rs6000_components_for_bb): Use the field.
22274 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22276 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22278 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22280 PR middle-end/80823
22281 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22283 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22286 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22287 address against address_operand predicate.
22288 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22290 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22292 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22293 parameters passed indirectly.
22295 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22297 * config/i386/i386.md (*movdi_internal): Remove SSE4
22298 alternative 18 (?r, *v). Update insn attributes.
22299 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22300 Update insn attributes.
22301 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22302 Update insn attributes.
22303 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22304 alternative 1 (r, v). Remove isa attribute.
22305 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22306 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22307 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22309 2017-05-23 Tom de Vries <tom@codesourcery.com>
22311 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22314 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22316 * cgraphunit.c (symbol_table::process_new_functions): Update.
22317 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22318 (inline_generate_summary): Rename to ...
22319 (ipa_fn_summary_generate): ... this one.
22320 (inline_read_summary): Rename to ...
22321 (ipa_fn_summary_read): ... this one.
22322 (inline_write_summary): Rename to ...
22323 (ipa_fn_summary_write): ... this one.
22324 (inline_free_summary): Rename to ...
22325 (ipa_free_fn_summary): ... this one.
22326 (pass_data_local_fn_summary, pass_local_fn_summary,
22327 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22328 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22329 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22330 make_pass_ipa_fn_summary): New.
22331 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22332 inline_write_summary, inline_free_summary): Remove.
22333 (ipa_free_fn_summary) : New.
22334 * ipa-inline.c (ipa_inline): Update.
22335 (pass_ipa_inline): Do not generate summaries.
22336 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22338 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22339 and add pass_ipa_fn_summary.
22340 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22342 (make_pass_inline_parameters): Remove.
22344 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22346 * omp-low.c (struct omp_context): Remove "default_kind" member.
22349 * omp-offload.c (execute_oacc_device_lower): Remove the
22350 parallelism dimensions function attributes for unparallelized
22351 OpenACC kernels constructs.
22353 2017-05-23 Martin Liska <mliska@suse.cz>
22355 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22357 (cgraph_edge::make_speculative): Likewise.
22358 (cgraph_edge::resolve_speculation): Likewise.
22359 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22360 (cgraph_node::dump): Likewise.
22361 * cgraph.h: Likewise.
22362 * cgraphunit.c (analyze_functions): Likewise.
22363 (symbol_table::compile): Likewise.
22364 * ipa-cp.c (print_all_lattices): Likewise.
22365 (determine_versionability): Likewise.
22366 (initialize_node_lattices): Likewise.
22367 (ipcp_verify_propagated_values): Likewise.
22368 (estimate_local_effects): Likewise.
22369 (update_profiling_info): Likewise.
22370 (create_specialized_node): Likewise.
22371 (perhaps_add_new_callers): Likewise.
22372 (decide_about_value): Likewise.
22373 (decide_whether_version_node): Likewise.
22374 (identify_dead_nodes): Likewise.
22375 (ipcp_store_bits_results): Likewise.
22376 * ipa-devirt.c (dump_targets): Likewise.
22377 (ipa_devirt): Likewise.
22378 * ipa-icf.c (sem_item::dump): Likewise.
22379 (sem_function::equals): Likewise.
22380 (sem_variable::equals): Likewise.
22381 (sem_item_optimizer::read_section): Likewise.
22382 (sem_item_optimizer::execute): Likewise.
22383 (congruence_class::dump): Likewise.
22384 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22385 (dump_inline_summary): Likewise.
22386 (estimate_node_size_and_time): Likewise.
22387 (inline_analyze_function): Likewise.
22388 * ipa-inline-transform.c (inline_call): Likewise.
22389 * ipa-inline.c (report_inline_failed_reason): Likewise.
22390 (want_early_inline_function_p): Likewise.
22391 (edge_badness): Likewise.
22392 (update_edge_key): Likewise.
22393 (inline_small_functions): Likewise.
22394 * ipa-profile.c (ipa_profile): Likewise.
22395 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22396 (ipa_make_edge_direct_to_target): Likewise.
22397 (remove_described_reference): Likewise.
22398 (ipa_impossible_devirt_target): Likewise.
22399 (propagate_controlled_uses): Likewise.
22400 (ipa_print_node_params): Likewise.
22401 (ipcp_transform_function): Likewise.
22402 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22403 (propagate_pure_const): Likewise.
22404 * ipa-reference.c (generate_summary): Likewise.
22405 (read_write_all_from_decl): Likewise.
22406 (propagate): Likewise.
22407 (ipa_reference_read_optimization_summary): Likewise.
22408 * ipa-utils.c (ipa_merge_profiles): Likewise.
22409 * ipa.c (walk_polymorphic_call_targets): Likewise.
22410 (symbol_table::remove_unreachable_nodes): Likewise.
22411 (ipa_single_use): Likewise.
22412 * passes.c (execute_todo): Likewise.
22413 * predict.c (drop_profile): Likewise.
22414 * symtab.c (symtab_node::get_dump_name): New function.
22415 (symtab_node::dump_name): Likewise.
22416 (symtab_node::dump_asm_name): Likewise.
22417 (symtab_node::dump_references): Likewise.
22418 (symtab_node::dump_referring): Likewise.
22419 (symtab_node::dump_base): Likewise.
22420 (symtab_node::debug_symtab): Likewise.
22421 * tree-sra.c (convert_callers_for_node): Likewise.
22422 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22423 * value-prof.c (init_node_map): Likewise.
22425 2017-05-23 Martin Liska <mliska@suse.cz>
22427 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22428 and symtab_node::debug_symtab to symbol_table::debug.
22429 * cgraphunit.c (analyze_functions): Use the renamed function.
22430 (symbol_table::compile): Likewise.
22431 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22432 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22433 * passes.c (execute_todo): Likewise.
22434 * symtab.c (symbol_table::dump): New function.
22435 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22437 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22439 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22440 that nonconst implies exec.
22442 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22444 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22445 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22446 (inline_edge_summary_vec): Turn into ...
22447 (ipa_call_summaries): ... this one.
22448 (redirect_to_unreachable, edge_set_predicate,
22449 evaluate_properties_for_edge, inline_summary_alloc,
22450 reset_ipa_call_summary, reset_inline_summary,
22451 inline_summary_t::duplicate): Update.
22452 (inline_edge_duplication_hook): Turn to ...
22453 (ipa_call_summary_t::duplicate): ... this one.
22454 (inline_edge_removal_hook): Turn to ...
22455 (ipa_call_summary_t::remove): ... this one.
22456 (dump_inline_edge_summary): Turn to ...
22457 (dump_ipa_call_summary): ... this one.
22458 (estimate_function_body_sizes): Update.
22459 (inline_update_callee_summaries): Update.
22460 (remap_edge_change_prob): Update.
22461 (remap_edge_summaries): Update.
22462 (inline_merge_summary): Update.
22463 (do_estimate_edge_time): Update.
22464 (inline_generate_summary): Update.
22465 (inline_read_section): Update.
22466 (inline_read_summary): Update.
22467 (inline_free_summary): Update.
22468 * ipa-inline.c (can_inline_edge_p): Update.
22469 (compute_inlined_call_time): Update.
22470 (want_inline_small_function_p): Update.
22471 (edge_badness): Update.
22472 (early_inliner): Update.
22473 * ipa-inline.h (inline_edge_summary): Turn to ...
22474 (ipa_call_summary): ... this one.
22475 (ipa_call_summary_t): New class.
22476 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22477 (ipa_call_summaries): New.
22478 (inline_edge_summary): Remove.
22479 (estimate_edge_growth): Update.
22480 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22481 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22482 * ipa-split.c (execute_split_functions): Update.
22483 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22485 2017-05-23 Tom de Vries <tom@codesourcery.com>
22487 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22488 attributes): Document rdrand effective target.
22490 2017-05-23 Tom de Vries <tom@codesourcery.com>
22492 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22493 attributes): Sort alphabetically.
22495 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22497 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22499 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22502 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22503 V2DF/V2DI splat into two separate patterns, one that handles
22504 registers, and the other that only handles memory. Drop support
22505 for splatting from a GPR on ISA 2.07 and then splitting the
22506 splat into direct move and splat.
22507 (vsx_splat_<mode>_reg): Likewise.
22508 (vsx_splat_<mode>_mem): Likewise.
22510 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22512 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22514 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22516 PR middle-end/80809
22517 * omp-low.c (finish_taskreg_remap): New function.
22518 (finish_taskreg_scan): If unit size of ctx->record_type
22519 is non-constant, unshare the size expression and replace
22520 decls in it with possible outer var refs.
22522 PR middle-end/80809
22523 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22524 GOVD_SHARED rather than GOVD_PRIVATE with it.
22525 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22526 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22528 PR middle-end/80853
22529 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22530 as last argument to build_outer_var_ref for pointer bases of array
22531 section reductions.
22533 2017-05-19 Martin Sebor <msebor@redhat.com>
22535 * print-tree.c (print_node): Print DECL_READ_P flag.
22537 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22539 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22540 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22541 * cgraph.c: Likewise.
22542 * cgraphunit.c: Likewise.
22543 * gengtype.c: Likewise.
22544 * ipa-cp.c: Likewise.
22545 * ipa-devirt.c: Likewise.
22546 * ipa-icf.c: Likewise.
22547 * ipa-predicate.c: Likewise.
22548 * ipa-profile.c: Likewise.
22549 * ipa-prop.c: Likewise.
22550 * ipa-split.c: Likewise.
22552 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22553 edge_predicate_pool, dump_inline_hints,
22554 inline_summary::account_size_time, redirect_to_unreachable,
22555 edge_set_predicate, set_hint_predicate,
22556 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22557 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22558 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22559 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22560 ipa_call_summary_t::remove, initialize_growth_caches,
22561 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22562 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22563 mark_modified, unmodified_parm_1, unmodified_parm,
22564 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22565 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22566 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22567 will_be_nonconstant_predicate, record_modified_bb_info,
22568 get_minimal_bb, record_modified, param_change_prob,
22569 phi_result_unknown_predicate, predicate_for_phi_result,
22570 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22571 estimate_function_body_sizes, compute_inline_parameters,
22572 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22573 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22574 inline_update_callee_summaries, remap_edge_change_prob,
22575 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22576 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22577 inline_analyze_function, inline_summary_t::insert,
22578 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22579 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22580 inline_free_summary): Move to ipa-fnsummary.h
22581 (predicate_t): Remove.
22582 * ipa-fnsummary.c: New file.
22583 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22584 (enum inline_hints_vals, inline_hints, agg_position_info,
22585 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22586 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22587 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22588 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22589 inline_read_summary, inline_write_summary, inline_free_summary,
22590 inline_analyze_function, initialize_inline_failed,
22591 inline_merge_summary, inline_update_overall_summary,
22592 compute_inline_parameters): Move to ipa-fnsummary.h
22593 * ipa-fnsummary.h: New file.
22594 * ipa-inline-transform.h: Include ipa-inline.h.
22595 * ipa-inline.c: LIkewise.
22597 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22599 * ipa-inline.c (edge_badness): Use inlined_time instead of
22600 inline_summaries->get.
22602 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22604 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22606 2017-05-22 Nathan Sidwell <nathan@acm.org>
22608 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22609 (fdump-lang): Document 'raw' option.
22610 * dumpfile.h (TDI_tu): Delete.
22611 * dumpfile.c (dump_files): Remove translation-unit.
22612 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22614 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22616 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22617 command option from $(AWK) call.
22618 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22619 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22620 [FORMAT]: Remove handling of variable.
22621 * config/avr/t-multilib: Regenerate.
22623 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22625 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22627 (dump_inline_summary): Do not print self_time.
22628 (estimate_function_body_sizes): Do not set self_time.
22629 (compute_inline_parameters): Likewise.
22630 (inline_read_section, inline_write_summary): Do not stream self_time.
22631 * ipa-inline.h (inline_summary): Drop self_time.
22633 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22635 * ipa-inline-analysis.c (account_size_time): Rename to ...
22636 (inline_summary::account_size_time): ... this one.
22637 (reset_ipa_call_summary): Turn to ...
22638 (ipa_call_summary::reset): ... this one.
22639 (reset_inline_summary): Turn to ...
22640 (inline_summary::reset): ... this one.
22641 (inline_summary_t::remove): Update.
22642 (inline_summary_t::duplicate): Update.
22643 (ipa_call_summary_t::remove): Update.
22644 (dump_inline_summary): Update.
22645 (estimate_function_body_sizes): Update.
22646 (compute_inline_parameters): Update.
22647 (estimate_node_size_and_time): Update.
22648 (inline_merge_summary): Update.
22649 (inline_update_overall_summary): Update.
22650 (inline_read_section): Update.
22651 (inline_write_summary): Update.
22652 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22653 add account_size_time and reset member functions.
22654 (ipa_call_summary): Add reset function.
22655 * ipa-predicate.h (predicate::operator &): Constify.
22657 2017-05-22 Richard Biener <rguenther@suse.de>
22659 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22661 2017-05-19 Jason Merrill <jason@redhat.com>
22663 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22665 2017-05-19 Marek Polacek <polacek@redhat.com>
22668 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22669 TYPE_OVERFLOW_WRAPS checks.
22671 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22673 * tree-core.h (enum omp_clause_default_kind): Add
22674 "OMP_CLAUSE_DEFAULT_PRESENT".
22675 * tree-pretty-print.c (dump_omp_clause): Handle it.
22676 * gimplify.c (enum gimplify_omp_var_data): Add
22677 "GOVD_MAP_FORCE_PRESENT".
22678 (gimplify_adjust_omp_clauses_1): Map it to
22679 "GOMP_MAP_FORCE_PRESENT".
22680 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22682 * gimplify.c (oacc_default_clause): Clarify.
22684 2017-05-19 Nathan Sidwell <nathan@acm.org>
22686 LANG_HOOK_REGISTER_DUMPS
22687 * toplev.c (general_init): Call register dump lang hook.
22688 * doc/invoke.texi: Document -fdump-lang option family.
22689 * dumpfile.c (dump_files): Remove class dump here.
22690 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22691 * dumpfile.h (tree_dump_index): Remove TDI_class.
22692 * langhooks-def.h (lhd_register_dumps): Declare.
22693 (LANG_HOOKS_REGISTER_DUMPS): Define.
22694 (LANG_HOOKS_INITIALIZER): Add it.
22695 * langhooks.c (lhd_register_dumps): Define.
22696 * langhooks.h (struct lang_hooks): Add register_dumps.
22698 2017-05-19 Nathan Sidwell <nathan@acm.org>
22700 * context.h (context::set_passes): New.
22701 * context.c (context::context): Do not create pass manager.
22702 * toplev.c (general_init): Create pass manager here.
22704 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22706 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22707 use this splitter if two add or or instructions would also work for
22708 the constant we want to generate.
22710 2017-05-19 Richard Biener <rguenther@suse.de>
22713 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22714 predicate evaluation.
22716 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22718 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22720 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22723 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22725 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22726 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22727 (inline_edge_summary_vec): Turn into ...
22728 (ipa_call_summaries): ... this one.
22729 (redirect_to_unreachable, edge_set_predicate,
22730 evaluate_properties_for_edge, inline_summary_alloc,
22731 reset_ipa_call_summary, reset_inline_summary,
22732 inline_summary_t::duplicate): Update.
22733 (inline_edge_duplication_hook): Turn to ...
22734 (ipa_call_summary_t::duplicate): ... this one.
22735 (inline_edge_removal_hook): Turn to ...
22736 (ipa_call_summary_t::remove): ... this one.
22737 (dump_inline_edge_summary): Turn to ...
22738 (dump_ipa_call_summary): ... this one.
22739 (estimate_function_body_sizes): Update.
22740 (inline_update_callee_summaries): Update.
22741 (remap_edge_change_prob): Update.
22742 (remap_edge_summaries): Update.
22743 (inline_merge_summary): Update.
22744 (do_estimate_edge_time): Update.
22745 (inline_generate_summary): Update.
22746 (inline_read_section): Update.
22747 (inline_read_summary): Update.
22748 (inline_free_summary): Update.
22749 * ipa-inline.c (can_inline_edge_p): Update.
22750 (compute_inlined_call_time): Update.
22751 (want_inline_small_function_p): Update.
22752 (edge_badness): Update.
22753 (early_inliner): Update.
22754 * ipa-inline.h (inline_edge_summary): Turn to ...
22755 (ipa_call_summary): ... this one.
22756 (ipa_call_summary_t): New class.
22757 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22758 (ipa_call_summaries): New.
22759 (inline_edge_summary): Remove.
22760 (estimate_edge_growth): Update.
22761 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22762 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22763 * ipa-split.c (execute_split_functions): Update.
22764 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22766 2017-05-19 Richard Biener <rguenther@suse.de>
22768 PR middle-end/80764
22769 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22771 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22773 * config/rs6000/rs6000.c (struct machine_function): Add field
22774 fpr_is_wrapped_separately.
22775 (rs6000_get_separate_components): Use 64 components. Handle the
22776 new FPR components.
22777 (rs6000_components_for_bb): Handle the FPR components.
22778 (rs6000_emit_prologue_components): Handle the FPR components.
22779 (rs6000_emit_epilogue_components): Handle the FPR components.
22780 (rs6000_set_handled_components): Handle the FPR components.
22781 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22782 that are already separately shrink-wrapped.
22783 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22784 that are already separately shrink-wrapped.
22786 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22789 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22792 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22793 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22794 cases where the register allocator uses a move and an offsettable
22795 memory operation to/from a FPR register on ISA 2.06/2.07.
22796 (define_peephole2 for Altivec d-form store): Likewise.
22798 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22801 * config/i386/mmx.md (*mov<mode>_internal): Enable
22802 alternatives 11, 12, 13 and 14 also for 32bit targets.
22803 Remove alternatives 15, 16, 17 and 18.
22804 * config/i386/sse.md (vec_concatv2di): Change
22805 alternative (!x, *y) to (x, ?!*Yn).
22807 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22809 * dumpfile.h (enum dump_kind): Remove stray comma.
22811 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22813 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22814 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22815 predicate::num_conditions
22816 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22817 (CHANGED): turn into predicate::changed.
22818 (agg_position_info): Move to ipa-predicate.h
22819 (add_condition, predicate::add_clause, predicate::operator &=,
22820 predicate::or_with, predicate::evaluate, predicate::probability,
22821 dump_condition, dump_clause, predicate::dump,
22822 predicate::remap_after_duplication, predicate::remap_after_inlining,
22823 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22824 (evaluate_conditions_for_known_args): Update.
22825 (set_cond_stmt_execution_predicate): Update.
22826 * ipa-inline.h: Include ipa-predicate.h
22827 (condition, inline_param_summary, conditions, agg_position_info,
22828 predicate): Move to ipa-predicate.h
22829 * ipa-predicate.c: New file.
22830 * ipa-predicate.h: New file.
22832 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22834 * final.c (leaf_function_p): Check we are not in a sequence.
22836 2017-05-18 Martin Liska <mliska@suse.cz>
22838 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22839 * dumpfile.c (dump_register): Use new enum dump_kind.
22840 (get_dump_file_name): Likewise.
22841 (dump_enable_all): Likewise.
22842 (dump_switch_p_1): Likewise.
22843 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22844 * dumpfile.h (enum dump_kind): New enum type.
22845 (struct dump_file_info): Create constructor and
22846 format fields and comments.
22847 * passes.c (pass_manager::register_one_dump_file):
22849 * statistics.c (statistics_early_init): Likewise.
22850 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22851 TDF_TREE with TDF_SLIM.
22852 (gather_memory_references_ref): Likewise.
22854 2017-05-18 Martin Liska <mliska@suse.cz>
22856 * vec.h (struct vnull): Use it.
22858 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22860 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22861 (true_predicate, false_predicate, true_predicate_p,
22862 false_predicate_p): Remove.
22863 (single_cond_predicate, not_inlined_predicate): Turn to member function
22865 (add_condition): Update.
22866 (add_clause): Turn to...
22867 (predicate::add_clause): ... this one; update; allow passing NULL
22869 (and_predicates): Turn to ...
22870 (predicate::operator &=): ... this one.
22871 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22872 (or_predicates): Turn to ...
22873 (predicate::or_with): ... this one.
22874 (evaluate_predicate): Turn to ...
22875 (predicate::evaluate): ... this one.
22876 (predicate_probability): Turn to ...
22877 (predicate::probability): ... this one.
22878 (dump_condition): Update.
22879 (dump_predicate): Turn to ...
22880 (predicate::dump): ... this one.
22881 (account_size_time): Update.
22882 (edge_set_predicate): Update.
22883 (set_hint_predicate): UPdate.
22884 (evaluate_conditions_for_known_args): Update.
22885 (evaluate_properties_for_edge): Update.
22886 (remap_predicate_after_duplication): Turn to...
22887 (predicate::remap_after_duplication): ... this one.
22888 (remap_hint_predicate_after_duplication): Update.
22889 (inline_summary_t::duplicate): UPdate.
22890 (dump_inline_edge_summary): Update.
22891 (dump_inline_summary): Update.
22892 (set_cond_stmt_execution_predicate): Update.
22893 (set_switch_stmt_execution_predicate): Update.
22894 (compute_bb_predicates): Update.
22895 (will_be_nonconstant_expr_predicate): Update.
22896 (will_be_nonconstant_predicate): Update.
22897 (phi_result_unknown_predicate): Update.
22898 (predicate_for_phi_result): Update.
22899 (array_index_predicate): Update.
22900 (estimate_function_body_sizes): Update.
22901 (estimate_node_size_and_time): Update.
22902 (estimate_ipcp_clone_size_and_time): Update.
22903 (remap_predicate): Rename to ...
22904 (predicate::remap_after_inlining): ... this one.
22905 (remap_hint_predicate): Update.
22906 (inline_merge_summary): Update.
22907 (inline_update_overall_summary): Update.
22908 (estimate_size_after_inlining): Update.
22909 (read_predicate): Rename to ...
22910 (predicate::stream_in): ... this one.
22911 (read_inline_edge_summary): Update.
22912 (write_predicate): Rename to ...
22913 (predicate::stream_out): ... this one.
22914 (write_inline_edge_summary): Update.
22915 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22916 (clause_t): Turn to uint32_t
22917 (predicate): Turn to class; implement constructor and operators
22919 (size_time_entry): Update.
22920 (inline_summary): Update.
22921 (inline_edge_summary): Update.
22923 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22925 * fold-const.c (fold_binary_loc): Move transformation...
22926 * match.pd (C - X CMP X): ... here.
22928 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22930 * config/sparc/sparc.c (sparc_option_override): Set function
22931 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22932 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22934 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22936 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22938 2017-05-18 Marek Polacek <polacek@redhat.com>
22941 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22942 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22943 gimple_assign_load_p.
22945 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22947 PR middle-end/80692
22948 * real.c (do_compare): Give decimal_do_compare preference over
22949 comparing just the signs.
22951 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22953 * doc/md.texi (Canonicalization of Instructions): Describe the
22954 canonical form of instructions that inherently set a condition
22957 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22959 PR middle-end/80775
22960 * tree-cfg.c: Move deletion of unreachable case statements to after
22961 the merging of consecutive case labels.
22963 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22965 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22966 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22967 restoring of callee-saved registers.
22969 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22971 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22972 * config/visium/visium.c (single_set_and_flags): Likewise.
22973 * config/visium/visium.md (Substitutions): Likewise.
22975 2017-05-17 Martin Liska <mliska@suse.cz>
22977 * cfg.c: Introduce dump_flags_t type and
22978 use it instead of int type.
22980 * cfghooks.c: Likewise.
22981 * cfghooks.h (struct cfg_hooks): Likewise.
22982 * cfgrtl.c: Likewise.
22983 * cfgrtl.h: Likewise.
22984 * cgraph.c (cgraph_node::get_body): Likewise.
22985 * coretypes.h: Likewise.
22986 * domwalk.c: Likewise.
22987 * domwalk.h: Likewise.
22988 * dumpfile.c (struct dump_option_value_info): Likewise.
22989 (dump_enable_all): Likewise.
22990 (dump_switch_p_1): Likewise.
22991 (opt_info_switch_p): Likewise.
22992 * dumpfile.h (enum tree_dump_index): Likewise.
22993 (struct dump_file_info): Likewise.
22994 * genemit.c: Likewise.
22995 * generic-match-head.c: Likewise.
22996 * gengtype.c (open_base_files): Likewise.
22997 * gimple-pretty-print.c: Likewise.
22998 * gimple-pretty-print.h: Likewise.
22999 * graph.c (print_graph_cfg): Likewise.
23000 * graphite-scop-detection.c (dot_all_sese): Likewise.
23001 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23002 * loop-unroll.c (report_unroll): Likewise.
23003 * passes.c (pass_manager::register_one_dump_file): Likewise.
23004 * print-tree.c: Likewise.
23005 * statistics.c: Likewise.
23006 * tree-cfg.c: Likewise.
23007 * tree-cfg.h: Likewise.
23008 * tree-dfa.c: Likewise.
23009 * tree-dfa.h: Likewise.
23010 * tree-dump.c (dump_function): Likewise.
23011 * tree-dump.h (struct dump_info): Likewise.
23012 * tree-pretty-print.c: Likewise.
23013 * tree-pretty-print.h: Likewise.
23014 * tree-ssa-live.c: Likewise.
23015 * tree-ssa-live.h: Likewise.
23016 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23017 * tree-vect-loop.c: Likewise.
23018 * tree-vect-slp.c: Likewise.
23020 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23021 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23023 PR tree-optimization/80457
23024 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23025 of all arguments to a statement as scalar_to_vec operations.
23026 (vectorizable_call): Adjust call to vect_model_simple_cost for
23028 (vectorizable_conversion): Likewise.
23029 (vectorizable_assignment): Likewise.
23030 (vectorizable_shift): Likewise.
23031 (vectorizable_operation): Likewise.
23032 (vectorizable_comparison): Likewise.
23033 (vect_is_simple_cond): Record the def types for operands.
23034 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23035 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23036 for statement argument count.
23038 2017-05-16 Carl Love <cel@us.ibm.com>
23040 * config/rs6000/rs6000-c: Add support for built-in functions
23041 vector unsigned long long vec_bperm (vector unsigned long long,
23042 vector unsigned char)
23043 vector signed long long vec_mule (vector signed int,
23045 vector unsigned long long vec_mule (vector unsigned int,
23046 vector unsigned int)
23047 vector signed long long vec_mulo (vector signed int,
23049 vector unsigned long long vec_mulo (vector unsigned int,
23050 vector unsigned int)
23051 vector signed char vec_sldw (vector signed char,
23052 vector signed char,
23054 vector unsigned char vec_sldw (vector unsigned char,
23055 vector unsigned char,
23057 vector signed short vec_sldw (vector signed short,
23058 vector signed short,
23060 vector unsigned short vec_sldw (vector unsigned short,
23061 vector unsigned short,
23063 vector signed int vec_sldw (vector signed int,
23066 vector unsigned int vec_sldw (vector unsigned int,
23067 vector unsigned int,
23069 vector signed long long vec_sldw (vector signed long long,
23070 vector signed long long,
23072 vector unsigned long long vec_sldw (vector unsigned long long,
23073 vector unsigned long long,
23075 * config/rs6000/rs6000-c: Add support for built-in functions
23076 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23077 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23078 * doc/extend.texi: Update the built-in documentation for the
23079 new built-in functions.
23081 2017-05-16 Marek Polacek <polacek@redhat.com>
23085 * tree.c (save_expr): Don't fold the expression.
23087 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23089 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23090 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23091 and (?*y,m). Update insn attributes.
23093 2017-05-16 Martin Liska <mliska@suse.cz>
23095 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23096 flags argument of print_gimple_stmt, print_gimple_expr,
23097 print_generic_stmt and print_generic_expr.
23098 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23099 * coretypes.h: Likewise.
23100 * except.c (dump_eh_tree): Likewise.
23101 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23102 * gimple-pretty-print.h: Likewise.
23103 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23104 (backprop::push_to_worklist): Likewise.
23105 (backprop::pop_from_worklist): Likewise.
23106 (backprop::process_use): Likewise.
23107 (backprop::intersect_uses): Likewise.
23108 (note_replacement): Likewise.
23109 * gimple-ssa-store-merging.c
23110 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23111 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23112 (pass_store_merging::execute): Likewise.
23113 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23114 (ssa_base_cand_dump_callback): Likewise.
23115 (dump_incr_vec): Likewise.
23116 (replace_refs): Likewise.
23117 (replace_mult_candidate): Likewise.
23118 (create_add_on_incoming_edge): Likewise.
23119 (create_phi_basis): Likewise.
23120 (insert_initializers): Likewise.
23121 (all_phi_incrs_profitable): Likewise.
23122 (introduce_cast_before_cand): Likewise.
23123 (replace_one_candidate): Likewise.
23124 * gimplify.c (gimplify_expr): Likewise.
23125 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23126 (set_rename): Likewise.
23127 (rename_uses): Likewise.
23128 (copy_loop_phi_nodes): Likewise.
23129 (add_close_phis_to_merge_points): Likewise.
23130 (copy_loop_close_phi_args): Likewise.
23131 (copy_cond_phi_args): Likewise.
23132 (graphite_copy_stmts_from_block): Likewise.
23133 (translate_pending_phi_nodes): Likewise.
23134 * graphite-poly.c (print_pdr): Likewise.
23135 (dump_gbb_cases): Likewise.
23136 (dump_gbb_conditions): Likewise.
23137 (print_scop_params): Likewise.
23138 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23139 (build_cross_bb_scalars_use): Likewise.
23140 (gather_bbs::before_dom_children): Likewise.
23141 * hsa-dump.c (dump_hsa_immed): Likewise.
23142 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23143 (get_replacement_map): Likewise.
23144 * ipa-inline-analysis.c (dump_condition): Likewise.
23145 (estimate_function_body_sizes): Likewise.
23146 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23147 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23148 * ipa-prop.c (ipa_dump_param): Likewise.
23149 (ipa_print_node_jump_functions_for_edge): Likewise.
23150 (ipa_modify_call_arguments): Likewise.
23151 (ipa_modify_expr): Likewise.
23152 (ipa_dump_param_adjustments): Likewise.
23153 (ipa_dump_agg_replacement_values): Likewise.
23154 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23155 * ipa-pure-const.c (check_stmt): Likewise.
23156 (pass_nothrow::execute): Likewise.
23157 * ipa-split.c (execute_split_functions): Likewise.
23158 * omp-offload.c (dump_oacc_loop_part): Likewise.
23159 (dump_oacc_loop): Likewise.
23160 * trans-mem.c (tm_log_emit): Likewise.
23161 (tm_memopt_accumulate_memops): Likewise.
23162 (dump_tm_memopt_set): Likewise.
23163 (dump_tm_memopt_transform): Likewise.
23164 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23165 (print_loop): Likewise.
23166 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23167 (chkp_gather_checks_info): Likewise.
23168 (chkp_get_check_result): Likewise.
23169 (chkp_remove_check_if_pass): Likewise.
23170 (chkp_use_outer_bounds_if_possible): Likewise.
23171 (chkp_reduce_bounds_lifetime): Likewise.
23172 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23173 (chkp_mark_completed_bounds): Likewise.
23174 (chkp_register_incomplete_bounds): Likewise.
23175 (chkp_mark_invalid_bounds): Likewise.
23176 (chkp_maybe_copy_and_register_bounds): Likewise.
23177 (chkp_build_returned_bound): Likewise.
23178 (chkp_get_bound_for_parm): Likewise.
23179 (chkp_build_bndldx): Likewise.
23180 (chkp_get_bounds_by_definition): Likewise.
23181 (chkp_generate_extern_var_bounds): Likewise.
23182 (chkp_get_bounds_for_decl_addr): Likewise.
23183 * tree-chrec.c (chrec_apply): Likewise.
23184 * tree-data-ref.c (dump_data_reference): Likewise.
23185 (dump_subscript): Likewise.
23186 (dump_data_dependence_relation): Likewise.
23187 (analyze_overlapping_iterations): Likewise.
23188 * tree-inline.c (expand_call_inline): Likewise.
23189 (tree_function_versioning): Likewise.
23190 * tree-into-ssa.c (dump_defs_stack): Likewise.
23191 (dump_currdefs): Likewise.
23192 (dump_names_replaced_by): Likewise.
23193 (dump_update_ssa): Likewise.
23194 (update_ssa): Likewise.
23195 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23196 * tree-parloops.c (build_new_reduction): Likewise.
23197 (try_create_reduction_list): Likewise.
23198 (ref_conflicts_with_region): Likewise.
23199 (oacc_entry_exit_ok_1): Likewise.
23200 (oacc_entry_exit_single_gang): Likewise.
23201 * tree-pretty-print.h: Likewise.
23202 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23203 (get_scalar_evolution): Likewise.
23204 (add_to_evolution): Likewise.
23205 (get_loop_exit_condition): Likewise.
23206 (analyze_evolution_in_loop): Likewise.
23207 (analyze_initial_condition): Likewise.
23208 (analyze_scalar_evolution): Likewise.
23209 (instantiate_scev): Likewise.
23210 (number_of_latch_executions): Likewise.
23211 (gather_chrec_stats): Likewise.
23212 (final_value_replacement_loop): Likewise.
23213 (scev_const_prop): Likewise.
23214 * tree-sra.c (dump_access): Likewise.
23215 (disqualify_candidate): Likewise.
23216 (create_access): Likewise.
23217 (reject): Likewise.
23218 (maybe_add_sra_candidate): Likewise.
23219 (create_access_replacement): Likewise.
23220 (analyze_access_subtree): Likewise.
23221 (analyze_all_variable_accesses): Likewise.
23222 (sra_modify_assign): Likewise.
23223 (initialize_constant_pool_replacements): Likewise.
23224 (find_param_candidates): Likewise.
23225 (decide_one_param_reduction): Likewise.
23226 (replace_removed_params_ssa_names): Likewise.
23227 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23228 * tree-ssa-copy.c (dump_copy_of): Likewise.
23229 (copy_prop_visit_cond_stmt): Likewise.
23230 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23231 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23232 (record_equivalences_from_stmt): Likewise.
23233 * tree-ssa-dse.c (compute_trims): Likewise.
23234 (delete_dead_call): Likewise.
23235 (delete_dead_assignment): Likewise.
23236 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23237 (forward_propagate_into_cond): Likewise.
23238 (pass_forwprop::execute): Likewise.
23239 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23240 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23242 (move_computations_worker): Likewise.
23243 (execute_sm): Likewise.
23244 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23245 (remove_exits_and_undefined_stmts): Likewise.
23246 (remove_redundant_iv_tests): Likewise.
23247 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23248 (adjust_iv_update_pos): Likewise.
23249 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23250 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23251 (value_replacement): Likewise.
23252 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23253 * tree-ssa-pre.c (print_pre_expr): Likewise.
23254 (get_representative_for): Likewise.
23255 (create_expression_by_pieces): Likewise.
23256 (insert_into_preds_of_block): Likewise.
23257 (eliminate_insert): Likewise.
23258 (eliminate_dom_walker::before_dom_children): Likewise.
23259 (eliminate): Likewise.
23260 (remove_dead_inserted_code): Likewise.
23261 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23262 * tree-ssa-reassoc.c (get_rank): Likewise.
23263 (eliminate_duplicate_pair): Likewise.
23264 (eliminate_plus_minus_pair): Likewise.
23265 (eliminate_not_pairs): Likewise.
23266 (undistribute_ops_list): Likewise.
23267 (eliminate_redundant_comparison): Likewise.
23268 (update_range_test): Likewise.
23269 (optimize_range_tests_var_bound): Likewise.
23270 (optimize_vec_cond_expr): Likewise.
23271 (rewrite_expr_tree): Likewise.
23272 (rewrite_expr_tree_parallel): Likewise.
23273 (linearize_expr): Likewise.
23274 (break_up_subtract): Likewise.
23275 (linearize_expr_tree): Likewise.
23276 (attempt_builtin_powi): Likewise.
23277 (attempt_builtin_copysign): Likewise.
23278 (transform_stmt_to_copy): Likewise.
23279 (transform_stmt_to_multiply): Likewise.
23280 (dump_ops_vector): Likewise.
23281 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23282 (print_scc): Likewise.
23283 (set_ssa_val_to): Likewise.
23284 (visit_reference_op_store): Likewise.
23285 (visit_use): Likewise.
23286 (sccvn_dom_walker::before_dom_children): Likewise.
23287 (run_scc_vn): Likewise.
23288 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23290 (expr_hash_elt::print): Likewise.
23291 (const_and_copies::pop_to_marker): Likewise.
23292 (const_and_copies::record_const_or_copy_raw): Likewise.
23293 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23294 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23295 (dump_predicates): Likewise.
23296 (find_uninit_use): Likewise.
23297 (warn_uninitialized_phi): Likewise.
23298 (pass_late_warn_uninitialized::execute): Likewise.
23299 * tree-ssa.c (verify_vssa): Likewise.
23300 (verify_ssa): Likewise.
23301 (maybe_optimize_var): Likewise.
23302 * tree-vrp.c (dump_value_range): Likewise.
23303 (dump_all_value_ranges): Likewise.
23304 (dump_asserts_for): Likewise.
23305 (register_edge_assert_for_2): Likewise.
23306 (vrp_visit_cond_stmt): Likewise.
23307 (vrp_visit_switch_stmt): Likewise.
23308 (vrp_visit_stmt): Likewise.
23309 (vrp_visit_phi_node): Likewise.
23310 (simplify_cond_using_ranges_1): Likewise.
23311 (fold_predicate_in): Likewise.
23312 (evrp_dom_walker::before_dom_children): Likewise.
23313 (evrp_dom_walker::push_value_range): Likewise.
23314 (evrp_dom_walker::pop_value_range): Likewise.
23315 (execute_early_vrp): Likewise.
23317 2017-05-16 Richard Biener <rguenther@suse.de>
23319 * dwarf2out.c (loc_list_from_tree_1): Do not create
23320 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23322 2017-05-16 Richard Biener <rguenther@suse.de>
23324 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23326 (note_variable_value_in_expr): If we resolved the decl ref
23327 do not push to the stack.
23329 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23331 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23332 operations in fast-math mode.
23333 (vaddq_f16): Likewise.
23334 (vmul_f16): Likewise.
23335 (vmulq_f16): Likewise.
23336 (vsub_f16): Likewise.
23337 (vsubq_f16): Likewise.
23338 * config/arm/neon.md (add<mode>3): New.
23340 (fma:<VH:mode>3): New. Also remove outdated comment.
23343 2017-05-16 Martin Liska <mliska@suse.cz>
23347 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23348 (odr_types_equivalent_p): Likewise.
23350 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23352 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23354 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23357 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23358 non-interunit SSE move alternatives with '?'.
23359 (zero-extendsidi peephole2): New peephole to skip intermediate
23360 general register in SSE zero-extend sequence.
23362 2017-05-15 Jeff Law <law@redhat.com>
23364 * reorg.c (relax_delay_slots): Create a new variable to hold
23365 the temporary target rather than clobbering TARGET_LABEL.
23367 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23368 missing argument to extract_bit_field call.
23369 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23371 2017-05-15 Martin Liska <mliska@suse.cz>
23374 * gcc.c (process_command): Do not allow empty argument of -o option.
23376 2017-05-15 Renlin Li <renlin.li@arm.com>
23378 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23379 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23380 * config/aarch64/constraints.md (Usf): Add long call check.
23381 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23382 (call_value): Likewise.
23383 (sibcall): Likewise.
23384 (sibcall_value): Likewise.
23386 (call_value_insn): New.
23387 (sibcall_insn): Update rtx pattern.
23388 (sibcall_value_insn): Likewise.
23389 (call_internal): Remove.
23390 (call_value_internal): Likewise.
23391 (sibcall_internal): Likewise.
23392 (sibcall_value_internal): Likewise.
23393 (call_reg): Likewise.
23394 (call_symbol): Likewise.
23395 (call_value_reg): Likewise.
23396 (call_value_symbol): Likewise.
23398 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23401 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23403 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23405 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23406 compatible with CCGOCmode and with CCZmode.
23408 2017-05-14 Martin Sebor <msebor@redhat.com>
23410 PR middle-end/77671
23411 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23412 (gimple_fold_builtin_snprintf): Same.
23413 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23414 (gimple_fold_builtin_snprintf): Same.
23415 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23416 of character types.
23417 (is_call_safe): New function.
23418 (try_substitute_return_value): Call it.
23419 (try_simplify_call): New function.
23420 (pass_sprintf_length::handle_gimple_call): Call it.
23422 2017-05-14 Martin Sebor <msebor@redhat.com>
23424 PR middle-end/80669
23425 * builtins.c (expand_builtin_stpncpy): Simplify.
23427 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23429 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23430 * config/i386/i386.h
23431 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23432 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23433 (struct machine_function): Add new members call_ms2sysv,
23434 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23435 (struct machine_frame_state): New fields sp_realigned and
23436 sp_realigned_offset.
23437 * config/i386/i386.c
23438 (enum xlogue_stub): New enum.
23439 (enum xlogue_stub_sets): New enum.
23440 (class xlogue_layout): New class.
23441 (struct ix86_frame): New fields stack_realign_allocate_offset,
23442 stack_realign_offset and outlined_save_offset. Modify comments to
23443 detail stack layout when using out-of-line stubs.
23444 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23445 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23446 -mcall-ms2sysv-xlogues.
23447 (stub_managed_regs): New static variable.
23448 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23449 registers managed by out-of-line stub.
23450 (disable_call_ms2sysv_xlogues): New function.
23451 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23452 m->call_ms2sysv when appropriate and compute frame layout for
23454 (sp_valid_at, fp_valid_at): New inline functions.
23455 (choose_basereg): New function.
23456 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23458 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23459 Use align parameter of choose_baseaddr to generated aligned SSE movs
23461 (pro_epilogue_adjust_stack): Modify to track
23462 machine_frame_state::sp_realigned.
23463 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23464 (ix86_nsaved_sseregs): Likewise.
23465 (ix86_emit_save_regs): Likewise.
23466 (ix86_emit_save_regs_using_mov): Likewise.
23467 (ix86_emit_save_sse_regs_using_mov): Likewise.
23468 (get_scratch_register_on_entry): Likewise.
23469 (gen_frame_set): New function.
23470 (gen_frame_load): Likewise.
23471 (gen_frame_store): Likewise.
23472 (emit_outlined_ms2sysv_save): Likewise.
23473 (emit_outlined_ms2sysv_restore): Likewise.
23474 (ix86_expand_prologue): Modify stack re-alignment code and call
23475 emit_outlined_ms2sysv_save when appropriate.
23476 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23477 parameter rtx_insn *insn, which allows the function to be used to only
23478 generate the notes.
23479 (ix86_expand_epilogue): Modify validity checks of frame and stack
23480 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23481 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23482 * config/i386/predicates.md
23483 (save_multiple): New predicate.
23484 (restore_multiple): Likewise.
23485 * config/i386/sse.md
23486 (save_multiple<mode>): New pattern.
23487 (save_multiple_realign<mode>): Likewise.
23488 (restore_multiple<mode>): Likewise.
23489 (restore_multiple_and_return<mode>): Likewise.
23490 (restore_multiple_leave_return<mode>): Likewise.
23491 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23493 2017-05-14 Julia Koval <julia.koval@intel.com>
23495 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23496 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23497 (__builtin_ia32_xsetbv): New builtins.
23498 * config/i386/i386.c (ix86_expand_special_args_builtin):
23500 (ix86_expand_builtin): Special expand for new intrinsics.
23501 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23502 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23503 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23505 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23507 * cfganal.c (inverted_post_order_compute): Change argument type
23509 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23510 * df-core.c (rest_of_handle_df_initialize): Adjust.
23511 (rest_of_handle_df_finish): Likewise.
23512 (df_analyze_1): Likewise.
23513 (df_analyze): Likewise.
23514 (loop_inverted_post_order_compute): Change argument to be a vec *.
23515 (df_analyze_loop): Adjust.
23516 (df_get_n_blocks): Likewise.
23517 (df_get_postorder): Likewise.
23518 * df.h (struct df_d): Change field to be a vec.
23519 * lcm.c (compute_laterin): Adjust.
23520 (compute_available): Likewise.
23521 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23522 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23523 * tree-ssa-pre.c (compute_antic): Likewise.
23525 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23527 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23528 (depth_first_search::depth_first_search): Change structure init
23529 function to this constructor.
23530 (depth_first_search::add_bb): Rename function to this member.
23531 (depth_first_search::execute): Likewise.
23532 (flow_dfs_compute_reverse_finish): Adjust.
23534 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23536 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23537 (longest_simple_path): Likewise.
23538 * shrink-wrap.c (spread_components): Likewise.
23539 (disqualify_problematic_components): Likewise.
23540 (emit_common_heads_for_components): Likewise.
23541 (emit_common_tails_for_components): Likewise.
23542 (insert_prologue_epilogue_for_components): Likewise.
23544 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23546 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23549 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23551 * df-core.c (df_set_blocks): Start using auto_bitmap.
23552 (df_compact_blocks): Likewise.
23553 * df-problems.c (df_rd_confluence_n): Likewise.
23554 * df-scan.c (df_insn_rescan_all): Likewise.
23555 (df_process_deferred_rescans): Likewise.
23556 (df_update_entry_block_defs): Likewise.
23557 (df_update_exit_block_uses): Likewise.
23558 (df_entry_block_bitmap_verify): Likewise.
23559 (df_exit_block_bitmap_verify): Likewise.
23560 (df_scan_verify): Likewise.
23561 * lra-constraints.c (lra_constraints): Likewise.
23562 (undo_optional_reloads): Likewise.
23563 (lra_undo_inheritance): Likewise.
23564 * lra-remat.c (calculate_gen_cands): Likewise.
23565 (do_remat): Likewise.
23566 * lra-spills.c (assign_spill_hard_regs): Likewise.
23567 (spill_pseudos): Likewise.
23568 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23569 (bitmap_set_subtract_values): Likewise.
23571 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23573 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23574 management with auto_bitmap.
23575 (fix_inter_tick): Likewise.
23576 (fix_recovery_deps): Likewise.
23577 * ira.c (add_store_equivs): Likewise.
23578 (find_moveable_pseudos): Likewise.
23579 (split_live_ranges_for_shrink_wrap): Likewise.
23580 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23581 (rtx_reuse_manager::seen_def_p): Likewise.
23582 (rtx_reuse_manager::set_seen_def): Likewise.
23583 * print-rtl.h (class rtx_reuse_manager): Likewise.
23585 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23587 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23589 (migrate_btr_def): Likewise.
23590 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23591 * df-core.c (loop_post_order_compute): Likewise.
23592 (loop_inverted_post_order_compute): Likewise.
23593 * hsa-common.h: Likewise.
23594 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23595 * init-regs.c (initialize_uninitialized_regs): Likewise.
23596 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23597 (inline_small_functions): Likewise.
23598 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23599 * ira.c (combine_and_move_insns): Likewise.
23600 (build_insn_chain): Likewise.
23601 * loop-invariant.c (find_invariants): Likewise.
23602 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23603 * predict.c (tree_predict_by_opcode): Likewise.
23604 (predict_paths_leading_to): Likewise.
23605 (predict_paths_leading_to_edge): Likewise.
23606 (estimate_loops_at_level): Likewise.
23607 (estimate_loops): Likewise.
23608 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23609 (spread_components): Likewise.
23610 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23611 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23612 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23613 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23614 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23615 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23616 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23617 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23618 (mark_threaded_blocks): Likewise.
23619 (thread_through_all_blocks): Likewise.
23620 * tree-ssa.c (verify_ssa): Likewise.
23621 (execute_update_addresses_taken): Likewise.
23622 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23624 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23626 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23628 (post_order_compute): Likewise.
23629 (inverted_post_order_compute): Likewise.
23630 (pre_and_rev_post_order_compute_fn): Likewise.
23632 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23634 * genrecog.c (int_set::int_set): Explicitly construct our
23635 auto_vec base class.
23636 * vec.h (auto_vec::auto_vec): New constructor.
23638 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23640 * bitmap.h (class auto_bitmap): New constructor taking
23641 bitmap_obstack * argument.
23643 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23645 * bitmap.h (class auto_bitmap): Change type of m_bits to
23646 bitmap_head, and adjust ctor / dtor and member operators.
23648 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23650 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23651 when returned register mode doesn't match original mode.
23653 2017-05-12 Jeff Law <law@redhat.com>
23654 Jakub Jelinek <jakub@redhat.com>
23656 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23657 we look for cc setter after the compare-elim changes.
23658 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23659 within the vector to match what compare-elim now expects.
23660 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23661 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23663 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23664 after the compare-elim changes.
23665 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23666 the vector to match what compare-elim now expects.
23667 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23668 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23669 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23670 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23671 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23673 * config/visium/visium.c (single_set_and_flags): Fix where
23674 we look for cc setter after the compare-elim changes.
23675 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23676 with the vector to match what compare-elim now expects.
23677 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23678 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23679 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23680 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23681 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23682 (neg<mode>2_insn_set_overflow): Likewise.
23684 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23686 PR middle-end/79794
23687 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23688 maybe_expand_insn call, set ops[0].target. If still set after call,
23689 set alt_rtl. Add extra arg to recursive calls.
23690 (extract_bit_field): Add alt_rtl argument. Pass to
23692 * expmed.h (extract_bit_field): Fix prototype.
23693 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23694 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23695 to extract_bit_field_calls.
23696 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23697 Pass alt_rtl to extract_bit_field calls.
23698 * calls.c (store_unaligned_arguments_into_psuedos)
23699 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23700 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23702 * optabs.h (struct expand_operand): Add target bitfield.
23704 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23706 * compare-elim.c (try_eliminate_compare): Canonicalize
23707 operation with embedded compare to
23708 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23709 (set (reg) (operation)].
23711 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23713 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23716 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23717 cost of adding a carry flag for ADC instruction.
23718 [case MINUS]: Ignore the cost of subtracting a carry flag
23719 for SBB instruction.
23721 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23723 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23725 * config/rs6000/bmiintrin.h: New file.
23726 * config/rs6000/bmi2intrin.h: New file.
23727 * config/rs6000/x86intrin.h: New file.
23729 2017-05-12 Jeff Law <law@redhat.com>
23731 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23734 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23736 PR middle-end/80707
23737 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23739 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23741 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23742 early expansion of vector divide builtins.
23743 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23744 builtins identified as having unsigned arguments.
23746 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23748 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23749 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23750 expansion of vector logical operations (and, andc, or, xor,
23753 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23755 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23756 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23758 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23760 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23761 early GIMPLE expansion of vector multiplies.
23763 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23765 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23766 TARGET_HAVE_MOVT conditional.
23767 (movt splitter): Likewise.
23769 2017-05-12 Richard Biener <rguenther@suse.de>
23771 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23772 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23773 Fold all stmts not inplace.
23775 2017-05-12 Richard Biener <rguenther@suse.de>
23777 PR tree-optimization/80713
23778 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23779 inserted_exprs bit for not removed stmts.
23781 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23783 PR middle-end/69921
23784 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23785 parallelized" attribute for parallelized OpenACC kernels.
23786 * omp-offload.c (execute_oacc_device_lower): Use it.
23788 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23789 Set "oacc kernels" attribute.
23790 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23791 parameter. Adjust all users.
23792 (oacc_fn_attrib_kernels_p): Remove function.
23793 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23794 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23795 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23796 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23797 assert "oacc kernels" attribute is set.
23799 2017-05-11 Carl Love <cel@us.ibm.com>
23801 * config/rs6000/rs6000-c: Add support for built-in functions
23802 vector unsigned char vec_popcnt (vector signed char)
23803 vector unsigned char vec_popcnt (vector unsigned char)
23804 vector unsigned short vec_popcnt (vector signed short)
23805 vector unsigned short vec_popcnt (vector unsigned short)
23806 vector unsigned int vec_popcnt (vector signed int)
23807 vector unsigned int vec_popcnt (vector unsigned int)
23808 vector unsigned long long vec_popcnt (vector signed long long)
23809 vector unsigned long long vec_popcnt (vector unsigned long long)
23810 vector signed long long vec_slo (vector signed long long,
23811 vector signed char)
23812 vector signed long long vec_slo (vector signed long long,
23813 vector unsigned char)
23814 vector unsigned long long vec_slo (vector unsigned long long,
23815 vector signed char)
23816 vector unsigned long long vec_slo (vector unsigned long long,
23817 vector unsigned char)
23818 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23819 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23820 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23821 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23822 * doc/extend.texi: Update the built-in documentation file for the
23823 new built-in functions.
23825 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23827 * attribs.h (sorted_attr_string): Move machine independent
23828 functions for target clone support from the i386 port to common
23829 code. Rename ix86_function_versions to common_function_versions.
23830 Rename make_name to make_unique_name.
23831 (common_function_versions): Likewise.
23832 (make_unique_name): Likewise.
23833 (make_dispatcher_decl): Likewise.
23834 (is_function_default_version): Likewise.
23835 * attribs.c (attr_strcmp): Likewise.
23836 (sorted_attr_string): Likewise.
23837 (common_function_versions): Likewise.
23838 (make_unique_name): Likewise.
23839 (make_dispatcher_decl): Likewise.
23840 (is_function_default_version): Likewise.
23841 * config/i386/i386.c (attr_strcmp): Likewise.
23842 (sorted_attr_string): Likewise.
23843 (ix86_function_versions): Likewise.
23844 (make_name): Likewise.
23845 (make_dispatcher_decl): Likewise.
23846 (is_function_default_version): Likewise.
23847 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23849 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23852 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23853 Account for direct move costs for vec_construct of integer
23856 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23859 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23860 (UNSPEC_STX_ATOMIC): Ditto.
23861 (loaddi_via_sse): New insn.
23862 (storedi_via_sse): Ditto.
23863 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23864 Update corresponding peephole2 patterns.
23865 (atomic_storedi_fpu): Ditto.
23867 2017-05-11 Julia Koval <julia.koval@intel.com>
23869 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23870 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23872 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23873 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23874 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23876 2017-05-11 Nathan Sidwell <nathan@acm.org>
23878 * graphite-poly.c: Include dumpfile.h.
23880 * dumpfle.h (dump_function): Declare here ...
23881 * tree-dump.h (dump_function): ... not here.
23882 * dumpfile.c: #include tree-cfg.h.
23883 (dump_function): Move here from ...
23884 * tree-dump.c (dump_function): ... here.
23885 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23886 * graphite-poly.c: Don't include tree-dump.h.
23887 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23888 * print-tree.c: Likewise.
23889 * stor-layout.c: Likewise.
23890 * tree-nested.c: Likewise.
23892 * dumpfile.c (dump_start): Use TDF_FLAGS.
23893 (dump_enable_all): Fix TDF_KIND check thinko.
23895 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23897 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23898 array entries to represent two legal parameterizations of the
23899 overloaded __builtin_cmpb function, as represented by the
23900 P6_OV_BUILTIN_CMPB constant.
23901 (altivec_resolve_overloaded_builtin): Add special case handling
23902 for the __builtin_cmpb function, as represented by the
23903 P6_OV_BUILTIN_CMPB constant.
23904 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23905 (BU_P6_64BIT_2): New macro.
23906 (BU_P6_OVERLOAD_2): New macro
23907 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23908 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23909 (CMPB): Add overload support to represent both 32-bit and 64-bit
23910 compare-bytes function.
23911 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23912 support for TARGET_CMPB.
23913 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23914 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23915 documentation of the __builtin_cmpb overloaded built-in function.
23917 2017-05-11 Richard Biener <rguenther@suse.de>
23919 PR tree-optimization/80705
23920 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23921 bases are not vectorizable.
23923 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23925 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23926 when counting register pressure.
23928 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23930 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23931 (iv_ca_more_deps): Renamed to ...
23932 (iv_ca_compare_deps): ... this.
23933 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23935 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23937 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23939 (determine_group_iv_costs): ... here.
23940 (find_inv_vars_cb): Record inv var if it's not recorded before.
23942 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23944 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23945 (get_shiftadd_cost): Ditto.
23947 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23949 * tree-ssa-address.c: Include header file.
23950 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23952 (add_to_parts): Refactor.
23953 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23954 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23957 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23959 PR tree-optimization/53090
23960 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23962 (extract_cond_operands): Detect condition with IV on both sides
23963 and return COMP_IV_EXPR_2.
23964 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23965 (rewrite_use_compare): Simplify by removing call to function
23966 extract_cond_operands.
23968 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23970 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23971 (extract_cond_operands): Detect condition comparing against non-
23972 invariant bound and return appropriate enum value.
23973 (find_interesting_uses_cond): Update use of extract_cond_operands.
23974 Handle its return value accordingly.
23975 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23977 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23979 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23980 nonlinear iv_use computation in loop invariant sensitive way.
23982 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23984 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23985 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23987 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23989 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23990 (dump_cand): Support iv_cand.inv_exprs.
23991 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23993 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23996 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23998 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24000 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24001 as local function. Include necessary header files.
24002 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24004 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24006 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24008 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24010 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24011 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24012 RSHIFT_EXPR and BIT_NOT_EXPR.
24014 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24016 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24017 (adjust_setup_cost): New parameter supporting round up adjustment.
24018 (struct address_cost_data): Delete.
24019 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24020 (split_address_cost, ptr_difference_cost): Delete.
24021 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24022 (struct ainc_cost_data): New struct.
24023 (get_address_cost_ainc): New function.
24024 (get_address_cost, get_computation_cost): Reimplement.
24025 (determine_group_iv_cost_address): Record inv_expr for all uses of
24027 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24028 (iv_ca_has_deps): Reimplemented to ...
24029 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24031 (iv_ca_extend): Call iv_ca_more_deps.
24033 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24035 * tree-ssa-address.c (struct mem_address): Move to header file.
24036 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24037 * tree-ssa-address.h (struct mem_address): Move from C file.
24038 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24040 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24042 * tree-affine.h (aff_combination_type): New interface.
24043 (aff_combination_zero_p): Remove static.
24044 (aff_combination_const_p): New interface.
24045 (aff_combination_singleton_var_p): New interfaces.
24047 2017-05-11 Richard Biener <rguenther@suse.de>
24049 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24050 Skip unreachable blocks and destinations.
24051 (eliminate): Move stmt removal and fixup ...
24052 (fini_eliminate): ... here. Skip inserted exprs.
24053 (pass_pre::execute): Move fini_pre after fini_eliminate.
24054 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24055 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24056 PRE to get rid of dead code that has invalid SSA form and
24057 split critical edges again.
24059 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24061 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24063 2017-05-11 Richard Biener <rguenther@suse.de>
24065 * passes.c (execute_function_todo): Verify loops if they are
24066 said to be up-to-date.
24067 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24068 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24070 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24073 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24074 handle calling assemble_external ourself.
24077 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24078 modes with zero size. Enhance comment.
24080 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24082 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24083 built-ins for vec_xl and vec_xst with short and char pointer
24086 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24088 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24089 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24090 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24091 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24092 (_mm_maskz_min_round_ss): New intrinsics.
24093 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24094 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24095 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24096 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24097 (__builtin_ia32_minss_mask_round): New builtins.
24098 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24099 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24100 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24102 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24103 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24105 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24108 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24110 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24111 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24112 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24113 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24114 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24115 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24116 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24117 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24118 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24119 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24120 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24121 (__builtin_ia32_mulss_mask_round): New builtins.
24122 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24123 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24124 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24126 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24127 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24129 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24132 2017-05-10 Julia Koval <julia.koval@intel.com>
24134 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24135 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24136 (_mm256_setr_m128i): New intrinsics.
24138 2017-05-10 Julia Koval <julia.koval@intel.com>
24140 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24141 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24142 (_mm_maskz_rcp14_ss): New intrinsics.
24143 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24144 (__builtin_ia32_rcp14ss_mask): New builtins.
24145 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24147 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24149 PR tree-optimization/51513
24150 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24151 (assert_unreachable_fallthru_edge_p): Use it.
24152 (group_case_labels_stmt): Likewise.
24153 * tree-cfg.h: Prototype it.
24154 * stmt.c: Include cfghooks.h and tree-cfg.h.
24155 (emit_case_dispatch_table) <gap_label>: New local variable.
24156 Use it to fill dispatch table gaps.
24157 Test for default_label before updating probabilities.
24158 (expand_case) <default_label>: Remove unneeded initialization.
24159 Test for unreachable default case statement and remove its edge.
24160 Set default_label accordingly.
24161 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24163 2017-05-10 Carl Love <cel@us.ibm.com>
24165 * config/rs6000/rs6000-c: Add support for built-in functions
24166 vector signed char vec_neg (vector signed char)
24167 vector signed short int vec_neg (vector short int)
24168 vector signed int vec_neg (vector signed int)
24169 vector signed long long vec_neg (vector signed long long)
24170 vector float vec_neg (vector float)
24171 vector double vec_neg (vector double)
24172 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24174 * config/rs6000/altivec.h: Add define for vec_neg
24175 * doc/extend.texi: Update the built-in documentation for the
24176 new built-in functions.
24178 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24180 PR tree-optimization/77644
24181 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24183 2017-05-10 Nathan Sidwell <nathan@acm.org>
24185 * dumpfile.h (TDI_lang_all): New.
24186 (TDF_KIND): New. Renumber others
24187 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24189 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24191 (get_dump_file_name): Adjust suffix generation.
24192 (dump_enable_all): Use TDF_KIND.
24193 * doc/invoke.texi (-fdump-lang-all): Document.
24195 * dumpfile.h: Tabify.
24197 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24200 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24201 Move member access before delete.
24203 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24205 * tree-inline.c (expand_call_inline): Split block at stmt
24208 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24211 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24212 are now unused after splitting mov{sf,sd}_hardfloat.
24213 (f32_lr2): Likewise.
24214 (f32_lm): Likewise.
24215 (f32_lm2): Likewise.
24216 (f32_li): Likewise.
24217 (f32_li2): Likewise.
24218 (f32_lv): Likewise.
24219 (f32_sr): Likewise.
24220 (f32_sr2): Likewise.
24221 (f32_sm): Likewise.
24222 (f32_sm2): Likewise.
24223 (f32_si): Likewise.
24224 (f32_si2): Likewise.
24225 (f32_sv): Likewise.
24226 (f32_dm): Likewise.
24227 (f32_vsx): Likewise.
24228 (f32_av): Likewise.
24229 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24230 For movsf, order stores so the VSX stores occur before the GPR
24231 store which encourages the register allocator to use a traditional
24232 FPR instead of a GPR. For movsd, order the stores so that the GPR
24233 store comes before the VSX stores to allow the power6 to work.
24234 This is due to the power6 not having a 32-bit integer store
24235 instruction from a FPR.
24236 (movsf_hardfloat): Likewise.
24237 (movsd_hardfloat): Likewise.
24239 2017-05-09 Martin Sebor <msebor@redhat.com>
24241 PR translation/80280
24242 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24245 PR translation/80280
24246 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24247 data member added in r247778.
24248 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24250 2017-05-09 Nathan Sidwell <nathan@acm.org>
24252 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24254 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24257 2017-05-09 Marek Polacek <polacek@redhat.com>
24259 * doc/invoke.texi: Fix typo.
24261 2017-05-09 Richard Biener <rguenther@suse.de>
24263 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24264 (vrp_val_is_min): Likewise.
24265 (set_value_range_to_value): Likewise.
24266 (set_value_range_to_nonnegative): Likewise.
24267 (gimple_assign_nonzero_p): Likewise.
24268 (gimple_stmt_nonzero_p): Likewise.
24269 (vrp_int_const_binop): Likewise. Remove unreachable case.
24270 (adjust_range_with_scev): Adjust comments.
24271 (compare_range_with_value): Likewise.
24272 (extract_range_from_phi_node): Likewise.
24273 (test_for_singularity): Likewise.
24275 2017-05-09 Richard Biener <rguenther@suse.de>
24277 * tree-vrp.c (get_single_symbol): Add assert that we don't
24278 get overflowed constants as invariant part.
24279 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24280 checks. Use wi::cmp instead of recursing for integer constants.
24281 (compare_values): Just ignore whether we assumed undefined
24282 overflow instead of failing the compare.
24283 (extract_range_for_var_from_comparison_expr): Add comment before the
24284 TREE_NO_WARNING sets.
24285 (test_for_singularity): Likewise.
24286 (extract_range_from_comparison): Do not disable optimization
24287 when we assumed undefined overflow.
24288 (extract_range_basic): Remove init of unused var.
24290 2017-05-09 Richard Biener <rguenther@suse.de>
24292 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24293 (extract_range_from_multiplicative_op_1): Adjust.
24294 (extract_range_from_binary_expr_1): Use int_const_binop.
24296 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24299 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24300 rs6000_store_data_bypass_p in seven define_bypass directives and
24301 in several comments.
24302 * config/rs6000/rs6000-protos.h: Add prototype for
24303 rs6000_store_data_bypass_p function.
24304 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24305 function implements slightly different (rs6000-specific) semantics
24306 than store_data_bypass_p, returning false rather than aborting
24307 with assertion error when arguments do not satisfy the
24308 requirements of store data bypass.
24309 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24310 rs6000_store_data_bypass_p.
24312 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24314 * config/xtensa/xtensa-protos.h
24315 (xtensa_initial_elimination_offset): New declaration.
24316 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24317 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24318 macro definition, add case for FRAME_POINTER_REGNUM when
24319 FRAME_GROWS_DOWNWARD.
24320 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24321 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24322 xtensa_initial_elimination_offset.
24324 2017-05-08 Nathan Sidwell <nathan@acm.org>
24326 * doc/invoke.texi: Alphabetize -fdump options.
24328 2017-05-08 Martin Sebor <msebor@redhat.com>
24330 PR translation/80280
24331 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24333 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24335 * target.def (compute_frame_layout): New optional target hook.
24336 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24337 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24338 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24340 * reload1.c (verify_initial_elim_offsets): Likewise.
24341 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24342 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24343 (arm_get_frame_offsets): Split up into this ...
24344 (arm_compute_frame_layout): ... and this function.
24346 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24348 * config/aarch64/constraints.md (Usa): New constraint.
24349 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24351 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24353 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24354 with_multilib_list after it has been checked.
24356 2017-05-08 Richard Biener <rguenther@suse.de>
24358 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24359 (bitmap_set_subtract_values): Likewise.
24361 2017-05-08 Richard Biener <rguenther@suse.de>
24363 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24364 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24366 (gimple_stmt_nonzero_warnv_p): Rename to ...
24367 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24369 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24370 (extract_range_basic): Adjust, do not disable propagation on
24371 strict overflow sensitive simplification.
24372 (vrp_visit_cond_stmt): Likewise.
24374 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24376 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24377 body size unconditionally.
24379 2017-05-07 Jeff Law <law@redhat.com>
24382 2017-05-06 Jeff Law <law@redhat.com>
24383 PR tree-optimization/78496
24384 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24387 PR tree-optimization/78496
24388 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24389 (simplify_stmt_using_ranges): Call it.
24390 (vrp_dom_walker::before_dom_children): Extract equivalences
24391 from an ASSERT_EXPR with an equality comparison against a
24394 2017-05-06 Jeff Law <law@redhat.com>
24396 PR tree-optimization/78496
24397 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24400 PR tree-optimization/78496
24401 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24402 (simplify_stmt_using_ranges): Call it.
24403 (vrp_dom_walker::before_dom_children): Extract equivalences
24404 from an ASSERT_EXPR with an equality comparison against a
24407 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24409 * lra-constraints.c (lra_copy_reg_equiv): New function.
24410 (split_reg): Use it to copy equivalence information from the
24411 original register to the spill register.
24413 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24415 PR rtl-optimization/75964
24416 * simplify-rtx.c (simplify_const_relational_operation): Remove
24417 invalid handling of comparisons of integer ABS.
24419 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24421 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24422 initialize to zero.
24423 (init_regs): Remove declaration.
24424 (function_arg_advance_32): Initialize error_p as boolean variable.
24426 2017-05-05 Nathan Sidwell <nathan@acm.org>
24428 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24429 lines. Use for (;;).
24431 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24433 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24434 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24435 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24436 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24437 VF=2 that require versioning.
24439 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24441 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24444 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24446 * diagnostic.h (diagnostic_override_option_index): Convert from
24447 macro to inline function.
24449 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24451 * diagnostic.c (last_module_changed_p): New function.
24452 (set_last_module): New function.
24453 (diagnostic_report_current_module): Convert macro usage to
24454 the above functions.
24455 * diagnostic.h (diagnostic_context::last_module): Strengthen
24456 from const line_map * to const line_map_ordinary *.
24457 (diagnostic_last_module_changed): Delete macro.
24458 (diagnostic_set_last_module): Delete macro.
24460 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24462 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24463 with diagnostic_report_diagnostic.
24464 (diagnostic_n_impl_richloc): Likewise.
24465 * diagnostic.h (report_diagnostic): Delete macro.
24466 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24467 with diagnostic_report_diagnostic.
24468 * substring-locations.c (format_warning_va): Likewise.
24470 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24472 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24473 save/restor of format_spec. Move option-printing code to...
24474 (print_option_information): ...this new function, and
24475 reimplement by simply printing to the pretty_printer,
24476 rather than appending to the format string.
24478 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24480 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24481 handling logic into...
24482 (update_effective_level_from_pragmas): ...this new function.
24484 2017-05-04 Andrew Waterman <andrew@sifive.com>
24486 * config/riscv/riscv.opt (mstrict-align): New option.
24487 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24488 (SLOW_UNALIGNED_ACCESS): Define.
24489 (riscv_slow_unaligned_access): Declare.
24490 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24492 (riscv_slow_unaligned_access): New variable.
24493 (rocket_tune_info): Set slow_unaligned_access to true.
24494 (optimize_size_tune_info): Set slow_unaligned_access to false.
24495 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24496 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24497 (riscv_option_override): Set riscv_slow_unaligned_access.
24498 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24500 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24502 * config/riscv/riscv.md: Unify indentation.
24504 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24509 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24511 (extendsi<mode>2): Add support for doing sign extension via
24512 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24513 don't have ISA 3.0 instructions.
24514 (extendsi<mode>2 splitter): Likewise.
24515 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24516 generate the normal insns since SImode can now go in vector
24517 registers. Disallow the special UNSPECs needed for previous
24518 machines to hide SImode being used. Add new insns
24519 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24520 (fix_trunc<mode>si2_stfiwx): Likewise.
24521 (fix_trunc<mode>si2_internal): Likewise.
24522 (fixuns_trunc<mode>si2): Likewise.
24523 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24524 (fctiw<u>z_<mode>_smallint): Likewise.
24525 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24526 of floating point to 32-bit integer from doing a direct move to
24527 the GPR registers to do a store.
24528 (fctiwz_<mode>): Break long line.
24530 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24532 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24533 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24534 (addr_list, addr_offset_valid_p): New.
24535 (split_address_groups): Check offset validity with above function.
24536 (gt-tree-ssa-loop-ivopts.h): Include header file.
24538 2017-05-05 Nathan Sidwell <nathan@acm.org>
24540 * config.gcc (arm*-*-*): Add missing 'fi'.
24542 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24544 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24545 included in -fopt-info does not matter.
24546 * doc/optinfo.texi (-fopt-info): Fix description of default
24547 behavour. Explicitly say order of options included in -fopt-info
24550 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24552 * config.gcc: Allow combinations of aprofile and rmprofile values for
24553 --with-multilib-list.
24554 * config/arm/t-multilib: New file.
24555 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24556 variables. Remove setting of ISA and floating-point ABI in
24557 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24558 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24559 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24560 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24562 * config/arm/t-rmprofile: Likewise except for the matches changes.
24563 * doc/install.texi (--with-multilib-list): Document the combination of
24564 aprofile and rmprofile values and warn about pitfalls in doing that.
24566 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24568 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24569 (movdi_aarch64): Likewise.
24571 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24573 PR tree-optimization/80632
24574 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24576 (build_arrays): Initialize it for virtual phis.
24577 (fix_phi_nodes): Use it for virtual phis.
24579 PR tree-optimization/80558
24580 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24581 [x, y] op z into [x op, y op z] for op & or | if conditions
24584 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24585 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24588 * config/arm/arm.md (use_literal_pool): Remove.
24589 (64-bit immediate split): No longer takes cost into consideration
24590 if arm_disable_literal_pool is enabled.
24591 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24592 used when arm_disable_literal_pool is enabled.
24593 (arm_max_const_double_inline_cost): Remove use of
24594 arm_disable_literal_pool.
24595 (push_minipool_fix): Add assert.
24596 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24597 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24598 (no_literal_pool_sf_immediate): New.
24600 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24602 PR tree-optimization/80613
24603 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24604 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24606 2017-05-05 Richard Biener <rguenther@suse.de>
24608 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24610 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24612 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24613 of this flag from insn conditions due to removal from r247495.
24615 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24617 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24619 (arm_early_store_addr_dep_ptr): Likewise.
24620 * config/arm/aarch-common-protos.h
24621 (arm_early_load_addr_dep_ptr): Add prototype.
24622 (arm_early_store_addr_dep_ptr): Likewise.
24623 * config/arm/cortex-a53.md: Add new bypasses.
24625 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24627 * tree.c (next_type_uid): Change type to unsigned.
24628 (type_hash_canon): Decrement back next_type_uid if
24629 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24630 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24633 2017-05-04 Martin Sebor <msebor@redhat.com>
24635 * builtins.c: Fix a trivial typo in a comment.
24637 PR middle-end/79234
24638 * builtins.c (check_sizes): Adjust to handle reading past the end.
24639 Avoid printing excessive upper bound of ranges. Use %E to print
24640 tree nodes instead of converting them to %wu.
24641 (expand_builtin_memchr): New function.
24642 (compute_dest_size): Rename...
24643 (compute_objsize): ...to this.
24644 (expand_builtin_memcpy): Adjust.
24645 (expand_builtin_mempcpy): Adjust.
24646 (expand_builtin_strcat): Adjust.
24647 (expand_builtin_strcpy): Adjust.
24648 (check_strncat_sizes): Adjust.
24649 (expand_builtin_strncat): Adjust.
24650 (expand_builtin_strncpy): Adjust and simplify.
24651 (expand_builtin_memset): Adjust.
24652 (expand_builtin_bzero): Adjust.
24653 (expand_builtin_memcmp): Adjust.
24654 (expand_builtin): Handle memcmp.
24655 (maybe_emit_chk_warning): Check strncat just once.
24657 2017-05-04 Martin Sebor <msebor@redhat.com>
24659 PR preprocessor/79214
24660 PR middle-end/79222
24661 PR middle-end/79223
24662 * builtins.c (check_sizes): Add inlining context and issue
24663 warnings even when -Wno-system-headers is set.
24664 (check_strncat_sizes): Same.
24665 (expand_builtin_strncat): Same.
24666 (expand_builtin_memmove): New function.
24667 (expand_builtin_stpncpy): Same.
24668 (expand_builtin): Handle memmove and stpncpy.
24670 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24672 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24673 which is not used any more.
24675 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24677 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24679 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24681 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24682 (cortexa53_tunings): Likewise.
24683 (cortexa57_tunings): Likewise.
24684 (cortexa72_tunings): Likewise.
24685 (cortexa73_tunings): Likewise.
24687 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24689 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24690 Set loop alignment to 8.
24692 2017-05-04 Martin Sebor <msebor@redhat.com>
24694 PR translation/80280
24695 * builtins.c (expand_builtin_object_size): Add missing quoting to
24696 %D and like directives.
24697 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24698 (hsa_type_for_tree_type): Same.
24699 (verify_function_arguments): Same.
24700 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24701 * varasm.c (get_section): Same.
24704 2017-05-04 Martin Sebor <msebor@redhat.com>
24706 PR translation/80280
24707 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24709 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24711 * config/aarch64/aarch64.c (generic_addrcost_table):
24712 Change HI/TI mode setting.
24714 2017-05-04 Martin Jambor <mjambor@suse.cz>
24716 PR tree-optimization/80622
24717 * tree-sra.c (comes_initialized_p): New function.
24718 (build_accesses_from_assign): Only set write lazily when
24719 comes_initialized_p is false.
24720 (analyze_access_subtree): Use comes_initialized_p.
24721 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24722 instead of testing for PARM_DECL.
24724 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24726 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24727 constraint on operand 0 to allow more general addressing modes.
24728 Adjust output template.
24729 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24731 * config/aarch64/aarch64-protos.h
24732 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24733 * config/aarch64/constraints.md (Dp): New address constraint.
24734 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24737 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24739 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24740 update use of estimate_ipcp_clone_size_and_time.
24741 (estimate_local_effects): Update use of
24742 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24743 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24744 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24745 Return nonspecialized time.
24747 2017-05-04 Richard Biener <rguenther@suse.de>
24749 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24750 for the last VUSE which def dominates the PHI. Directly call
24752 (get_continuation_for_phi_1): Remove.
24754 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24756 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24757 to explain the use of truncating division. Cap the number of
24758 iterations to the maximum given by nb_iterations_upper_bound,
24761 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24763 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24764 * configure: Regenerate.
24765 * config.in: Regenerate.
24766 * config/i386/driver-mingw32.c: new file.
24767 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24768 * config.host: Link driver-mingw32.o on MinGW host.
24769 * doc/install.texi: Document new --enable-mingw-wildcard configure
24772 2017-05-04 Marek Polacek <polacek@redhat.com>
24774 PR tree-optimization/80612
24775 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24777 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24778 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24780 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24781 (movt splitter): Likewise.
24782 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24783 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24784 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24785 block for Thumb-1 with MOVT.
24786 (thumb2_legitimate_address_p): Move code block ...
24787 (can_avoid_literal_pool_for_label_p): ... into this new function.
24788 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24790 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24791 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24792 "M-profile targets with the MOVT instruction".
24794 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24796 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24797 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24798 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24800 2017-05-04 Martin Liska <mliska@suse.cz>
24802 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24803 variable cond_code.
24805 2017-05-04 Richard Biener <rguenther@suse.de>
24807 * tree.c (array_at_struct_end_p): Handle arrays at struct
24808 end with flexarrays more conservatively. Refactor and treat
24809 arrays of arrays or aggregates more strict. Fix
24810 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24811 * tree.c (array_at_struct_end_p): Adjust prototype.
24812 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24813 * gimple-fold.c (get_range_strlen): Likewise.
24814 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24816 2017-05-04 Richard Biener <rguenther@suse.de>
24818 PR tree-optimization/31130
24819 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24821 (supports_overflow_infinity): Likewise.
24822 (is_negative_overflow_infinity): Likewise.
24823 (is_positive_overflow_infinity): Likewise.
24824 (is_overflow_infinity): Likewise.
24825 (stmt_overflow_infinity): Likewise.
24826 (overflow_infinity_range_p): Likewise.
24827 (usable_range_p): Remove as always returning true.
24828 (make_overflow_infinity): Remove.
24829 (negative_overflow_infinity): Likewise.
24830 (positive_overflow_infinity): Likewise.
24831 (avoid_overflow_infinity): Likewise.
24832 (set_value_range): Adjust accordingly.
24833 (set_value_range_to_nonnegative): Likewise, remove now unused
24834 overflow_infinity arg.
24835 (vrp_operand_equal_p): Adjust.
24836 (update_value_range): Likewise.
24837 (range_int_cst_singleton_p): Likewise.
24838 (operand_less_p): Likewise.
24839 (compare_values_warnv): Likewise.
24840 (extract_range_for_var_from_comparison_expr): Likewise.
24841 (vrp_int_const_binop): Likewise.
24842 (zero_nonzero_bits_from_vr): Likewise.
24843 (extract_range_from_multiplicative_op_1): Likewise.
24844 (extract_range_from_binary_expr_1): Likewise.
24845 (extract_range_from_unary_expr): Likewise.
24846 (extract_range_from_comparison): Likewise.
24847 (extract_range_basic): Likewise.
24848 (adjust_range_with_scev): Likewise.
24849 (compare_ranges): Likewise.
24850 (compare_range_with_value): Likewise.
24851 (dump_value_range): Likewise.
24852 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24854 (simplify_cond_using_ranges): Adjust.
24856 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24858 * brig-builtins.def: Added a builtin for class_f64.
24859 * builtin-types.def: Added a builtin type needed by class_f64.
24861 2017-05-03 Jason Merrill <jason@redhat.com>
24863 * timevar.def: Add TV_CONSTEXPR.
24865 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24867 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24869 2017-05-03 Martin Jambor <mjambor@suse.cz>
24871 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24872 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24873 * ipa-cp.c (ipcp_propagate_stage): Do not call
24874 ipa_update_after_lto_read.
24875 * ipa-inline.c (ipa_inline): Likewise.
24877 2017-05-03 Martin Jambor <mjambor@suse.cz>
24879 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24880 tag. Added a default constructor and a destructor.
24881 (ipa_edge_args_sum_t): New class;
24882 (ipa_edge_args_sum): Declare.
24883 (ipa_edge_args_vector): Remove declaration.
24884 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24885 (ipa_free_edge_args_substructures): Remove declaration.
24886 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24887 (ipa_edge_args_info_available_for_edge_p): Likewise.
24888 * ipa-prop.c (ipa_edge_args_vector): Removed.
24889 (edge_removal_hook_holder): Likewise.
24890 (edge_duplication_hook_holder): Likewise.
24891 (ipa_edge_args_sum): New variable.
24892 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24893 ipa_edge_args_vector.
24894 (ipa_free_edge_args_substructures): Likewise.
24895 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24896 ipa_edge_args_vector.
24897 (ipa_edge_removal_hook): Turned into method
24898 ipa_edge_args_sum_t::remove.
24899 (ipa_edge_duplication_hook): Turned into method
24900 ipa_edge_args_sum_t::duplicate.
24901 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24902 registering edge hooks.
24903 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24904 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24905 ipa_edge_args_sum instead of ipa_edge_args_vector.
24906 * ipa-profile.c (ipa_profile): Likewise.
24908 2017-05-03 Martin Jambor <mjambor@suse.cz>
24910 * symbol-summary.h (function_summary): New method exists.
24911 (function_summary::symtab_removal): Deallocate through release.
24912 (call_summary): New class.
24913 (gt_ggc_mx): New overload.
24914 (gt_pch_nx): Likewise.
24915 (gt_pch_nx): Likewise.
24917 2017-05-03 Jeff Law <law@redhat.com>
24919 PR tree-optimization/78496
24920 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24921 from simplify_cond_using_ranges. Split off code to walk
24922 backwards through casts into ...
24923 (simplify_cond_using_ranges_2): New function.
24924 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24925 (execute_vrp): After identifying jump threads, call
24926 simplify_cond_using_ranges_2.
24928 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24931 * ipa-inline.h (inline_summary): Add ctor.
24932 (create_ggc): Do not use ggc_cleared_alloc.
24934 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24935 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24937 * gcc.c (handle_braces): Support escaping in switch matching
24939 * doc/invoke.texi (Spec Files): Document it.
24940 Remove superfluous @code markup in items.
24942 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24944 * diagnostic-show-locus.c (struct column_range): New struct.
24945 (get_affected_columns): New function.
24946 (get_printed_columns): New function.
24947 (struct correction): New struct.
24948 (correction::ensure_capacity): New function.
24949 (correction::ensure_terminated): New function.
24950 (struct line_corrections): New struct.
24951 (line_corrections::~line_corrections): New dtor.
24952 (line_corrections::add_hint): New function.
24953 (layout::print_trailing_fixits): Reimplement in terms of the new
24955 (selftest::test_overlapped_fixit_printing): New function.
24956 (selftest::diagnostic_show_locus_c_tests): Call it.
24958 2017-05-03 Nathan Sidwell <nathan@acm.org>
24960 Canonicalize canonical type hashing
24961 * tree.h (type_hash_canon_hash): Declare.
24962 * tree.c (type_hash_list, attribute_hash_list): Move into
24963 type_hash_canon_hash.
24964 (build_type_attribute_qual_variant): Break out hash code calc into
24965 type_hash_canon_hash.
24966 (type_hash_canon_hash): New. Generic type hash computation.
24967 (build_range_type_1, build_array_type_1, build_function_type,
24968 build_method_type_directly, build_offset_type, build_complex_type,
24969 make_vector_type): Call it.
24971 2017-05-03 Richard Biener <rguenther@suse.de>
24973 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24974 When all DRs have unknown misaligned do not always peel
24975 when there is a store but apply the same costing model as if
24976 there were only loads.
24978 2017-05-03 Richard Biener <rguenther@suse.de>
24981 PR tree-optimization/80492
24982 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24983 compare_base_decls returning dont-know properly.
24985 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24987 * config/arm/iterators.md (CCSI): New mode iterator.
24988 (arch): New mode attribute.
24989 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24990 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24991 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24992 code iterator for success result mode.
24993 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24994 the corresponding new insn generators.
24996 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24999 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25000 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25002 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25004 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25005 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25006 (DDR_A): Wrap DDR argument in brackets.
25007 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25008 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25009 (DDR_REVERSED_P): Likewise.
25011 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25013 PR tree-optimization/79472
25014 * tree-switch-conversion.c (struct switch_conv_info): Add
25015 contiguous_range and default_case_nonstandard fields.
25016 (collect_switch_conv_info): Compute contiguous_range and
25017 default_case_nonstandard fields, don't clear final_bb if
25018 contiguous_range and only the default case doesn't have the required
25020 (check_all_empty_except_final): Set default_case_nonstandard instead
25021 of failing if contiguous_range and the default case doesn't have empty
25023 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25024 and only the default case doesn't have the required constants. Skip
25026 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25027 if default_case_nonstandard.
25028 (build_constructors): Build constant 1 just once. Assert that default
25029 values aren't inserted in between cases if contiguous_range. Skip
25031 (build_arrays): Skip virtual phis.
25032 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25033 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25034 Handle virtual phis.
25035 (gen_inbound_check): Handle default_case_nonstandard case.
25036 (process_switch): Adjust check_final_bb caller. Call
25037 gather_default_values with the first non-default case instead of
25038 default case if default_case_nonstandard.
25040 2017-05-02 Nathan Sidwell <nathan@acm.org>
25042 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25043 check. Fix formatting.
25045 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25047 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25048 errors when comparing specialized and unspecialized times.
25050 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25052 * diagnostic-show-locus.c
25053 (layout::should_print_annotation_line_p): Make private.
25054 (layout::print_annotation_line): Make private.
25055 (layout::annotation_line_showed_range_p): Make private.
25056 (layout::show_ruler): Make private.
25057 (layout::print_source_line): Make private. Pass in line and
25058 line_width, rather than calling location_get_source_line. Drop
25060 (layout::print_leading_fixits): New method.
25061 (layout::print_any_fixits): Rename to...
25062 (layout::print_trailing_fixits): ...this, and make private.
25063 Don't print newline fixits.
25064 (diagnostic_show_locus): Move logic for printing one row into...
25065 (layout::print_line): ...this new function. Move the
25066 location_get_source_line call and error-handling from
25067 print_source_line to here. Call print_leading_fixits, and rename
25068 print_any_fixits to print_trailing_fixits.
25069 (selftest::test_fixit_insert_containing_newline): Update now that
25070 newlines are partially supported.
25071 (selftest::test_fixit_insert_containing_newline_2): New test.
25072 (selftest::test_fixit_replace_containing_newline): Update comments.
25073 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25074 * edit-context.c (class added_line): New class.
25075 (class edited_line): Describe newline handling in comment.
25076 (edited_line::actually_edited_p): New method.
25077 (edited_line::print_content): Delete redundant decl.
25078 (edited_line::m_predecessors): New field.
25079 (edited_file::print_content): Call edited_line::print_content.
25080 (edited_file::print_diff): Update to support newlines.
25081 (edited_file::print_diff_hunk): Likewise.
25082 (edited_file::print_run_of_changed_lines): New function.
25083 (edited_file::print_diff_line): Convert to...
25084 (print_diff_line): ...this.
25085 (edited_file::get_effective_line_count): New function.
25086 (edited_line::edited_line): Initialize new field m_predecessors.
25087 (edited_line::~edited_line): Clean up m_predecessors.
25088 (edited_line::apply_fixit): Handle newlines.
25089 (edited_line::get_effective_line_count): New function.
25090 (edited_line::print_content): New function.
25091 (edited_line::print_diff_lines): New function.
25092 (selftest::test_applying_fixits_insert_containing_newline): New
25094 (selftest::test_applying_fixits_replace_containing_newline): New
25096 (selftest::insert_line): New function.
25097 (selftest::test_applying_fixits_multiple_lines): Add example of
25099 (selftest::edit_context_c_tests): Call the new tests.
25101 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25103 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25104 parameter cand. Update dump information.
25105 (get_computation_cost): Update uses.
25107 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25109 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25110 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25111 (get_computation_at, rewrite_use_address): Update use of
25112 get_computation_aff.
25114 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25116 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25117 (get_computation): Delete.
25118 (get_computation_cost): Implement like get_computation_cost_at.
25119 Use get_computation_at.
25120 (get_computation_cost_at): Delete.
25121 (rewrite_use_nonlinear_expr): Use get_computation_at.
25122 (rewrite_use_compare, remove_unused_ivs): Ditto.
25124 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25126 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25128 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25130 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25131 (ivopts_global_cost_for_size): Rename parameter and update uses.
25132 (iv_ca_recount_cost): Update uses.
25133 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25134 candidates seperately in n_invs and n_cands.
25135 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25137 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25139 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25140 (find_inv_vars_cb): New.
25141 (find_depends): Renamed to ...
25142 (find_inv_vars): ... this.
25143 (add_candidate_1, force_var_cost): Call find_inv_vars.
25144 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25146 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25148 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25149 inv_vars. Add inv_exprs.
25150 (struct iv_cand): Rename depends_on to inv_vars.
25151 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25152 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25153 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25154 (dump_cand): Dump inv_vars.
25155 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25156 (record_invariant, find_depends, add_candidate_1): Ditto.
25157 (set_group_iv_cost, force_var_cost): Ditto.
25158 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25159 (get_computation_cost_at, get_computation_cost): Ditto.
25160 (determine_group_iv_cost_generic): Ditto.
25161 (determine_group_iv_cost_address): Ditto.
25162 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25163 (determine_group_iv_costs): Ditto.
25164 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25165 (iv_ca_set_remove_invariants): Renamed to ...
25166 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25167 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25168 (iv_ca_set_add_invariants): Renamed to ...
25169 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25170 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25171 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25172 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25173 (create_new_ivs): Remove useless dump.
25175 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25177 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25179 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25180 (iv_ca_set_no_cp, create_new_iv): Ditto.
25182 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25184 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25186 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25188 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25189 function tree_check2.
25191 2017-05-02 Martin Liska <mliska@suse.cz>
25193 * doc/gcov.texi: Add missing preposition.
25194 * gcov.c (function_info::function_info): Properly fill up
25195 all member variables.
25197 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25199 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25201 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25203 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25205 2017-05-02 Martin Liska <mliska@suse.cz>
25208 * lto-streamer-in.c (lto_read_tree_1): Remove
25209 LTO_STREAMER_DEBUG.
25210 * lto-streamer.c (struct tree_hash_entry): Likewise.
25211 (struct tree_entry_hasher): Likewise.
25212 (tree_entry_hasher::hash): Likewise.
25213 (tree_entry_hasher::equal): Likewise.
25214 (lto_streamer_init): Likewise.
25215 (lto_orig_address_map): Likewise.
25216 (lto_orig_address_get): Likewise.
25217 (lto_orig_address_remove): Likewise.
25218 * lto-streamer.h: Likewise.
25219 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25220 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25222 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25224 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25225 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25226 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25227 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25228 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25229 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25230 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25231 (mm_maskz_sub_ss): New intrinsics.
25232 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25233 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25234 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25235 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25236 (__builtin_ia32_subss_mask_round): New builtins.
25237 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25238 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25239 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25241 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25242 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25244 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25247 2017-05-02 Martin Jambor <mjambor@suse.cz>
25249 PR tree-optimization/78687
25250 * tree-sra.c (access): New field parent.
25251 (process_subtree_disqualification): New function.
25252 (disqualify_candidate): Call it.
25253 (build_accesses_from_assign): Reset write flag if creating an
25255 (build_access_subtree): Fill in parent field and also prpagate
25256 down grp_write flag.
25257 (create_artificial_child_access): New parameter set_grp_write, set
25258 grp_write to its value.
25259 (propagate_subaccesses_across_link): Also propagate grp_write flag
25261 (propagate_all_subaccesses): Push the closest parent back to work
25262 queue if add_access_to_work_queue returned true.
25264 2017-05-02 Richard Biener <rguenther@suse.de>
25266 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25267 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25268 -fstrict-overflow documentation.
25269 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25270 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25271 flag_strict_overflow.
25272 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25273 * lto-opts.c (lto_write_options): Do not stream it.
25274 * lto-wrapper.c (merge_and_complain): Do not handle it.
25275 * opts.c (default_options_table): Do not set -fstrict-overflow.
25276 (finish_options): Likewise do not clear it when sanitizing.
25277 * simplify-rtx.c (simplify_const_relational_operation): Do not
25278 test flag_strict_overflow.
25280 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25282 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25283 using enabled attribute.
25284 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25285 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25286 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25287 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25288 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25289 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25290 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25291 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25292 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25293 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25295 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25297 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25299 2017-05-02 Richard Biener <rguenther@suse.de>
25301 PR tree-optimization/80591
25303 2017-04-10 Richard Biener <rguenther@suse.de>
25305 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25308 2017-05-02 Richard Biener <rguenther@suse.de>
25310 PR tree-optimization/80549
25311 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25312 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25313 headers if they do not have a preheader.
25315 2017-05-02 Martin Liska <mliska@suse.cz>
25318 * common.opt: Fix typo.
25319 * doc/invoke.texi: Likewise.
25321 2017-05-01 Jan Beulich <jbeulich@suse.com>
25323 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25324 swapping, add (x,x,m,x,n) alternative.
25326 2017-05-01 Nathan Sidwell <nathan@acm.org>
25328 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25329 unnecessary unadjusted_alignment check.
25331 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25334 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25336 * gimplify.c (gimplify_cilk_detach): New function.
25337 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25338 * tree-core.h: Document EXPR_CILK_SPAWN.
25339 * tree.h (EXPR_CILK_SPAWN): Define.
25341 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25343 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25344 to use new fixit_hint representation, using the "replace" logic.
25345 (get_line_span_for_fixit_hint): Likewise.
25346 (layout::print_any_fixits): Likewise.
25347 (selftest::test_one_liner_many_fixits): Rename to...
25348 (selftest::test_one_liner_many_fixits_1): ...this, and update
25349 comment and expected output to reflect that the multiple fix-it
25350 hints are now consolidated into one insertion.
25351 (selftest::test_one_liner_many_fixits_2): New test.
25352 (selftest::test_diagnostic_show_locus_one_liner): Update for
25354 (selftest::test_fixit_consolidation): Update for fix-it API
25356 * diagnostic.c (print_parseable_fixits): Likewise.
25357 * edit-context.c (edited_line::m_line_events): Convert from
25358 auto_vec <line_event *> to auto_vec <line_event>.
25359 (class line_event): Convert from abstract base class to a concrete
25360 class, taking over the role of replace_event.
25361 (class insert_event): Delete.
25362 (class replace_event): Rename to class line_event. Convert to
25364 (edit_context::add_fixits): Reimplement.
25365 (edit_context::apply_insert): Delete.
25366 (edit_context::apply_replace): Rename to...
25367 (edit_context::apply_fixit): ...this. Convert to half-open range.
25368 (edited_file::apply_insert): Delete.
25369 (edited_file::apply_replace): Rename to...
25370 (edited_file::apply_fixit): ...this.
25371 (edited_line::~edited_line): Drop deletion of events.
25372 (edited_line::apply_insert): Delete.
25373 (edited_line::apply_replace): Rename to...
25374 (edited_line::apply_fixit): ...this. Convert to half-open range.
25375 Update for change to type of m_line_events.
25376 * edit-context.h (edit_context::apply_insert): Delete.
25377 (edit_context::apply_replace): Rename to...
25378 (edit_context::apply_fixit): ...this.
25380 2017-05-01 Martin Sebor <msebor@redhat.com>
25382 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25385 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25388 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25389 __get_cpuid_max returns 0.
25390 (__get_cpuid_count): Ditto.
25392 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25394 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25395 replacement expression is another instance of one of its arguments.
25397 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25400 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25401 check for stack push/pop autoinc.
25402 * config/i386/i386.c (ix86_agi_dependent): Return false
25403 if the only reason why modified_in_p returned true is that
25404 addr is SP based and set_insn is a push or pop.
25406 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25408 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25411 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25414 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25415 (account_size_time): Use two predicates - exec_pred and
25417 (evaluate_conditions_for_known_args): Compute both clause and
25419 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25420 (inline_summary_t::duplicate): Update.
25421 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25423 (compute_inline_parameters): Likewise.
25424 (estimate_edge_size_and_time): Update caluclation of time.
25425 (estimate_node_size_and_time): Compute both time and nonspecialized
25427 (estimate_ipcp_clone_size_and_time): Update.
25428 (inline_merge_summary): Update.
25429 (do_estimate_edge_time): Update.
25430 (do_estimate_edge_size): Update.
25431 (do_estimate_edge_hints): Update.
25432 (inline_read_section, inline_write_summary): Stream both new predicates.
25433 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25435 (compute_inlined_call_time): Cleanup.
25436 (big_speedup_p): Update.
25437 (edge_badness): Update.
25438 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25439 (size_time_entry): Replace predicate by exec_predicate and
25440 nonconst_predicate.
25441 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25442 (estimate_edge_time): Return also nonspec_time.
25443 (reset_edge_growth_cache): Update.
25445 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25447 PR rtl-optimization/80491
25448 * ifcvt.c (noce_process_if_block): When looking for x setter
25449 with missing else_bb, don't check only the insn right before
25450 cond_earliest, but look for the last insn that x is modified in
25451 within the same bb.
25453 PR rtl-optimization/80491
25454 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25456 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25458 PR tree-optimization/80487
25459 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25461 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25463 PR tree-optimization/79697
25464 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25465 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25466 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25468 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25469 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25471 2017-04-28 Martin Sebor <msebor@redhat.com>
25473 PR tree-optimization/80523
25474 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25475 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25477 (maybe_warn, format_directive, parse_directive): Use new functions.
25478 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25480 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25482 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25484 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25486 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25487 target_header_dir): Set correctly.
25488 * configure: Regenerated.
25489 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25490 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25491 instead of SYSTEM_HEADER_DIR.
25493 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25495 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25496 (estimate_local_effects): Likewise.
25497 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25498 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25499 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25500 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25501 do_estimate_edge_time, estimate_edge_time): Likewise.
25502 * ipa-inline-analysis.c (estimate_node_size_and_time,
25503 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25504 (estimate_time_after_inlining): Remove.
25506 2017-04-28 Martin Liska <mliska@suse.cz>
25508 * doc/gcov.texi: Enhance documentation of gcov.
25510 2017-04-28 Martin Liska <mliska@suse.cz>
25512 * doc/gcov.texi: Sort options in alphabetic order.
25513 * doc/gcov-dump.texi: Likewise.
25514 * doc/gcov-tool.texi: Likewise.
25515 * gcov.c (print_usage): Likewise.
25516 * gcov-dump.c (print_usage): Likewise.
25517 * gcov-tool.c (print_merge_usage_message): Likewise.
25518 (print_rewrite_usage_message): Likewise.
25519 (print_overlap_usage_message): Likewise.
25521 2017-04-28 Martin Liska <mliska@suse.cz>
25523 PR gcov-profile/53915
25524 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25526 2017-04-28 Martin Liska <mliska@suse.cz>
25528 PR gcov-profile/79891
25529 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25530 is marked by compiler as living on a line.
25531 (get_cycles_count): Remove usage of the union.
25532 (output_intermediate_file): Likewise.
25533 (find_source): Fix GNU coding style.
25534 (accumulate_line_counts): Remove old non-all block mode.
25535 (output_lines): Remove usage of the union.
25536 * profile.c (output_location): Include all BBs, even if
25537 belonging to a same line (and file) as a previous BB.
25539 2017-04-28 Martin Liska <mliska@suse.cz>
25541 * gcov.c (process_args): Handle new argument 'w'.
25542 (read_graph_file): Assign ID to BBs.
25543 (output_branch_count): Display BB # if verbose flag is set.
25544 (output_lines): Likewise for arcs.
25545 (print_usage): Add '--verbose' option help.
25546 * doc/gcov.texi: Document --verbose (-w) option.
25548 2017-04-28 Martin Liska <mliska@suse.cz>
25550 * gcov.c (struct block_location_info): New struct.
25551 (process_file): Fill up the new structure.
25552 (read_graph_file): Replace usage of encoding by the newly added
25554 (add_line_counts): Likewise.
25555 (accumulate_line_counts): Remove usage of the union.
25556 (function_info::function_info): New function.
25557 (function_info::~function_info): Likewise.
25558 (process_file): Call delete instead of release_function.
25559 (release_function): Release the function.
25560 (release_structures): Call delete instead of release_function.
25561 (solve_flow_graph): Replace usage of num_blocks.
25562 (find_exception_blocks): Likewise.
25563 (output_lines): Fix GNU coding style.
25565 2017-04-28 Martin Liska <mliska@suse.cz>
25568 * coverage.c (coverage_remove_note_file): New function.
25569 * coverage.h: Declare the function.
25570 * toplev.c (finalize): Clean if an error has been seen.
25572 2017-04-28 Martin Liska <mliska@suse.cz>
25574 PR gcov-profile/80031
25575 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25576 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25577 * gcov.c (read_graph_file): Read just number of blocks.
25578 * profile.c (branch_prob): Do not stream 0 flags per a basic
25581 2017-04-28 Martin Liska <mliska@suse.cz>
25583 * gcov-dump.c (tag_*): Add new argument to declarations.
25584 (dump_gcov_file): Likewise.
25585 (tag_blocks): Add and use new argument depth.
25586 (tag_arcs): Likewise.
25587 (tag_lines): Likewise.
25588 (tag_counters): Likewise.
25589 (tag_summary): Likewise.
25590 (dump_working_sets): Use depth to do a proper indentation.
25592 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25595 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25596 * symtab.c (symtab_node::debug_symtab): Move definition here.
25598 2017-04-28 Richard Biener <rguenther@suse.de>
25600 * lto-streamer.h (LTO_major_version): Bump to 7.
25602 2017-04-28 Richard Biener <rguenther@suse.de>
25604 * tree-vrp.c (assert_info): New struct.
25605 (add_assert_info): New helper.
25606 (register_edge_assert_for_2): Refactor to add asserts to a vector
25608 (register_edge_assert_for_1): Likewise.
25609 (register_edge_assert_for): Likewise.
25610 (finish_register_edge_assert_for): New helper actually registering
25611 asserts where live on edge.
25612 (find_conditional_asserts): Adjust.
25613 (find_switch_asserts): Likewise.
25614 (evrp_dom_walker::try_find_new_range): Generalize.
25615 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25617 2017-04-27 Marek Polacek <polacek@redhat.com>
25620 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25621 arg10 and arg11 to itype.
25623 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25625 * doc/extend.texi (Object Size Checking): Improve grammar.
25627 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25630 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25631 that the logic for permitting reciprocal estimates matches that
25634 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25637 * tree.c (type_cache_hasher::equal): Only compare
25638 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25639 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25640 non-aggregate element types.
25641 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25642 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25644 2017-04-27 Richard Biener <rguenther@suse.de>
25646 PR middle-end/80533
25647 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25648 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25649 keeping a reference to a trailing array.
25651 2017-04-27 Richard Biener <rguenther@suse.de>
25653 PR middle-end/80539
25654 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25655 being in loop-closed SSA form conservatively.
25656 (chrec_fold_multiply_poly_poly): Likewise.
25658 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25660 PR middle-end/79665
25661 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25662 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25664 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25667 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25668 (aarch64_function_arg_alignment): Return unsigned int again, but still
25669 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25670 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25671 Don't emit -Wpsabi note.
25672 (aarch64_function_arg_boundary): Likewise.
25673 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25676 2017-04-26 Nathan Sidwell <nathan@acm.org>
25678 * tree.h (crc32_unsigned_n): Declare.
25679 (crc32_unsigned, crc32_unsigned): Make inline.
25680 * tree.c (crc32_unsigned_bits): Replace with ...
25681 (crc32_unsigned_n): ... this.
25682 (crc32_unsigned, crc32_byte): Remove.
25683 (crc32_string): Remove unnecessary braces.
25685 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25687 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25688 * ipa-inline-analysis.c (MAX_TIME): Remove.
25689 (account_size_time): Use sreal for time.
25690 (dump_inline_summary): Update.
25691 (estimate_function_body_sizes): Update.
25692 (estimate_edge_size_and_time): Update.
25693 (estimate_calls_size_and_time): Update.
25694 (estimate_node_size_and_time): Update.
25695 (inline_merge_summary): Update.
25696 (inline_update_overall_summary): Update.
25697 (estimate_time_after_inlining): Update.
25698 (inline_read_section): Update.
25699 (inline_write_summary): Update.
25700 * ipa-inline.c (compute_uninlined_call_time): Update.
25701 (compute_inlined_call_time): Update.
25702 (recursive_inlining): Update.
25703 (inline_small_functions): Update.
25704 (dump_overall_stats): Update.
25705 * ipa-inline.h: Include sreal.h.
25706 (size_time_entry): Turn time to sreal.
25707 (inline_summary): Turn self_time nad time to sreal.
25709 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25711 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25713 (sreal::stream_out, sreal::stream_in): New.
25714 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25716 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25718 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25721 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25724 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25726 (dimode_scalar_chain::compute_convert_gain): Ditto.
25727 (dimode_scalar_chain::make_vector_copies): Ditto.
25728 (dimode_scalar_chain::convert_reg): Ditto.
25729 (dimode_scalar_chain::convert_insn): Ditto.
25730 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25731 (VI248_AVX512BW_1): New mode iterator.
25732 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25733 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25736 2017-04-25 Martin Sebor <msebor@redhat.com>
25738 PR tree-optimization/80497
25739 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25740 constants are representable in HOST_WIDE_INT.
25741 (parse_directive): Ditto.
25743 2017-04-25 Martin Sebor <msebor@redhat.com>
25746 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25747 (new_zero_array): Adjust signature.
25748 (dom_info::dom_init): Used unsigned rather that size_t.
25749 (dom_info::dom_info): Same.
25751 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25752 Jakub Jelinek <jakub@redhat.com>
25755 * config/arm/arm.c: Include gimple.h.
25756 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25757 returns negative, increment ncrn only if it returned positive.
25758 (arm_needs_doubleword_align): Return int instead of bool,
25759 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25760 members, but if there is any such non-FIELD_DECL
25761 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25762 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25763 returns negative, increment nregs only if it returned positive.
25764 (arm_setup_incoming_varargs): Likewise.
25765 (arm_function_arg_boundary): Emit -Wpsabi note if
25766 arm_needs_doubleword_align returns negative, return
25767 DOUBLEWORD_ALIGNMENT only if it returned positive.
25769 2017-04-25 Marek Polacek <polacek@redhat.com>
25772 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25773 first argument to type.
25775 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25778 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25779 type checks to test for compatibility instead of equality.
25781 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25782 Jakub Jelinek <jakub@redhat.com>
25785 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25787 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25788 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25789 the alignment computation, but return their maximum in warn_alignment.
25790 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25791 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25793 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25794 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25797 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25799 * config/arc/simdext.md (dmpyh): Fix typo.
25801 2017-04-25 Richard Biener <rguenther@suse.de>
25803 PR tree-optimization/80492
25804 * alias.c (compare_base_decls): Handle registers with asm
25805 specification conservatively.
25806 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25807 compare_base_decls returning dont-know properly.
25809 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25811 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25812 (legitimate_offset_address_p): New function.
25813 (arc_legitimate_address_p): Use above function.
25815 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25817 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25819 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25821 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25822 ACCH registers whenever they are available.
25824 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25826 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25827 double regs fix when not used.
25829 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25831 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25833 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25834 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25836 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25838 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25839 when emitting short ADD instructions.
25841 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25843 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25845 (cmpsi_cc_c_insn): Likewise.
25846 (cbranchsi4_scratch): Compute proper instruction length using
25847 compact_hreg_operand.
25848 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25850 2017-04-25 Richard Biener <rguenther@suse.de>
25852 PR middle-end/80509
25853 * passes.c (pass_manager::pass_manager): Initialize
25854 m_name_to_pass_map.
25856 2017-04-25 Richard Biener <rguenther@suse.de>
25858 PR tree-optimization/79201
25859 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25861 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25864 * config/s390/vector.md: Split MEM->GPR vector moves for
25865 non-s_operand addresses.
25867 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25870 * config/s390/predicates.md (reload_const_wide_int_operand): New
25872 * config/s390/s390.md ("movti"): Remove d/P alternative.
25873 ("movti_bigconst"): New pattern definition.
25875 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25878 * s390-protos.h (s390_expand_cs_hqi): Removed.
25879 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25880 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25881 modes as well as CCZ1mode and CCZmode.
25882 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25883 signature of s390_emit_compare_and_swap.
25884 (s390_expand_cs_hqi): Likewise, make static.
25885 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25886 compare-and-swap, in some cases.
25887 (s390_expand_cs): Wrapper function.
25888 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25890 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25891 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25892 patterns for small and large integers. Forbid symref memory operands.
25893 Move expander to s390.c. Require cc register.
25894 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25895 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25896 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25897 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25898 symref memory operands. Remove CC mode and call s390_match_ccmode
25900 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25902 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25904 * config/s390/s390.md (define_peephole2): New peephole to help
25905 combining the load-and-test pattern with volatile memory.
25907 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25909 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25910 with CCZmode for TARGET_Z196.
25912 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25914 PR rtl-optimization/80501
25915 * combine.c (make_compound_operation_int): Set subreg_code to SET
25916 even for AND with mask of the sign bit of mode.
25918 PR rtl-optimization/80500
25919 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25920 sum's initial value.
25922 2017-04-25 Julian Brown <julian@codesourcery.com>
25923 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25925 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25927 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25929 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25931 2017-04-25 Julian Brown <julian@codesourcery.com>
25932 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25934 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25935 (thunderx2t99_sha): New Reservation.
25937 2017-04-25 Julian Brown <julian@codesourcery.com>
25938 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25940 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25941 type for 1-element load.
25943 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25945 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25947 2017-04-24 Martin Jambor <mjambor@suse.cz>
25949 PR tree-optimization/80293
25950 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25951 char arrays not totally scalarizable if it is false.
25952 (analyze_all_variable_accesses): Pass correct value in the new
25953 parameter. Add a statistics counter.
25955 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25957 PR middle-end/79931
25958 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25960 2017-04-24 Richard Biener <rguenther@suse.de>
25962 PR tree-optimization/80494
25963 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25964 out for complex types.
25966 2017-04-24 Richard Biener <rguenther@suse.de>
25968 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25969 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25970 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25971 (DFS): Adjust and never fail.
25972 (sccvn_dom_walker::fail): Remove.
25973 (sccvn_dom_walker::before_dom_children): Adjust.
25974 (run_scc_vn): Likewise and never fail.
25975 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25976 (pass_fre::execute): Likewise.
25978 2017-04-24 Richard Biener <rguenther@suse.de>
25980 PR tree-optimization/79725
25981 * tree-ssa-sink.c (statement_sink_location): Return whether
25982 failure reason was zero uses. Move that check later.
25983 (sink_code_in_bb): Deal with zero uses by removing the stmt
25986 2017-04-24 Richard Biener <rguenther@suse.de>
25989 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25990 pointer-based references.
25992 2017-04-24 Richard Biener <rguenther@suse.de>
25995 * pass_manager.h (pass_manager::operator new): Remove.
25996 (pass_manager::operator delete): Likewise.
25997 * passes.c (pass_manager::operator new): Remove.
25998 (pass_manager::operator delete): Likewise.
25999 (pass_manager::pass_manager): Zero individual pass members.
26001 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26004 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26005 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26006 Check "XEXP (src, 1)" operand here.
26007 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26008 Check "XEXP (src, 1)" operand here.
26009 (dimode_scalar_chain::make_vector_copies): Detect count register
26010 of a shift instruction. Zero extend count register from QImode
26011 to DImode to satisfy vector shift pattern count operand predicate.
26012 Substitute vector shift count operand with a DImode copy.
26013 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26016 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26018 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26019 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26020 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26021 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26022 (UNSPEC_NOREX_MEM): Remove definition.
26024 2017-04-21 Richard Biener <rguenther@suse.de>
26026 PR tree-optimization/79547
26027 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26028 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26029 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26030 without any constraints.
26032 2017-04-21 Richard Biener <rguenther@suse.de>
26034 PR tree-optimization/78847
26035 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26037 2017-04-21 Richard Biener <rguenther@suse.de>
26039 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26040 (build_distinct_type_copy): Likewise.
26041 (build_variant_type_copy): Likewise.
26042 * tree.c (build_qualified_type): Pass down mem-stat info.
26043 (build_distinct_type_copy): Likewise.
26044 (build_variant_type_copy): Likewise.
26046 2017-04-21 Richard Biener <rguenther@suse.de>
26048 PR tree-optimization/80237
26049 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26051 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26052 for a simplified result.
26054 2016-04-21 Richard Biener <rguenther@suse.de>
26056 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26057 sth as strict as a simple_iv but a chrec without symbols and an
26058 operand defined in the loop we are peeling (and not some subloop).
26059 (propagate_constants_for_unrolling): Propagate all constants.
26061 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26064 * config/i386/i386.c (print_reg): Remove assert for disalowed
26065 regno values, call output_operand_lossage instead.
26067 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26070 * config/i386/constraints.md (Yc): New register constraint.
26071 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26072 Use Yc constraint for alternative 2 of operand 0. Remove
26073 preferred_for_speed attribute.
26075 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26077 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26078 lastprivate clauses in SIMT case.
26080 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26082 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26084 2017-04-20 Richard Biener <rguenther@suse.de>
26086 PR tree-optimization/57796
26087 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26088 as N scalar stores.
26089 (vect_model_load_cost): Cost gathers as N scalar loads.
26091 2017-04-20 Richard Biener <rguenther@suse.de>
26093 * ggc-page.c (ggc_allocated_p): Rename to ...
26094 (safe_lookup_page_table_entry): ... this and return the lookup
26096 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26098 2017-04-20 Richard Biener <rguenther@suse.de>
26100 PR tree-optimization/80453
26101 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26102 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26103 from the conditions.
26104 (vn_phi_eq): Pass them down.
26105 (vn_phi_lookup): Record them.
26106 (vn_phi_insert): Likewise.
26108 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26110 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26111 uninitialized variable warning to avoid buffer overrun.
26113 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26116 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26117 is suppressed for '{ 0 }' in C.
26119 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26121 * BASE-VER: Set to 8.0.0.
26123 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26125 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26126 priority .init_array and .fini_array section with SECTION_NOTYPE
26129 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26131 PR middle-end/80423
26132 * tree.h (build_array_type): Add typeless_storage default argument.
26133 * tree.c (type_cache_hasher::equal): Also compare
26134 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26135 (build_array_type): Add typeless_storage argument, set
26136 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26138 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26139 (build_array_type): Likewise. Add typeless_storage argument.
26141 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26142 Jakub Jelinek <jakub@redhat.com>
26144 PR tree-optimization/80426
26145 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26146 operation on symbolic operands, also compute the overflow for the
26147 invariant part when the operation degenerates into a negation.
26149 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26152 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26153 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26156 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26158 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26161 * config/avr/avr.c (tree.h): Include it.
26162 (cgraph.h): Include it.
26163 (avr_encode_section_info): Don't warn for uninitialized progmem
26164 variable if it's just an alias.
26166 2017-04-19 Richard Biener <rguenther@suse.de>
26169 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26170 when needed by AutoPGO.
26172 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26175 * doc/lto.texi: Remove an extra 'that'.
26177 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26179 PR rtl-optimization/80429
26180 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26181 are only used in debug insns.
26183 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26184 Vladimir Makarov <vmakarov@redhat.com>
26186 * config/sparc/predicates.md (input_operand): Add comment. Return
26187 true for any memory operand when LRA is in progress.
26188 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26190 2017-04-18 Jeff Law <law@redhat.com>
26193 * mips.md ({return,simple_return}_internal): Do not overwrite
26196 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26198 PR tree-optimization/80443
26199 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26200 instead of adding 1, subtract -1 and similarly instead of subtracting
26203 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26205 PR rtl-optimization/80357
26206 * haifa-sched.c (tmp_bitmap): New variable.
26207 (model_recompute): Handle duplicate use records.
26208 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26209 (free_global_sched_pressure_data): Free it.
26211 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26214 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26215 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26216 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26217 instead of SYSTEM_HEADER_DIR.
26219 2017-04-18 Jeff Law <law@redhat.com>
26221 PR middle-end/80422
26222 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26223 predecessors after walking up the insn chain.
26225 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26228 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26229 sizetype type into debug info.
26231 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26234 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26235 unneeded test for TARGET_UPPER_REGS_SF.
26236 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26238 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26241 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26242 instead of gsi_after_labels.
26244 2017-04-18 Jeff Law <law@redhat.com>
26246 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26250 2017-04-13 Jeff Law <law@redhat.com>
26251 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26252 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26254 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26257 * config/avr/avr.c (intl.h): Include it.
26258 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26260 2017-04-18 Martin Liska <mliska@suse.cz>
26262 PR gcov-profile/78783
26263 * gcov-tool.c (gcov_output_files): Validate that destination
26264 file is either removed by the tool or by a user.
26266 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26267 Guy Benyei <guybe@mellanox.com>
26269 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26270 block, and do not negate it, the stored id is already negative.
26272 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26274 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26276 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26279 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26280 masks of options that should be turned off if the VSX vector
26281 options are turned off.
26282 (OTHER_P8_VECTOR_MASKS): Likewise.
26283 (OTHER_VSX_VECTOR_MASKS): Likewise.
26284 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26285 rs6000_disable_incompatible_switches to validate no type switches
26287 (rs6000_incompatible_switch): New function to disallow turning on
26288 other vector options if -mno-vsx, -mno-power8-vector, or
26289 -mno-power9-vector are specified.
26291 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26293 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26295 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26297 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26298 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26299 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26300 (ARG_POINTER_CFA_OFFSET): Likewise.
26302 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26304 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26305 conditions to take advantage of various optimizations.
26307 2017-04-13 Jeff Law <law@redhat.com>
26309 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26310 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26311 (zero_extendsidi2_dext): Likewise.
26313 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26316 * fold-const.c (fold_ternary_loc): Revert
26317 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26320 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26322 PR rtl-optimization/80343
26323 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26324 new scratch pseudo.
26326 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26329 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26330 to ubsan_encode_value.
26332 2017-04-13 Jeff Law <law@redhat.com>
26334 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26335 appearing in DEBUG_INSNs.
26337 2017-04-13 Martin Liska <mliska@suse.cz>
26339 PR gcov-profile/80413
26340 * gcov-io.c (gcov_write_string): Copy to buffer just when
26341 allocated size is greater than zero.
26343 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26346 * dwarf2out.c (decls_for_scope): Ignore declarations of
26347 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26349 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26352 * ipa-visibility.c (non_local_p): Fix typos.
26353 (localize_node): When localizing symbol in same comdat group,
26354 dissolve the group only when we know external symbols are going
26356 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26358 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26360 PR tree-optimization/79390
26361 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26362 order does not result in usable sequence, retry with reversed operand
26368 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26369 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26370 op0 instead of fold_convert_loc (loc, type, arg0).
26372 2017-04-12 Jeff Law <law@redhat.com>
26374 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26375 has a delay slot in the generated code.
26377 * config/cris/cris.md (cris_preferred_reload_class): Return
26378 GENNONACR_REGS rather than GENERAL_REGS.
26380 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26383 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26384 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26385 signedness of the result type.
26387 2017-04-12 Richard Biener <rguenther@suse.de>
26388 Jeff Law <law@redhat.com>
26390 PR tree-optimization/80359
26391 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26392 trim stores to TARGET_MEM_REFs.
26394 2017-04-12 Richard Biener <rguenther@suse.de>
26396 PR tree-optimization/79390
26397 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26398 threading case even more.
26400 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26403 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26404 for quad_address_p for TImode, instead of just not indexed_address.
26406 2017-04-12 Richard Biener <rguenther@suse.de>
26407 Bernd Edlinger <bernd.edlinger@hotmail.de>
26409 PR middle-end/79671
26410 * alias.c (component_uses_parent_alias_set_from): Handle
26411 TYPE_TYPELESS_STORAGE.
26412 (get_alias_set): Likewise.
26413 * tree-core.h (tree_type_common): Add typeless_storage flag.
26414 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26415 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26416 for types containing members with TYPE_TYPELESS_STORAGE.
26417 (place_field): Likewise.
26418 (layout_type): Likewise for ARRAY_TYPE.
26419 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26420 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26421 TYPE_TYPELESS_STORAGE.
26422 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26424 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26427 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26428 first argument to type.
26430 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26434 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26435 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26436 (rs6000_expand_binop_builtin): Likewise.
26437 (rs6000_expand_ternop_builtin): Likewise; also add missing
26438 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26439 vshasigma built-ins.
26440 * doc/extend.texi: Document that vec_xxpermdi's third argument
26441 must be a constant.
26443 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26445 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26446 Use shift_const cost parameter when calculating gain of STV shifts.
26448 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26450 PR rtl-optimization/70478
26451 * lra-constraints.c (process_alt_operands): Check memory for
26452 disfavoring memory insn operand.
26454 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26456 PR middle-end/80100
26457 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26458 left shift in unsigned HOST_WIDE_INT type.
26460 PR rtl-optimization/80385
26461 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26462 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26465 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26466 if they have any depend clauses.
26468 2017-04-11 Martin Liska <mliska@suse.cz>
26471 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26472 * ipa-split.c (split_function): Create a local comdat symbol
26473 if caller is in a comdat group.
26475 2017-04-11 Martin Liska <mliska@suse.cz>
26478 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26481 2017-04-11 Martin Sebor <msebor@redhat.com>
26483 PR middle-end/80364
26484 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26485 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26487 (directive::set_width, directive::set_precision, format_character):
26489 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26492 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26495 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26496 conflict, set target->arch_name instead of target->cpu_name.
26498 2017-04-11 Richard Biener <rguenther@suse.de>
26500 PR tree-optimization/80374
26501 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26502 build_zero_cst, remove fold_convertible_p check again.
26504 2017-04-11 Martin Liska <mliska@suse.cz>
26507 * ubsan.c (instrument_object_size): Do not instrument register
26510 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26513 * config/i386/i386-builtin-types.def
26514 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26515 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26516 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26517 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26518 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26519 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26520 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26521 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26522 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26523 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26524 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26525 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26526 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26527 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26528 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26529 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26530 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26531 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26532 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26533 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26534 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26535 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26536 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26537 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26538 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26539 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26540 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26541 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26542 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26543 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26544 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26545 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26546 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26547 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26548 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26549 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26550 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26551 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26552 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26553 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26554 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26555 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26556 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26557 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26558 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26559 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26561 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26562 flag to second_arg_count, handle 4 argument function type _COUNT
26563 aliases, handle second_arg_count on second argument rather than last.
26565 2017-04-10 Jeff Law <law@redhat.com>
26567 PR tree-optimization/80374
26568 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26569 record anything if we can not convert integer_zero_node to the
26572 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26575 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26576 Enhance special handling given to the TARGET_P9_MINMAX option in
26577 relation to certain other options.
26579 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26581 PR tree-optimization/80153
26582 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26583 remove POINTER_PLUS_EXPR's base part directly, rather than through
26586 2017-04-10 Richard Biener <rguenther@suse.de>
26587 Bin Cheng <bin.cheng@arm.com>
26589 PR tree-optimization/80153
26590 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26591 the first element of pointer type aff_tree. Build result expr in
26593 (add_elt_to_tree): Convert to type unconditionally. Remove other
26594 fold_convert calls.
26595 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26596 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26598 2017-04-10 Richard Biener <rguenther@suse.de>
26600 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26603 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26605 PR rtl-optimization/70478
26606 * lra-constraints.c (curr_small_class_check): New.
26607 (update_and_check_small_class_inputs): New.
26608 (process_alt_operands): Update curr_small_class_check. Disfavor
26609 alternative insn memory operands. Check available regs for small
26612 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26615 * config/mips/mips.opt (-mvirt): Update description.
26616 * doc/invoke.texi (-mvirt): Likewise.
26618 2017-04-10 Richard Biener <rguenther@suse.de>
26620 PR middle-end/80362
26621 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26622 looking for NEGATE_EXPR in -A / -B to A / B folding.
26624 2017-04-10 Martin Liska <mliska@suse.cz>
26626 PR gcov-profile/80224
26627 * gcov.c (print_usage): Fix usage string.
26628 (get_gcov_intermediate_filename): Remove.
26629 (output_gcov_file): Use both for normal and intermediate format.
26630 (generate_results): Do not initialize special file for
26631 intermediate format.
26633 2017-04-10 Richard Biener <rguenther@suse.de>
26635 PR tree-optimization/80304
26636 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26639 2017-04-10 Nathan Sidwell <nathan@acm.org>
26642 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26643 (rs6000_init_builtins): Use it.
26645 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26647 * config/arm/arm.md (<mrc>): Add mode to SET source.
26648 (<mrrc>): Likewise.
26650 2017-04-10 Richard Biener <rguenther@suse.de>
26652 PR middle-end/80344
26653 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26655 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26658 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26659 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26660 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26661 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26662 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26663 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26664 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26665 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26666 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26667 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26668 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26669 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26670 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26671 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26672 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26673 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26674 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26675 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26676 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26677 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26678 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26679 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26680 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26682 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26684 PR rtl-optimization/70478
26685 * lra-constraints.c: Reverse the last patch.
26687 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26689 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26690 Add comment for WCHAR_T.
26692 2017-04-08 Martin Liska <mliska@suse.cz>
26695 2017-04-07 Martin Liska <mliska@suse.cz>
26698 * ipa-split.c (split_function): Add function part to a same comdat
26701 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26704 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26706 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26708 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26709 (vec_pairing): ...to this.
26710 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26711 (rs6000_sched_init): Adjust for name change.
26712 (struct rs6000_sched_context): Likewise.
26713 (rs6000_init_sched_context): Likewise.
26714 (rs6000_set_sched_context): Likewise.
26716 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26722 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26724 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26725 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26726 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26728 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26730 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26732 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26734 PR rtl-optimization/70703
26735 * ira-color.c (update_conflict_hard_regno_costs): Use
26736 int64_t instead of HOST_WIDE_INT.
26738 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26740 PR rtl-optimization/70478
26741 * lra-constraints.c (process_alt_operands): Disfavor alternative
26742 insn memory operands.
26744 2017-04-07 Jeff Law <law@redhat.com>
26746 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26747 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26749 2017-04-07 Martin Liska <mliska@suse.cz>
26752 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26753 Show error message instead of an ICE.
26755 2017-04-07 Martin Liska <mliska@suse.cz>
26758 * ipa-split.c (split_function): Add function part to a same comdat
26761 2017-04-07 Richard Biener <rguenther@suse.de>
26763 PR middle-end/80341
26764 * tree.c (get_unwidened): Also handle ! for_type case for
26766 * convert.c (do_narrow): Split out from ...
26767 (convert_to_integer_1): ... here. Do not pass final truncation
26768 type to get_unwidened for TRUNC_DIV_EXPR.
26770 2017-04-07 Richard Biener <rguenther@suse.de>
26772 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26774 (aff_combination_const): Adjust.
26775 (aff_combination_scale): Likewise.
26776 (aff_combination_add_elt): Likewise.
26777 (aff_combination_add_cst): Likewise.
26778 (aff_combination_convert): Likewise.
26779 (add_elt_to_tree): Likewise. Remove unused argument.
26780 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26782 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26784 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26786 * config/arm/arm.c (arm_default_short_enums): Use
26787 ARM_DEFAULT_SHORT_ENUMS.
26788 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26790 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26793 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26794 members with redundant out-of-class redeclaration.
26796 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26799 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26800 * config/i386/i386.md (*zero_extendsidi2):
26801 Add (?*x,*x) and (?*v,*v) alternatives.
26803 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26806 * config/i386/i386.c (ix86_expand_builtin)
26807 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26808 mode from insn data. Convert operands to insn operand mode.
26809 Copy operands that don't satisfy insn predicate to a register.
26811 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26813 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26816 2017-04-06 Richard Biener <rguenther@suse.de>
26818 PR tree-optimization/80334
26819 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26820 preserve alignment of accesses.
26822 2017-04-06 Richard Biener <rguenther@suse.de>
26824 PR tree-optimization/80262
26825 * tree-sra.c (build_ref_for_offset): Preserve address-space
26827 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26828 Drop useless address-space information on MEM_REF offsets.
26830 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26832 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26834 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26836 PR rtl-optimization/70703
26837 * ira-color.c (update_conflict_hard_regno_costs): Use
26838 HOST_WIDE_INT instead of long.
26840 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26843 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26844 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26846 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26847 for x86_64 target. Handle -m3dnowa option.
26849 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26851 PR rtl-optimization/70703
26852 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26853 (update_conflict_hard_regno_costs): Use long instead of unsigned
26854 arithmetic for cost calculation.
26856 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26857 Bernd Edlinger <bernd.edlinger@hotmail.de>
26860 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26863 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26866 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26867 ptr_mode with Pmode throughout.
26868 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26869 into probe_stack_range and use DImode.
26871 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26874 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26875 call_eh_return is true.
26877 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26879 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26880 Initialize last_match_fntype_index.
26882 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26885 * tree-nvr.c: Include internal-fn.h.
26886 (pass_return_slot::execute): Ignore internal calls without
26889 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26890 Richard Biener <rguenther@suse.de>
26893 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26894 captures used multiple times, except for the last use.
26895 * generic-match-head.c: Include gimplify.h.
26897 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26899 PR tree-optimization/79390
26900 * target.h (struct noce_if_info): Declare.
26901 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26902 * target.def (noce_conversion_profitable_p): New target hook.
26903 * ifcvt.h (struct noce_if_info): New type, moved from ...
26904 * ifcvt.c (struct noce_if_info): ... here.
26905 (noce_conversion_profitable_p): Renamed to ...
26906 (default_noce_conversion_profitable_p): ... this. No longer
26908 (noce_try_store_flag_constants, noce_try_addcc,
26909 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26910 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26911 instead of noce_conversion_profitable_p.
26912 * config/i386/i386.c: Include ifcvt.h.
26913 (ix86_option_override_internal): Don't override
26914 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26915 (ix86_noce_conversion_profitable_p): New function.
26916 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26917 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26918 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26919 * doc/tm.texi: Regenerated.
26921 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26923 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26926 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26929 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26930 instructions for small multiply cores.
26932 2017-04-04 Jeff Law <law@redhat.com>
26934 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26936 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26937 that are not set by the loop over the elements.
26939 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26942 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26943 int mode, convert_modes it to mode as unsigned, otherwise use
26944 lowpart_subreg to mode rather than SImode.
26945 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26946 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26947 Use DImode instead of SImode for the shift count operand.
26948 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26951 2017-04-04 Richard Biener <rguenther@suse.de>
26953 PR middle-end/80281
26954 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26955 arithmetic done for the negate or the plus. Simplify.
26956 (A - (-B) -> A + B): Likewise.
26957 * fold-const.c (split_tree): Make sure to not negate pointers.
26959 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26961 PR rtl-optimization/60818
26962 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26963 a compare of comparisons with the thing compared if this results
26964 in a different machine mode.
26966 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26968 * alias.c (base_alias_check): Fix typo in comment.
26969 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26970 * cgraphunit.c (symbol_table::compile): Likewise.
26971 * collect2.c (maybe_run_lto_and_relink): Likewise.
26972 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26973 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26974 * config/avr/avr.c (avr_map_op_t): Likewise.
26975 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26976 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26977 * config/epiphany/epiphany.md (movcc): Likewise.
26978 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26979 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26981 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26982 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26983 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26984 * config/sh/sh.c (sh_rtx_costs): Likewise.
26985 * fold-const.c (fold_truth_andor): Likewise.
26986 * genautomata.c (collapse_flag): Likewise.
26987 * gengtype.h (struct type::u::s): Likewise.
26988 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26989 * input.c (FORMAT_AMOUNT): Likewise.
26990 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26991 (known_aggs_to_agg_replacement_list): Likewise.
26992 * ipa-inline-analysis.c: Likewise.
26993 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26994 * ipa-polymorphic-call.c
26995 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26996 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26997 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26999 * modulo-sched.c (apply_reg_moves): Likewise.
27000 * omp-expand.c (build_omp_regions_1): Likewise.
27001 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27002 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27003 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27004 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27005 * value-prof.c: Likewise.
27006 * var-tracking.c (val_reset): Likewise.
27008 2017-04-03 Richard Biener <rguenther@suse.de>
27010 PR tree-optimization/80275
27011 * fold-const.c (split_address_to_core_and_offset): Handle
27014 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27016 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27017 descriptors is at least equal to that of functions.
27019 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27021 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27023 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27026 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27027 (mov<IMOD4:mode>): New expander.
27028 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27030 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27032 PR rtl-optimization/79405
27033 * fwprop.c (propagations_left): New variable.
27034 (forward_propagate_into): Decrement it.
27035 (fwprop_init): Initialize it.
27036 (fw_prop): If the variable has reached zero, stop propagating.
27037 (fwprop_addr): Ditto.
27039 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27042 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27043 a FUNCTION_DECL, pass it as decl instead of origin to
27046 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27048 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27051 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27054 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27055 TARGET_VSX_SMALL_INTEGER.
27057 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27059 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27060 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27062 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27064 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27065 extraction from odd-numbered MSA register.
27067 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27069 PR middle-end/80173
27070 * expmed.c (store_bit_field_1): Don't attempt to create
27071 a word subreg out of hard registers wider than word if they
27072 have HARD_REGNO_NREGS of 1 for their mode.
27074 PR middle-end/80163
27075 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27076 conversions to integer types wider than word and pointer.
27079 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27080 (rtx_equal_for_cselib_p): Pass 0 to it.
27081 * cselib.c (cselib_hasher::equal): Likewise.
27082 (rtx_equal_for_cselib_1): Add depth argument. If depth
27083 is 128, don't look up VALUE locs and punt. Increment
27084 depth in recursive calls when walking VALUE locs.
27086 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27088 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27089 (make_gcov_file_name): Use the canonical path name for generating
27091 (read_line): Fix handling of files with ascii null bytes.
27093 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27095 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27096 to initialise a vector register instead
27097 of using a const_int.
27099 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27101 PR translation/80189
27102 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27103 diagnostic messages.
27105 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27108 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27109 (dfp_diex_<mode>): Update mode of operand 1.
27110 * doc/extend.texi (dxex, dxexq): Document change to return type.
27111 (diex, diexq): Document change to argument type.
27113 2017-03-30 Martin Jambor <mjambor@suse.cz>
27116 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27117 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27118 it reflects the signature changes performed at the callee side.
27119 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27120 to cgraph_build_function_type_skip_args.
27121 (build_function_decl_skip_args): Adjust call to the above function.
27123 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27126 * config/i386/sse.md
27127 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27128 register as dest whenever it is a MEM not rtx_equal_p to the
27129 corresponding dup operand, and when forcing into reg move the
27130 reg into the memory afterwards.
27131 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27132 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27133 for the force_reg mode.
27134 (avx512vl_vextractf128<mode>): Use register as dest either
27135 always when a MEM, or when it is a MEM not rtx_equal_p to the
27136 corresponding dup operand, or even not when it is a CONST_VECTOR
27137 depending on the mode and lo vs. hi.
27138 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27140 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27141 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27142 Likewise. Require that operands[2] is even.
27143 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27144 Remove extraneous parens. Require that operands[2] is a multiple
27146 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27147 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27148 disallow memory then.
27150 2017-03-30 Richard Biener <rguenther@suse.de>
27152 PR tree-optimization/77498
27153 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27154 to non-constants over backedges.
27156 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27158 PR rtl-optimization/80233
27159 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27160 as last_combined_insn. Do not test for BARRIER_P separately.
27162 2017-03-29 Andreas Schwab <schwab@suse.de>
27165 * calls.c (prepare_call_address): Convert funexp to Pmode before
27166 copying to temp reg.
27168 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27170 PR tree-optimization/80158
27171 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27172 Handle possible future case of more than one alternate
27174 (replace_rhs_if_not_dup): Likewise.
27175 (replace_one_candidate): Likewise.
27177 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27179 PR rtl-optimization/80193
27180 * ira.c (ira): Do not check allocation for LRA.
27182 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27184 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27185 (nvptx_output_simt_exit): Declare.
27186 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27187 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27188 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27189 (nvptx_declare_function_name): ...here. Emit declaration of local
27190 memory space buffer for omp_simt_enter insn.
27191 (nvptx_output_unisimt_switch): New.
27192 (nvptx_output_softstack_switch): New.
27193 (nvptx_output_simt_enter): New.
27194 (nvptx_output_simt_exit): New.
27195 * config/nvptx/nvptx.h (struct machine_function): New fields
27196 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27197 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27198 (UNSPECV_SIMT_EXIT): Ditto.
27199 (omp_simt_enter_insn): New insn.
27200 (omp_simt_enter): New expansion.
27201 (omp_simt_exit): New insn.
27202 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27204 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27205 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27206 (expand_GOMP_SIMT_EXIT): New.
27207 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27208 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27209 (GOMP_SIMT_EXIT): Ditto.
27210 * target-insns.def (omp_simt_enter): New insn.
27211 (omp_simt_exit): Ditto.
27212 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27214 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27215 (lower_rec_input_clauses): Likewise.
27216 (lower_lastprivate_clauses): Handle SIMT privatization.
27218 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27219 (ompdevlow_adjust_simt_enter): New.
27220 (find_simtpriv_var_op): New.
27221 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27222 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27224 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27225 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27226 (copy_decl_for_dup_finish): Ditto.
27228 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27230 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27233 * config/i386/i386.c (ix86_option_override_internal): Always
27234 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27236 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27238 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27240 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27242 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27243 mark new edge's irreducible flag accordign to it.
27244 (vect_do_peeling): Check loop preheader edge's irreducible flag
27245 and pass it to function slpeel_add_loop_guard.
27247 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27249 PR tree-optimization/80218
27250 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27251 Update block frequencies and counts.
27253 2017-03-28 Richard Biener <rguenther@suse.de>
27255 PR tree-optimization/78644
27256 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27257 of a simplification result we may not use it at all.
27259 2017-03-28 Richard Biener <rguenther@suse.de>
27262 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27263 without arguments, generate default definition of a SSA name.
27265 2017-03-28 Richard Biener <rguenther@suse.de>
27267 PR middle-end/80222
27268 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27269 TYPE_REF_CAN_ALIAS_ALL references.
27270 * fold-const.c (fold_indirect_ref_1): Likewise.
27272 2017-03-28 Martin Liska <mliska@suse.cz>
27275 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27276 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27278 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27279 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27281 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27282 (EXTRA_SPECS): Define.
27283 (SUBTARGET_EXTRA_SPECS): Likewise.
27284 (SUBTARGET_CPP_SPEC): Likewise.
27285 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27286 SUBTARGET_EXTRA_SPECS.
27287 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27289 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27291 * config/arc/simdext.md (vst64_insn): Update pattern.
27292 (vld32wh_insn): Likewise.
27293 (vld32wl_insn): Likewise.
27294 (vld64_insn): Likewise.
27295 (vld32_insn): Likewise.
27297 2017-03-28 Marek Polacek <polacek@redhat.com>
27300 * fold-const.c (fold_comparison): Use protected_set_expr_location
27301 instead of SET_EXPR_LOCATION.
27303 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27305 * tree.c (add_expr): Avoid name lookup warning.
27307 2017-03-27 Jeff Law <law@redhat.com>
27309 PR tree-optimization/80216
27310 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27311 function name. Limit recursion depth.
27312 (record_temporary_equivalences): Corresponding changes.
27314 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27316 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27319 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27322 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27324 * cfgcleanup.c (reg_note_cfa_p): New array.
27325 (insns_have_identical_cfa_notes): New function.
27326 (old_insns_match_p): Don't cross-jump in between /f
27327 and non-/f instructions. If both i1 and i2 are frame related,
27328 verify all CFA notes, their order and content.
27330 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27333 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27334 HImode and SImode with zero extend to DImode to one insn.
27335 (bswap<mode>2_extenddi): Likewise.
27336 (bswapsi2_extenddi): Likewise.
27337 (bswaphi2_extendsi): Likewise.
27338 (bswaphi2): Combine bswap HImode and SImode into one insn.
27339 Separate memory insns from swapping register.
27340 (bswapsi2): Likewise.
27341 (bswap<mode>2): Likewise.
27342 (bswaphi2_internal): Delete, no longer used.
27343 (bswapsi2_internal): Likewise.
27344 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27345 store, and gpr<-gpr swap insns.
27346 (bswap<mode>2_store): Likewise.
27347 (bswaphi2_reg): Register only splitter, combine with the splitter.
27348 (bswaphi2 splitter): Likewise.
27349 (bswapsi2_reg): Likewise.
27350 (bswapsi2 splitter): Likewise.
27351 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27352 the insns into load, store, and register/register insns.
27353 (bswapdi2_ldbrx): Likewise.
27354 (bswapdi2_load): Likewise.
27355 (bswapdi2_store): Likewise.
27356 (bswapdi2_reg): Likewise.
27358 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27360 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27361 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27363 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27366 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27368 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27369 special handling for target option conflicts between dform
27370 options (-mpower9-dform, -mpower9-dform-vector,
27371 -mpower9-dform-scalar) and -mno-direct-move.
27373 2017-03-27 Richard Biener <rguenther@suse.de>
27375 PR tree-optimization/80181
27376 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27378 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27380 * config/arc/predicates.md (move_double_src_operand): Replace the
27381 call to move_double_src_operand with a call to address_operand.
27383 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27385 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27386 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27387 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27389 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27391 * config/arc/predicates.md (long_immediate_loadstore_operand):
27392 Consider scaled addresses cases.
27394 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27396 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27397 restored when in interrupt.
27398 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27399 doesn't have delay slot.
27401 2017-03-27 Richard Biener <rguenther@suse.de>
27404 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27405 inlined thunk clones.
27407 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27410 * asan.c (instrument_derefs): Copy over last operand from
27411 original COMPONENT_REF to the new COMPONENT_REF with
27412 DECL_BIT_FIELD_REPRESENTATIVE.
27413 * ubsan.c (instrument_object_size): Likewise.
27415 2017-03-27 Richard Biener <rguenther@suse.de>
27417 PR tree-optimization/80170
27418 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27419 sure DR/SCEV didnt fold in constants we do not see when looking
27420 at the reference base alignment.
27422 2017-03-27 Richard Biener <rguenther@suse.de>
27424 PR middle-end/80171
27425 * gimple-fold.c (fold_ctor_reference): Properly guard against
27426 NULL return value from canonicalize_constructor_val.
27428 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27431 * config/i386/i386.c (ix86_expand_builtin)
27432 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27433 flags reg setting and flags reg using instructions.
27434 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27435 clobbering instructions to zero extend op2.
27437 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27439 * doc/install.texi (Configuration) <--with-aix-soname>:
27440 Update link to AIX ld.
27442 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27444 PR rtl-optimization/80160
27445 PR rtl-optimization/80159
27446 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27447 reg_alternate_class into account.
27449 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27452 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27453 to consider in curr_insn_transform.
27455 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27457 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27458 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27459 and emit_mode_inner.
27461 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27463 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27464 argument to the overloaded builtin variants. Use the new flag to
27465 deprecate certain builtin variants.
27466 * config/s390/s390-builtin-types.def: Add new builtin types.
27467 * config/s390/s390-builtins.h: Support new flags field for
27468 overloaded builtins.
27469 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27470 (s390_macro_to_expand): Enable vector float data type.
27471 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27472 builtins by incrementing the __VEC__ version number.
27473 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27475 (s390_resolve_overloaded_builtin): Emit error messages depending
27476 on the builtin flags.
27477 * config/s390/s390.c (s390_expand_builtin): Support additional
27478 flags argument. Change error message to match the messages
27479 emitted in s390-c.c.
27480 * config/s390/s390.md: New UNSPEC_* constants.
27481 (op_type): Add new instruction types.
27482 * config/s390/vecintrin.h: Add new builtins and test data class
27484 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27485 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27486 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27487 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27488 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27489 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27491 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27492 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27493 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27494 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27496 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27497 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27498 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27499 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27501 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27502 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27503 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27504 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27505 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27506 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27507 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27509 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27510 ("vec_scatter_element<V_HW_4:mode>_DI")
27511 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27512 ("vec_fpint<mode>", "vflls")
27513 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27514 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27515 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27516 ("*vec_cmphe<mode>_cc"): ... these.
27518 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27519 mode constant instead of magic value.
27521 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27523 * config/s390/s390.c (s390_expand_vec_compare): Support other
27524 vector floating point modes than just V2DF.
27525 (s390_expand_vcond): Likewise.
27526 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27527 (s390_cannot_change_mode_class): Prevent mode changes between TF
27528 and V1TF in vector registers.
27529 * config/s390/s390.md (DF, SF): New mode attributes.
27530 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27531 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27532 SFmode support for VRs.
27533 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27535 (VFT, VF_HW): New mode iterators.
27536 (vw, sdx): New mode attributes.
27537 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27538 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27539 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27540 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27541 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27542 also the new vector floating point modes. Renaming to ...
27544 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27545 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27546 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27547 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27548 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27549 ("vec_unordered<mode>"): ... these.
27551 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27552 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27553 ("*vec_extendv2df"): New insn definitions.
27555 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27557 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27558 ("mulditi3_2", "*muldi3_sign"): New patterns.
27559 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27560 rename the pattern definition.
27562 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27564 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27566 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27568 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27570 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27571 instruction if possible.
27572 * config/s390/vector.md (vec_halfnumelts): New mode
27574 ("*vec_vllezlf<mode>"): New pattern.
27576 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27578 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27579 ("popcountv4si2", "popcountv2di2"): Rename to ...
27580 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27581 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27583 ("popcount<mode>2_vxe"): New pattern.
27585 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27587 * common/config/s390/s390-common.c (processor_flags_table): Add
27589 * config.gcc: Add arch12.
27590 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27591 Default to arch12 for unknown CPU model numbers.
27592 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27593 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27594 PROCESSOR_max sanity check.
27595 * config/s390/s390-opts.h (enum processor_type): Add
27597 * config/s390/s390.c (processor_table): Add arch12.
27598 (s390_expand_builtin): Add check for B_VXE flag.
27599 (s390_issue_rate): Add PROCESSOR_ARCH12.
27600 (s390_get_sched_attrmask): Likewise.
27601 (s390_get_unit_mask): Likewise.
27602 (s390_sched_score): Enable z13 scheduling for arch12.
27603 (s390_sched_reorder): Likewise.
27604 (s390_sched_variable_issue): Likewise.
27605 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27607 (s390_tune_attr): Use z13 scheduling also for arch12.
27608 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27609 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27610 (TARGET_VXE_P): New macros.
27611 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27612 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27613 * config/s390/s390.opt: Add arch12 as processor_type.
27615 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27617 * config/s390/s390.md
27618 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27619 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27620 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27622 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27623 Rename expanders to ...
27625 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27626 ("fixuns_truncdddi2_emu"): ... these.
27628 ("fixuns_trunc<mode>si2_emu"): New expander.
27630 ("*fixuns_truncdfdi2_z13"): Rename to ...
27631 ("*fixuns_truncdfdi2_vx"): ... this.
27633 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27635 * config/s390/2964.md: Remove the single element vector compare
27636 instructions which are no longer used.
27637 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27639 (s390_canonicalize_comparison): Remove handling of DFmode
27641 (s390_expand_vec_compare_scalar): Remove function.
27642 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27643 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27645 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27647 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27649 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27651 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27652 will anyway by matched by mov<mode>_64dfp.
27654 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27656 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27657 vlef/vstef. Add missing operand to vleif.
27659 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27661 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27662 pair for all vector types with 64 bit elements.
27663 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27664 * config/s390/vector.md (V_HW_64): ... here.
27665 (V_128_NOSINGLE): New mode iterator.
27666 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27667 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27668 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27669 ("*vec_load_pairv2di"): Change to ...
27670 ("*vec_load_pair<mode>"): ... this one.
27672 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27674 * config/s390/constraints.md: Add comments.
27675 (jKK): Reject element sizes > 8 bytes.
27676 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27678 * config/s390/s390.md: Add the s_operand checks formerly in
27679 s390_split_ok_p to various splitters where they are still
27681 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27682 for 128 bit vectors. Plus two splitters.
27684 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27686 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27689 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27692 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27693 error if the boundary argument is not constant.
27695 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27697 PR rtl-optimization/80112
27698 * loop-doloop.c (doloop_condition_get): Don't check condition
27699 if cmp isn't SET with IF_THEN_ELSE src.
27701 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27703 PR tree-optimization/80158
27704 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27705 replacing a candidate statement, also replace it for the
27706 candidate's alternate interpretation.
27707 (replace_rhs_if_not_dup): Likewise.
27708 (replace_one_candidate): Likewise.
27710 2017-03-24 Richard Biener <rguenther@suse.de>
27712 PR tree-optimization/80167
27713 * graphite-isl-ast-to-gimple.c
27714 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27716 (translate_isl_ast_to_gimple::get_rename): Likewise.
27718 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27720 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27721 handling of certain combinations of target options, including the
27722 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27723 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27725 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27728 * config/arm/arm.md (*load_multiple): Add reload_completed to
27729 matching condition.
27731 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27732 Richard Biener <rguenth@suse.de>
27734 PR tree-optimization/79908
27735 PR tree-optimization/80136
27736 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27737 been cast away, gimplify_and_add suffices.
27739 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27741 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27743 2017-03-23 Richard Biener <rguenther@suse.de>
27745 PR tree-optimization/80032
27746 * gimplify.c (gimple_push_cleanup): Forced unconditional
27747 cleanups still have to go to the conditional_cleanups
27750 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27752 PR tree-optimization/80072
27753 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27755 (next_operand_entry_id): Change type to unsigned int.
27756 (sort_by_operand_rank): Make sure to return the right return value
27757 even if unsigned fields are bigger than INT_MAX.
27758 (struct oecount): Change cnt and id type to unsigned int.
27759 (oecount_hasher::equal): Formatting fix.
27760 (oecount_cmp): Make sure to return the right return value
27761 even if unsigned fields are bigger than INT_MAX.
27762 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27765 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27766 TREE_READONLY on result if writing it more than once.
27769 * doc/invoke.texi (-fsanitize=thread): Document that with
27770 -fnon-call-exceptions atomics are not able to throw
27774 * tsan.c: Include tree-eh.h.
27775 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27776 maybe_clean_or_replace_eh_stmt where needed.
27777 (instrument_memory_accesses): Add cfg_changed argument.
27778 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27779 if it returned true.
27780 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27782 PR rtl-optimization/63191
27783 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27784 wrapper function, moved the whole old content into ...
27785 (ix86_delegitimize_address_1): ... this. New inline function.
27786 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27787 true as last argument instead of ix86_delegitimize_address.
27789 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27791 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27792 cortexa57_branch_cost.
27794 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27796 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27798 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27801 * doc/md.texi (Constraints): Document wA constraint.
27802 * config/rs6000/constraints.md (wA): New.
27803 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27804 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27805 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27806 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27808 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27811 * gimplify.c (is_oacc_declared): New function.
27812 (oacc_default_clause): Use it to set default flags for acc declared
27813 variables inside parallel regions.
27814 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27815 declared variables.
27816 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27817 declare attribute to any decl as necessary.
27819 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27822 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27823 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27824 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27825 (arm_arch_lpae): This.
27826 * config/arm/arm.c (arm_arch7ve): Rename into ...
27827 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27828 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27831 2017-03-22 Martin Liska <mliska@suse.cz>
27834 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27835 error message instead of an ICE.
27837 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27839 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27841 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27843 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27845 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27848 2017-03-21 Martin Sebor <msebor@redhat.com>
27850 * doc/extend.texi: Use "cannot" instead of "can't."
27851 * doc/hostconfig.texi: Same.
27852 * doc/install.texi: Same.
27853 * doc/invoke.texi: Same.
27854 * doc/loop.texi: Same.
27855 * doc/md.texi: Same.
27856 * doc/objc.texi: Same.
27857 * doc/rtl.texi: Same.
27858 * doc/tm.texi: Same.
27859 * doc/tm.texi.in: Same.
27860 * doc/trouble.texi: Same.
27862 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27865 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27866 (collect_checksum_attributes): Set it.
27867 (die_checksum_ordered): Use it.
27869 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27871 PR tree-optimization/79908
27872 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27873 change: For a VA_ARG whose LHS has been cast away, use
27874 force_gimple_operand to construct the side effects.
27876 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27878 PR translation/80001
27879 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27880 more amenable to translation.
27881 (oacc_loop_auto_partitions): Likewise.
27883 2017-03-21 Marek Polacek <polacek@redhat.com>
27884 Martin Sebor <msebor@redhat.com>
27886 PR tree-optimization/80109
27887 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27888 on INTEGRAL_TYPE_P.
27890 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27891 Segher Boessenkool <segher@kernel.crashing.org>
27894 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27895 check reg_used_between_p between insn and one of succ or succ2
27896 depending on if succ is artificial insn not inserted into insn
27899 2017-03-21 Martin Liska <mliska@suse.cz>
27901 PR gcov-profile/80081
27902 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27903 * doc/gcc.texi: Include gcov-dump stuff.
27904 * doc/gcov-dump.texi: New file.
27906 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27908 PR rtl-optimization/79150
27909 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27910 conditional jump, if the jump is the last insn of the loop.
27912 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27913 Richard Biener <rguenth@suse.de>
27915 PR tree-optimization/79908
27916 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27917 been cast away, use force_gimple_operand to construct the side
27920 2017-03-21 Martin Liska <mliska@suse.cz>
27922 PR libfortran/79956
27923 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27926 2017-03-21 Brad Spengler <spender@grsecurity.net>
27929 * plugin.c (htab_hash_plugin): New function.
27930 (add_new_plugin): Use it and adjust.
27931 (parse_plugin_arg_opt): Adjust.
27932 (init_one_plugin): Likewise.
27934 2017-03-21 Richard Biener <rguenther@suse.de>
27936 PR tree-optimization/80032
27937 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27938 if set force the cleanup to happen unconditionally.
27939 (gimplify_target_expr): Push inserted clobbers with force_uncond
27940 to avoid them being removed by control-dependent DCE.
27942 2017-03-21 Richard Biener <rguenther@suse.de>
27944 PR tree-optimization/80122
27945 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27946 va_arg_pack_len when the inlined call stmt requires pack
27948 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27950 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27953 * tsan.c (instrument_builtin_call): If the memory model argument
27954 is not a constant, assume it is valid.
27957 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27960 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27962 PR rtl-optimization/79910
27963 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27964 if its dest is used by an insn before I2 (other than the combined
27965 insns themselves, which are properly handled already).
27967 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27970 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27972 * combine.c (record_used_regs): New static function.
27973 (try_combine): Handle situations where there is an additional
27974 instruction between I2 and I3 which needs to have a LOG_LINK
27978 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27980 * combine.c (try_combine): Delete redundant i1 test. Call
27981 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27983 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27986 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27987 alternatives 13/14.
27989 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27991 PR tree-optimization/80054
27992 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27993 the optimization if a PHI or any of its arguments is not dominated
27994 by the candidate's basis. Use gphi* rather than gimple* as
27996 (replace_profitable_candidates): Clean up a gimple* variable that
27997 should be a gphi* variable.
27999 2017-03-20 Martin Sebor <msebor@redhat.com>
28002 * doc/extend.texi (attribute constructor): Document present limitation.
28004 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28007 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28008 __POWER9_VECTOR__ #ifdef control, change template definition to
28009 use Power9-specific built-in function.
28010 (vec_any_eq): Likewise.
28011 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28012 to control outcomes from this test.
28013 (vector_ae_<mode>p): For VEC_F modes, likewise.
28015 2017-03-20 Ian Lance Taylor <iant@google.com>
28017 * config/i386/i386.c (ix86_function_regparm): Save an extra
28018 register for -fsplit-stack with DECL_STATIC_CHAIN.
28020 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28023 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28024 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28026 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28028 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28030 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28033 2017-03-20 Marek Polacek <polacek@redhat.com>
28036 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28038 2017-03-20 Richard Biener <rguenther@suse.de>
28040 PR tree-optimization/80113
28041 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28042 allocate extra SSA name for PHI def.
28043 (add_close_phis_to_outer_loops): Likewise.
28044 (add_close_phis_to_merge_points): Likewise.
28045 (copy_loop_close_phi_args): Likewise.
28046 (copy_cond_phi_nodes): Likewise.
28048 2017-03-20 Martin Liska <mliska@suse.cz>
28050 PR middle-end/79753
28051 * tree-chkp.c (chkp_build_returned_bound): Do not build
28052 returned bounds for a LHS that's not a BOUNDED_P type.
28054 2017-03-20 Martin Liska <mliska@suse.cz>
28058 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28059 COMPLEX_CST and VECTOR_CST.
28061 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28064 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28065 target operand. A new splitter adds the clobber statement in case
28066 the target operand is dead anyway.
28068 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28070 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28071 to age-old versions of binutils and glibc.
28073 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28075 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28077 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28079 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28081 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28083 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28084 requirement for binutils 2.13.
28086 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28088 * combine.c (try_combine): Delete redundant i1 test. Call
28089 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28091 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28093 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28094 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28096 <riscv64-*-elf>: Re-arrange section
28097 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28098 <riscv32-*-linux>: Likewise.
28099 <riscv64-*-elf>: Likewise
28100 <riscv64-*-linux>: Likewise.
28102 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28105 * aarch64.opt(verbose-cost-dump): Fix typo.
28107 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28110 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28111 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28113 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28115 * reload.c (find_reloads): When reloading a nonoffsettable address,
28116 use RELOAD_OTHER for it and its address reloads.
28118 PR rtl-optimization/79910
28119 * combine.c (record_used_regs): New static function.
28120 (try_combine): Handle situations where there is an additional
28121 instruction between I2 and I3 which needs to have a LOG_LINK
28124 2017-03-17 Jeff Law <law@redhat.com>
28126 PR tree-optimization/71437
28127 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28128 conditional in the hash table first.
28129 (vrp_dom_walker::before_dom_children): Extract condition from
28130 ASSERT_EXPR. Record condition, its inverion and any implied
28131 conditions as well.
28133 2017-03-17 Marek Polacek <polacek@redhat.com>
28134 Markus Trippelsdorf <markus@trippelsdorf.de>
28136 PR tree-optimization/80079
28137 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28140 2017-03-17 Richard Biener <rguenther@suse.de>
28142 PR middle-end/80075
28143 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28144 Properly verify the LHS before the RHS possibly claims to be
28146 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28149 2017-03-17 Martin Jambor <mjambor@suse.cz>
28151 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28152 (List of -O2 options): Likewise.
28153 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28156 2017-03-17 Tom de Vries <tom@codesourcery.com>
28158 * gcov-dump.c (print_usage): Print bug_report_url.
28160 2017-03-17 Richard Biener <rguenther@suse.de>
28162 PR middle-end/80050
28163 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28164 (parser::peek): Likewise.
28166 2017-03-17 Richard Biener <rguenther@suse.de>
28168 PR tree-optimization/80048
28169 * sese.c (free_sese_info): Properly release rename_map and
28170 copied_bb_map elements.
28172 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28174 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28175 Add linked-list forward and backlinks. Insert on
28176 construction, remove on destruction.
28177 (class pass_store_merging): Add m_stores_head field.
28178 (pass_store_merging::terminate_and_process_all_chains):
28179 Iterate over m_stores_head list.
28180 (pass_store_merging::terminate_all_aliasing_chains):
28182 (pass_store_merging::execute): Check for debug stmts first.
28183 Push new chains onto the m_stores_head stack.
28185 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28188 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28189 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28190 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28192 2017-03-16 Jeff Law <law@redhat.com>
28194 PR tree-optimization/71437
28195 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28196 member function. Implementation moved into after_dom_children
28197 member function and into the threader's thread_outgoing_edges
28199 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28200 some code into new thread_outgoing_edges.
28201 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28202 definition. Simplify marker handling (do it here). Assume we always
28203 have the available expression and the const/copies tables.
28204 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28206 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28207 * tree-vrp.c (equiv_stack): No longer file scoped.
28208 (vrp_dom_walker): New class.
28209 (vrp_dom_walker::before_dom_children): New member function.
28210 (vrp_dom_walker::after_dom_children): Likewise.
28211 (identify_jump_threads): Setup domwalker. Use it rather than
28212 walking edges in a random order by hand. Simplify setup/finalization.
28213 (finalize_jump_threads): Remove.
28214 (vrp_finalize): Do not call identify_jump_threads here.
28215 (execute_vrp): Do it here instead and call thread_through_all_blocks
28218 PR tree-optimization/71437
28219 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28221 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28223 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28224 (dom_opt_dom_walker::thread_across_edge): Remove
28225 handle_dominating_asserts argument. All callers changed.
28226 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28227 changes. Remove calls to lhs_of_dominating_assert. Other
28228 uses of handle_dominating_asserts turn into unconditional code
28229 (simplify_control_stmt_condition_1): Likewise.
28230 (simplify_control_stmt_condition): Likewise.
28231 (thread_through_normal_block, thread_across_edge): Likewise.
28232 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28233 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28234 object if it is not an SSA_NAME.
28235 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28236 before calling into the VRP specific simplifiers.
28237 (identify_jump_threads): Remove handle_dominating_asserts
28240 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28243 * tree-diagnostic.c (default_tree_printer): No longer static.
28244 * tree-diagnostic.h (default_tree_printer): New prototype.
28246 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28248 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28249 Change ins into fmov.
28251 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28253 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28254 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28255 Use h_con constraint for operand 1.
28256 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28257 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28259 2017-03-15 Jeff Law <law@redhat.com>
28261 PR tree-optimization/71437
28262 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28263 (record_temporary_equivalences): Use it.
28265 PR tree-optimization/71437
28266 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28267 tree-ssa-scopedtables.
28268 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28269 (record_conditions, record_cond, vuse_eq): Likewise.
28270 (record_edge_info): Adjust to API tweak of record_conditions.
28271 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28272 (record_temporary_equivalences, optimize_stmt): Likewise.
28273 (eliminate_redundant_computations): Likewise.
28274 (record_equivalences_from_stmt): Likewise.
28275 * tree-ssa-scopedtables.c: Include options.h and params.h.
28276 (vuse_eq): New function, moved from tree-ssa-dom.c
28277 (build_and_record_new_cond): Likewise.
28278 (record_conditions): Likewise. Accept vector of conditions rather
28279 than edge_equivalence structure for first argument.
28280 for the first argument.
28281 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28282 from tree-ssa-dom.c.
28283 (avail_exprs_stack::record_cond): Likewise.
28284 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28285 from tree-ssa-dom.c.
28286 (avail_exprs_stack): Add new member functions lookup_avail_expr
28288 (record_conditions): Declare.
28290 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28293 * lra-constraints.c (process_alt_operands): Increase reject for
28294 reloading an input/output operand.
28296 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28299 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28300 insns to convert from signed/unsigned char/short to IEEE 128-bit
28302 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28304 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28307 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28308 subreg of inner mode for values already in registers.
28310 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28312 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28313 iteration reg is used after the loop.
28315 2017-03-14 Martin Sebor <msebor@redhat.com>
28317 PR tree-optimization/79800
28318 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28319 precision in negative-positive range.
28320 (format_floating): Call non-const overload with adjusted precision.
28322 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28325 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28328 2017-03-14 Martin Sebor <msebor@redhat.com>
28330 PR middle-end/80020
28331 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28332 * builtins.def (aligned_alloc): Use it.
28335 * Makefile.in (GTFILES): Add calls.c.
28336 * calls.c: Include "gt-calls.h".
28338 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28340 PR rtl-optimization/79728
28341 * regs.h (struct target_regs): New field
28342 x_contains_allocatable_regs_of_mode.
28343 (contains_allocatable_regs_of_mode): New macro.
28344 * reginfo.c (init_reg_sets_1): Initialize it, and change
28345 contains_reg_of_mode so it includes global regs as well.
28346 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28347 rather than contains_regs_of_mode.
28349 2017-03-14 Martin Liska <mliska@suse.cz>
28351 * doc/invoke.texi: Document options that can't be combined with
28352 -fcheck-pointer-bounds.
28354 2017-03-14 Martin Liska <mliska@suse.cz>
28356 PR middle-end/79831
28357 * doc/invoke.texi (-Wchkp): Document the option.
28359 2017-03-14 Martin Liska <mliska@suse.cz>
28361 * Makefile.in: Install gcov-dump.
28363 2017-03-14 Martin Liska <mliska@suse.cz>
28365 * multiple_target.c (expand_target_clones): Bail out for
28366 an invalid attribute.
28368 2017-03-14 Richard Biener <rguenther@suse.de>
28370 * alias.c (struct alias_set_entry): Pack properly.
28371 * cfgloop.h (struct loop): Likewise.
28372 * cse.c (struct set): Likewise.
28373 * ipa-utils.c (struct searchc_env): Likewise.
28374 * loop-invariant.c (struct invariant): Likewise.
28375 * lra-remat.c (struct cand): Likewise.
28376 * recog.c (struct change_t): Likewise.
28377 * rtl.h (struct address_info): Likewise.
28378 * symbol-summary.h (function_summary): Likewise.
28379 * tree-loop-distribution.c (struct partition): Likewise.
28380 * tree-object-size.c (struct object_size_info): Likewise.
28381 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28382 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28383 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28384 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28385 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28386 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28387 (struct _stmt_vec_info): Likewise.
28389 2017-03-14 Martin Liska <mliska@suse.cz>
28392 * multiple_target.c (create_dispatcher_calls): Check that
28393 a target can create a function dispatcher.
28395 2017-03-14 Martin Liska <mliska@suse.cz>
28398 * multiple_target.c (expand_target_clones): Drop local.local
28399 flag for default implementation.
28401 2017-03-14 Richard Biener <rguenther@suse.de>
28403 PR tree-optimization/80030
28404 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28406 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28408 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28409 gcc_fallthrough() instead of __attribute__((fallthrough));
28411 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28413 * doc/gcc.texi: Remove "up" link to (DIR).
28414 * doc/gccint.texi: Ditto.
28416 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28418 * doc/install.texi (Specific) <avr>: Remove reference to
28421 2017-03-13 Jeff Law <law@redhat.com>
28423 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28424 attribute rather than comments.
28426 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28427 match_scratch operand is highest.
28429 2017-03-13 Martin Liska <mliska@suse.cz>
28431 PR middle-end/78339
28432 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28433 is a CHKP clone, use original declaration.
28435 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28437 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28438 (arc_conditional_register_usage): Use a different allocation order
28439 when optimizing for size.
28440 * common/config/arc/arc-common.c (arc_option_optimization_table):
28441 Section anchors default on when optimizing for size.
28443 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28445 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28447 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28449 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28450 * config/arc/arc.md (cpu_facility): Add cd variant.
28451 (*movqi_insn): Add code density variant.
28452 (*movhi_insn): Likewise.
28453 (*movqi_insn): Likewise.
28454 (*addsi3_mixed): Likewise.
28455 (subsi3_insn): Likewise.
28457 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28459 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28461 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28463 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28464 expressions with MINUS and UNARY ops.
28466 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28469 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28471 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28472 between vec_select and vector argument.
28473 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28474 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28475 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28476 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28477 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28478 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28480 2017-03-13 Richard Biener <rguenther@suse.de>
28483 * params.def (vect-max-peeling-for-alignment): Fix typo.
28485 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28487 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28488 issue that only occurred with binutils below 2.18.
28490 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28492 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28493 refer to binutils 2.11/2.12 minimum.
28495 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28497 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28498 ftp.kernel.org and simplify binutils requirement.
28500 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28502 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28504 (Optimize Options): Ditto. Also remove redundancy.
28506 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28508 PR translation/79848
28509 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28511 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28512 to G_ to avoid double translation.
28514 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28516 PR translation/79923
28517 * auto-profile.c (get_combined_location): Convert leading
28518 character of diagnostics to lower case and remove trailing period.
28519 (read_profile): Likewise for various diagnostics.
28520 * config/arm/arm.c (arm_option_override): Remove trailing period
28521 from various diagnostics.
28522 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28523 (msp430_expand_delay_cycles): Likewise.
28525 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28528 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28529 full command-line argument, rather than just "str".
28530 (aarch64_validate_march): Likewise.
28531 (aarch64_validate_mtune): Likewise.
28533 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28535 PR rtl-optimization/78911
28536 * lra-assigns.c (must_not_spill_p): New function.
28537 (spill_for): Use it.
28539 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28541 PR tree-optimization/79981
28542 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28543 ATOMIC_COMPARE_EXCHANGE ifn result.
28544 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28545 IFN_ATOMIC_COMPARE_EXCHANGE.
28547 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28550 * opts.c (parse_sanitizer_options): Add missing question mark to
28551 "did you mean" message.
28553 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28555 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28557 (VMULEUH_UNS): Likewise.
28558 (VMULOUB_UNS): Likewise.
28559 (VMULOUH_UNS): Likewise.
28560 * config/rs6000/rs6000.c (builtin_function_type): Remove
28561 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28563 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28566 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28567 x with result of extra_parsing_for_operand_code_0.
28568 (function_reader::extra_parsing_for_operand_code_0): Convert
28569 return type from void to rtx, returning x. When reading
28570 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28571 larger size containing struct block_symbol.
28573 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28575 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28576 -mfloat128-hardware without -m64.
28578 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28581 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28582 entries to the case statement that marks unsigned arguments to
28583 overloaded functions.
28585 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28587 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28588 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28590 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28593 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28594 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28596 2017-03-10 Martin Liska <mliska@suse.cz>
28600 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28601 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28604 2017-03-10 Olivier Hainque <hainque@adacore.com>
28606 * tree-switch-conversion (array_value_type): Start by resetting
28607 candidate type to it's main variant.
28609 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28611 PR rtl-optimization/79909
28612 * combine.c (try_combine): Use simplify_replace_rtx on individual
28613 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28614 of the whole CALL_INSN_FUNCTION_USAGE.
28616 PR tree-optimization/79972
28617 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28618 get_range_info on SSA_NAMEs. Formatting fixes.
28620 2017-03-10 Richard Biener <rguenther@suse.de>
28621 Jakub Jelinek <jakub@redhat.com>
28623 PR tree-optimization/77975
28624 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28625 edge to be constant.
28626 (get_val_for): For constant x return it. Formatting fix.
28627 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28628 would use the same bases as the current one.
28630 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28632 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28633 instead of vec_select for V1TImode.
28634 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28636 (VSX_LE_128): Add V1TI to this mode iterator.
28637 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28638 (*vsx_le_perm_store_<mode>): Likewise.
28639 (pre-reload splitter for VSX stores): Likewise.
28640 (post-reload splitter for VSX stores): Likewise.
28641 (*vsx_xxpermdi2_le_<mode>): Likewise.
28642 (*vsx_lxvd2x2_le_<mode>): Likewise.
28643 (*vsx_stxvd2x2_le_<mode>): Likewise.
28645 2017-03-09 Michael Eager <eager@eagercon.com>
28647 Correct failures with --enable-checking=yes,rtl.
28649 * config/microblaze/microblaze.c (microblaze_expand_shift):
28650 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28651 test for const0_rtx.
28652 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28653 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28655 2017-03-09 Richard Biener <rguenther@suse.de>
28657 PR tree-optimization/79977
28658 * graphite-scop-detection.c (scop_detection::merge_sese):
28659 Handle the case of extra exits to blocks dominating the entry.
28661 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28663 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28666 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28668 PR rtl-optimization/79949
28669 * lra-constraints.c (process_alt_operands): Check memory when
28670 trying to predict a cycle. Print about the overall increase.
28672 2017-03-09 Richard Biener <rguenther@suse.de>
28674 PR middle-end/79971
28675 * gimple-expr.c (useless_type_conversion_p): Preserve
28676 TYPE_SATURATING for fixed-point types.
28678 2017-03-09 Richard Biener <rguenther@suse.de>
28681 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28682 alignment of BLKmode params.
28684 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28687 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28688 (VALL_NO_V2Q): Likewise.
28690 * config/aarch64/aarch64-simd.md
28691 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28693 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28694 VALL_NO_V2Q mode iterator.
28695 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28697 2017-03-09 Martin Liska <mliska@suse.cz>
28699 PR tree-optimization/79631
28700 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28701 tree_int_cst_sign_bit just for INTEGER constants.
28703 2017-03-09 Martin Liska <mliska@suse.cz>
28707 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28710 2017-03-09 Marek Polacek <polacek@redhat.com>
28713 * tree.c (inchash::add_expr): Handle TREE_VEC.
28715 2017-03-09 Martin Liska <mliska@suse.cz>
28718 (chkp_narrow_size_and_offset): New function.
28719 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28720 (void chkp_parse_bit_field_ref): New function.
28721 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28722 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28724 2017-03-09 Martin Liska <mliska@suse.cz>
28727 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28728 (chkp_find_bounds_1): Remove gcc_unreachable.
28730 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28733 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28734 BUILT_IN_SYNC*, determine the access type from the size suffix and
28735 always build a MEM_REF with that type. Handle forgotten
28736 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28739 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28740 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28741 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28742 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28743 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28744 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28745 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28746 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28747 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28748 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28749 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28750 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28751 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28752 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28753 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28754 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28755 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28756 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28757 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28758 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28759 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28760 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28761 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28762 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28763 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28764 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28765 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28766 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28767 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28768 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28769 definitions outside of __OPTIMIZE__ guarded section.
28772 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28773 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28774 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28775 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28778 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28780 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28781 ("vfenez<mode>"): Add missing constraints.
28783 2017-03-08 Martin Sebor <msebor@redhat.com>
28786 * config/nds32/nds32.c (nds32_option_override):
28787 Fix misspelled diagnostic.
28789 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28792 * gimplify.c (gimplify_omp_for): Replace index var in outer
28793 taskloop statement with an artificial variable and add
28794 OMP_CLAUSE_PRIVATE clause for it.
28796 2017-03-08 Richard Biener <rguenther@suse.de>
28798 PR tree-optimization/79955
28799 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28800 for accesses that are completely outside of the variable.
28802 2017-03-08 Andrew Haley <aph@redhat.com>
28804 PR tree-optimization/79943
28805 * tree-ssa-loop-split.c (compute_new_first_bound): When
28806 calculating the new upper bound, (END-BEG) should be added, not
28809 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28811 * config/avr/avr.md (setmemhi): Make sure match_dup
28812 operand number comes before match_scratch.
28814 2017-03-08 Richard Biener <rguenther@suse.de>
28816 PR tree-optimization/79920
28817 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28818 with ncopies == 1 to ...
28819 (vect_transform_slp_perm_load): ... here. Properly compute
28820 all element loads by iterating VF times over the group. Do
28821 not handle ncopies (computed in a broken way) in
28822 vect_create_mask_and_perm.
28824 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28827 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28828 is a uniform vector, use uniform_vector_p return value instead of
28829 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28831 2017-03-07 Marek Polacek <polacek@redhat.com>
28833 PR middle-end/79809
28834 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28835 (alloca_call_type): Likewise.
28837 2017-03-07 Martin Liska <mliska@suse.cz>
28839 * gcov.c (process_args): Put comment to correct location.
28841 2017-03-07 Martin Liska <mliska@suse.cz>
28843 PR middle-end/68270
28844 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28845 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28846 (chkp_narrow_bounds_for_field): Likewise.
28847 (chkp_parse_array_and_component_ref): Pass one more argument to
28850 2017-03-07 Richard Biener <rguenther@suse.de>
28852 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28855 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28857 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28858 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28860 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28863 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28864 to end of description.
28865 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28867 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28869 PR rtl-optimization/79901
28870 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28872 (*avx512f_<code><mode>3<mask_name>): ... this.
28873 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28874 iterator instead of VI8_AVX2_AVX512BW.
28876 PR rtl-optimization/79901
28877 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28878 min/max expander, expand it using expand_vec_cond_expr.
28881 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28884 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28887 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28888 to void * for PCH reasons.
28889 * dwarf2out.c (output_loc_operands, output_die): Cast
28890 v.val_vec.array to unsigned char *.
28892 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28895 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28898 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28900 PR rtl-optimization/79571
28901 * lra-constraints.c (process_alt_operands): Calculate static
28902 reject and subtract it from overall when only addresses will be
28905 2017-03-06 Julia Koval <julia.koval@intel.com>
28908 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28909 incoming stack boundary to 128 for 64-bit targets.
28911 2017-03-06 Richard Biener <rguenther@suse.de>
28913 PR tree-optimization/79894
28914 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28915 to NULL after folding it.
28917 2017-03-06 Richard Biener <rguenther@suse.de>
28919 PR tree-optimization/79824
28920 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28921 check disabling peeling for gaps.
28923 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28925 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28926 attributes): Document gettimeofday.
28928 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28930 * config/s390/s390.c (s390_option_override_internal): Set
28931 PARAM_MIN_VECT_LOOP_BOUND
28933 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28935 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28936 * config/s390/s390.md: Likewise.
28938 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28941 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28942 (<avx2_avx512>_perm<mode>): Rename to ...
28943 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28945 (<avx512>_perm<mode>_mask): Rename to ...
28946 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28948 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28949 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28950 instead of VI8F_256_512.
28951 (avx512f_perm<mode>): New define_expand.
28952 (avx512f_perm<mode>_mask): Likewise.
28953 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28954 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28956 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28958 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28959 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28961 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28963 2017-03-06 Martin Liska <mliska@suse.cz>
28966 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28967 when having a SSA NAME w/o VAR_DECL assigned to it.
28969 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28971 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28972 msa_dpsub_<su>_d): Fix MODE for vec_select.
28974 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28976 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28978 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28980 2017-03-06 Richard Biener <rguenther@suse.de>
28982 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28983 * plugin.c (register_plugin_info): Likewise.
28984 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28986 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28988 * config/i386/sse.md (sse_storehps, sse_storelps,
28989 avx_<castmode><avxsizesuffix>_<castmode>,
28990 avx512f_<castmode><avxsizesuffix>_<castmode>,
28991 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28992 in condition that at least one operand is not a MEM.
28994 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28996 PR middle-end/79805
28997 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28998 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29000 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29001 gimple_call_nothrow_p flag based on whether original builtin can throw.
29002 If it can, emit following stmts on the fallthrough edge.
29003 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29004 don't create new bb if inserting just debug stmts on the edge, try to
29005 insert them on the fallthru bb or just reset debug stmts.
29007 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29010 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29011 restore recog_data (including the operand rtxes inside it) around
29012 the call to get_insn_template.
29014 2017-03-03 Martin Sebor <msebor@redhat.com>
29016 PR tree-optimization/79699
29017 * context.c (context::~context): Free MPFR caches to avoid
29018 a memory leak on program exit.
29020 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29022 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29023 Use wide_int::ulow () instead of .elt (0).
29025 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29027 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29028 (*pushxf): Limit oF constraint to 32bit targets and add oC
29029 constraint for 64bit targets.
29030 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29031 (*pushdf): Change rmF constraint to rmC.
29033 2017-03-03 Martin Liska <mliska@suse.cz>
29035 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29036 Remove unused variable.
29038 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29041 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29042 is a memory operand, increase num_memory.
29043 (ix86_expand_args_builtin): Likewise.
29045 2017-03-03 Jan Hubicka <jh@suse.cz>
29048 * ipa-devirt.c (maybe_record_node): Properly handle
29049 __cxa_pure_virtual visibility.
29051 2017-03-03 Martin Liska <mliska@suse.cz>
29053 PR tree-optimization/79803
29054 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29056 (pass_loop_prefetch::execute): Disabled optimization if an
29057 assumption about L1 cache size is not met.
29059 2017-03-03 Martin Liska <mliska@suse.cz>
29061 PR rtl-optimization/79574
29062 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29063 (hash_scan_set): Likewise.
29064 (dump_hash_table): Likewise.
29065 (hoist_code): Likewise.
29067 2017-03-03 Richard Biener <rguenther@suse.de>
29069 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29070 in place of uhigh ().
29071 (fixed_convert_from_real): Likewise.
29073 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29076 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29078 2017-03-03 Richard Biener <rguenther@suse.de>
29080 PR middle-end/79818
29081 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29082 TYPE_OVERFLOW_UNDEFINED check.
29084 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29086 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29088 (vector_ae_<mode>_p): Likewise.
29089 (vector_nez_<mode>_p): Likewise.
29090 (vector_ne_v2di_p): Likewise.
29091 (vector_ae_v2di_p): Likewise.
29092 (vector_ne_<mode>_p): Likewise.
29093 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29095 (vsx_tsqrt<mode>2_fe): Likewise.
29097 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29100 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29102 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29104 PR rtl-optimization/79780
29105 * cprop.c (one_cprop_pass): When second and further conditional trap
29106 in a single basic block is turned into an unconditional trap, turn it
29107 into a deleted note to avoid RTL verification failures.
29109 2017-03-02 Richard Biener <rguenther@suse.de>
29111 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29113 2017-03-02 Richard Biener <rguenther@suse.de>
29115 PR tree-optimization/79345
29117 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29118 param and abort the walk, returning -1 if it is hit.
29119 (walk_aliased_vdefs): Take a limit param and pass it on.
29120 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29121 defaulting to 0 and return a signed int.
29122 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29123 (check_defs): New helper.
29124 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29125 about uninitialized memory.
29126 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29127 bogus uninitialized warning.
29128 (fixed_convert_from_real): Likewise.
29130 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29132 PR tree-optimization/66768
29133 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29134 iv_use if base object can't be determined.
29136 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29138 PR tree-optimization/79345
29139 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29140 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29141 (get_pattern_stats): Initialize it.
29142 * genemit.c (gen_expand): Verify match_scratch numbers come after
29143 match_operand/match_dup numbers.
29144 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29145 match_scratch numbers.
29146 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29148 * config/s390/s390.md (trunctdsd2): Likewise.
29150 2017-03-02 Richard Biener <rguenther@suse.de>
29152 * wide-int.h (wide_int_storage::operator=): Implement in terms
29155 2017-03-02 Richard Biener <rguenther@suse.de>
29157 PR tree-optimization/79777
29158 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29159 the to insert expression to sth existing.
29161 2017-03-01 Martin Sebor <msebor@redhat.com>
29163 PR middle-end/79692
29164 * gimple-ssa-sprintf.c
29165 (directive::known_width_and_precision): New function.
29166 (format_integer): Use it.
29167 (get_mpfr_format_length): Consider the full range of precision
29168 when computing %g output with the # flag. Set the likely byte
29169 count to 3 rather than 1 when precision is indeterminate.
29170 (format_floating): Correct the lower bound of precision.
29172 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29174 * doc/invoke.texi: Document default code model for 64-bit Linux.
29176 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29179 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29180 udiv rather than div since input pattern is unsigned.
29182 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29184 * config/i386/i386.c (print_reg): Warn for values of
29185 unsupported size in integer register.
29187 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29190 * config/rs6000/predicates.md (current_file_function_operand): Do
29191 not allow self calls to be local if the function is replaceable.
29193 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29196 * config/rs6000/altivec.h (vec_ctz and others): Change the
29197 preprocessor macro that controls conditional compilation from
29198 _ARCH_PWR9 to __POWER9_VECTOR__.
29199 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29200 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29201 control (instead of _ARCH_PWR9 control) so that template
29202 definition uses power9-specific function.
29203 (vec_any_eq): Likewise.
29204 (vec_all_ne): Change macro definition to use a power9-specific
29205 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29206 _ARCH_PWR9 control).
29207 (vec_any_eq) Likewise.
29208 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29209 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29210 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29211 support for xvcmpnedp instruction.
29212 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29213 macro expansion so that Power9 implementation of vec_all_ne does
29214 not use the AltiVec predicate framework.
29215 (VCMPNEH_P): Likewise.
29216 (VCMPNEW_P): Likewise.
29217 (VCMPNED_P): Likewise.
29218 (VCMPNEFP_P): Likewise.
29219 (VCMPNEDP_P): Likewise.
29220 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29221 implementation of vec_any_eq to not use AltiVec predicate
29223 (VCMPAEH_P): Likewise.
29224 (VCMPAEW_P): Likewise.
29225 (VCMPAED_P): Likewise.
29226 (VCMPAEFP_P): Likewise.
29227 (VCMPAEDP_P): Likewise.
29228 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29229 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29230 not use the AltiVec predicate framework.
29231 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29232 of vec_any_eq to not use AltiVec predicate framework.
29233 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29234 support for predefined __POWER9_VECTOR__ macro to indicate that
29235 Power9 instruction selection is enabled.
29236 (altivec_overloaded_builtins): Remove extraneous
29237 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29238 function argument types RS6000_BTI_bool_V16QI and
29239 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29240 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29241 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29242 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29243 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29244 Power9 for implementations of vec_cmpne. Change the signature for
29245 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29246 (representing vec_all_ne) to remove the previously described first
29247 argument of type RS6000_BTI_INTSI, as this was an artifact of
29248 reliance on the AltiVec predicate framework, which is no longer
29249 used in the implementation of these functions. Add
29250 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29251 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29252 since, unlike the AltiVec predicate framework implementation, we
29253 do not share function descriptors between vec_alle and vec_anyeq.
29254 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29255 set of modes that receive special treatment even when
29256 TARGET_P9_VECTOR is true. The special treatment emits code that
29257 does not depend on Power9 instructions.
29258 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29259 define_expand to not rely on AltiVec predicate framework.
29260 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29262 (vector_ne_v2di_p): Change this define_expand to not rely on
29263 AltiVec predicate framework.
29264 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29266 (vector_ne_<mode>_p): Change this define_expand to not rely on
29267 AltiVec predicate framework.
29268 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29270 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29271 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29272 define_insn pattern.
29273 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29274 define_insn pattern because the xvcmpne<VSs>. instruction is not
29276 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29277 instruction is not supported.
29279 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29281 * config/nvptx/nvptx.c: Include intl.h.
29283 2017-03-01 Martin Jambor <mjambor@suse.cz>
29286 * ipa-prop.h (ipa_bits): Removed field known.
29287 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29288 to pointers. Adjusted their comments to warn about their sharing.
29289 (ipcp_transformation_summary): Change bits to a vector of pointers.
29290 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29291 (ipa_get_ipa_bits_for_value): Declare.
29292 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29293 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29294 (ipa_bits_hash_table): Likewise.
29295 (ipa_vr_ggc_hash_traits): Likewise.
29296 (ipa_vr_hash_table): Likewise.
29297 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29298 being pointers and vr_known being removed.
29299 (ipa_set_jf_unknown): Likewise.
29300 (ipa_get_ipa_bits_for_value): New function.
29301 (ipa_set_jfunc_bits): Likewise.
29302 (ipa_get_value_range): New overloaded functions.
29303 (ipa_set_jfunc_vr): Likewise.
29304 (ipa_compute_jump_functions_for_edge): Use the above functions to
29305 construct bits and vr parts of jump functions.
29306 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29307 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29309 (ipcp_grow_transformations_if_necessary): Also allocate
29310 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29312 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29313 them. Fix too long lines.
29314 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29315 vr_known being removed.
29316 (ipa_read_jump_function): Use new setter functions to construct bits
29317 and vr parts of jump functions or set them to NULL.
29318 (write_ipcp_transformation_info): Adjust for bits being pointers.
29319 (read_ipcp_transformation_info): Likewise.
29320 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29322 Include gt-ipa-prop.h.
29323 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29325 (ipcp_store_bits_results): Likewise.
29326 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29327 Do not write to existing jump functions but use a temporary instead.
29329 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29332 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29333 attempt to use its first operand as BIT_FIELD_REF base.
29335 2017-03-01 Richard Biener <rguenther@suse.de>
29337 PR middle-end/79721
29338 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29339 interpolating formula in wrapping arithmetic.
29340 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29342 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29344 PR tree-optimization/79734
29345 * tree-vect-generic.c (expand_vector_condition): Optimize
29346 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29347 Handle VEC_COND_EXPR where comparison has different inner width from
29348 type's inner width.
29350 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29352 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29353 markup, and similar issues. Remove @opindex entries for things
29354 that aren't options. Add missing -mmpy-option entries.
29356 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29358 PR tree-optimization/79737
29359 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29360 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29361 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29362 instead of byte_size. Formatting fix.
29363 (shift_bytes_in_array_right): Formatting fix.
29365 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29368 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29369 condition on optimize for the leaf function test.
29371 2017-02-28 Martin Liska <mliska@suse.cz>
29374 * read-rtl-function.c (function_reader::handle_unknown_directive):
29375 Bail out when one uses -flto.
29377 2017-02-28 Martin Liska <mliska@suse.cz>
29379 * common.opt: Replace space with tabular for options of <number>
29381 * config/i386/i386.opt: Show <number> value for
29382 -mlarge-data-threshold.
29383 * opts.c (print_filtered_help): Do not display number in hexadecimal
29386 2017-02-28 Martin Liska <mliska@suse.cz>
29388 * common.opt: Fix --help=option -Q for options which are of
29391 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29393 * config/i386/i386.c (print_reg): Error out for values
29394 of 8-bit size in invalid integer register.
29396 2017-02-28 Martin Sebor <msebor@redhat.com>
29398 PR tree-optimization/79691
29399 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29401 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29404 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29405 gcc_unreachable with output_operand_lossage.
29407 2017-02-28 Richard Biener <rguenther@suse.de>
29409 PR tree-optimization/79740
29410 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29412 (visit_nary_op): Insert the nary into the hashtable if we
29413 pattern-matched sth.
29414 * tree-ssa-pre.c (eliminate_insert): Robustify.
29416 2017-02-28 Richard Biener <rguenther@suse.de>
29418 PR middle-end/79731
29419 * fold-const.c (decode_field_reference): Reject out-of-bound
29422 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29424 * config/i386/i386.c: Include intl.h.
29425 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29426 instead of just cond ? "..." : "...".
29427 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29428 * coverage.c (read_counts_file): Likewise.
29429 * omp-offload.c: Include intl.h.
29430 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29431 of just cond ? "..." : "...".
29432 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29433 of just cond ? "..." : "...".
29435 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29438 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29440 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29441 'tune for' CPU name.
29442 * config/arm/arm-cpu-data.h: Regenerated.
29444 2017-02-28 Richard Biener <rguenther@suse.de>
29446 PR tree-optimization/79732
29447 * tree-inline.c (expand_call_inline): Do not shadow var.
29449 2017-02-28 Richard Biener <rguenther@suse.de>
29451 PR tree-optimization/79723
29452 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29453 address-space properly.
29455 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29457 * doc/optinfo.texi (Optimization groups): Fix option used for
29459 * doc/invoke.texi (-fopt-info): Document "omp".
29460 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29461 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29462 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29463 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29464 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29466 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29468 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29469 OPTGROUP_OMP with "omp".
29471 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29474 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29475 for arithmetic shift of unsigned V2DI.
29477 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29479 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29480 arc/linux.h headers.
29481 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29482 (LINK_SPEC): Likewise.
29483 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29484 (EXTRA_SPECS): Likewise.
29485 (STARTFILE_SPEC): Likewise.
29486 (ENDFILE_SPEC): Likewise.
29487 (LIB_SPEC): Likewise.
29488 (TARGET_SDATA_DEFAULT): Likewise.
29489 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29490 (MULTILIB_DEFAULTS): Likewise.
29491 (DWARF2_UNWIND_INFO): Likewise.
29492 * config/arc/big.h: New file.
29493 * config/arc/elf.h: Likewise.
29494 * config/arc/linux.h: Likewise.
29495 * config/arc/t-uClibc: Remove.
29497 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29499 PR tree-optimization/77536
29500 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29501 (tree_transform_and_unroll_loop): Use above function to compute the
29502 estimated niter of unrolled loop and use it when scaling profile.
29503 Also use count info rather than frequency if it's non-zero.
29504 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29505 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29506 (vect_transform_loop): Call above function.
29508 2017-02-27 Richard Biener <rguenther@suse.de>
29510 PR tree-optimization/45397
29511 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29512 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29513 (visit_nary_op): Add pattern matching for CSEing sign-changed
29514 or truncated operations with wider ones.
29516 2017-02-27 Richard Biener <rguenther@suse.de>
29518 PR tree-optimization/79690
29519 * tree-vect-stmts.c (vectorizable_store): Use vector type
29520 built from the DR with address-space.
29522 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29524 * doc/invoke.texi (Optimize Options): Refine the description
29525 of asan-use-after-return.
29527 2017-02-25 Alan Modra <amodra@gmail.com>
29529 PR rtl-optimization/79584
29530 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29531 base, not ad->base_term, the reg within base. Remove assertion
29532 that ad->base == ad->base_term. Replace gen_int_mode using
29533 bogus mode with const0_rtx.
29535 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29537 PR middle-end/79396
29538 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29539 FMA_EXPR like tcc_binary or tcc_unary.
29541 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29544 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29546 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29547 (output_loc_operands): Handle DW_OP_call_ref and
29548 DW_OP_GNU_variable_value.
29549 (struct variable_value_struct): New type.
29550 (struct variable_value_hasher): Likewise.
29551 (variable_value_hash): New variable.
29552 (string_types): Remove.
29553 (copy_loc_descr): New function.
29554 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29555 (prepend_loc_descr_to_each): New function.
29556 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29557 instead of add_loc_descr_to_each if the first argument is single
29558 location list and the second has multiple.
29559 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29560 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29561 when looking for variable value which doesn't have other location info.
29562 (loc_list_from_tree): Formatting fix.
29563 (gen_array_type_die): Simplify DW_AT_string_length handling.
29564 (adjust_string_types): Remove.
29565 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29566 string_types. Call resolve_variable_values.
29567 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29568 (resolve_addr_in_expr): Likewise. Add A argument.
29569 (copy_deref_exprloc): Remove deref argument. Adjust for the
29570 original expression being DW_OP_GNU_variable_value with optionally
29571 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29572 optionally after it.
29573 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29574 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29575 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29576 (variable_value_hasher::hash, variable_value_hasher::equal): New
29578 (resolve_variable_value_in_expr, resolve_variable_value,
29579 resolve_variable_values, note_variable_value_in_expr,
29580 note_variable_value): New functions.
29581 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29584 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29587 * opts.h (handle_generated_option): Add GENERATED_P argument.
29588 * opts-common.c (handle_option): Adjust function comment.
29589 (handle_generated_option): Add GENERATED_P argument, pass it to
29591 (control_warning_option): Pass false to handle_generated_option
29593 * opts.c (maybe_default_option): Pass true to handle_generated_option
29595 * optc-gen.awk: Likewise.
29597 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29599 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29600 a REG, look at the REG it is a SUBREG of.
29601 (splitter for cmpeqsi_t): Ditto.
29603 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29605 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29606 the special USEs with the pattern of the insn, not the insn itself.
29608 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29611 * doc/invoke.texi: Document -mload-store-pairs.
29613 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29614 Sandra Loosemore <sandra@codesourcery.com>
29616 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29617 argument isn't a CONST_INT.
29618 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29619 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29620 (nios2_validate_compare): Bypass alternate compare logic if *op2
29621 is not a CONST_INT.
29622 (ldstwm_operation_p): Return false if first_base is not a REG or
29623 if first_offset is not a CONST_INT.
29625 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29627 * config/cris/cris.md: Use correct operand in a define_peephole2.
29629 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29631 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29633 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29635 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29636 this_insn if it is an INSN or JUMP_INSN.
29637 (force_offsettable): Look at base, not at addr.
29638 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29639 on things that aren't necessarily CONST_INTs.
29641 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29643 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29644 -mfpmath=sse is the default also for x86-32 targets with SSE2
29645 instruction set when @option{-ffast-math} is enabled
29647 2017-02-24 Jeff Law <law@redhat.com>
29649 PR rtl-optimizatoin/79286
29650 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29653 2017-02-24 Richard Biener <rguenther@suse.de>
29655 PR tree-optimization/79389
29656 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29659 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29661 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29662 function comment to reflect reality.
29663 (loop_exits_before_overflow): Fix typo in function description.
29665 2017-02-24 Richard Biener <rguenther@suse.de>
29667 PR tree-optimization/79389
29668 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29669 properly that a threading opportunity exists. Detect conditional
29670 copy/constant propagation opportunities.
29672 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29674 * config/visium/visium.md (type): Add trap.
29675 (b): New mode attribute.
29676 (*btst): Rename into...
29677 (*btst<mode>): ...this and adjust.
29678 (*cbranchsi4_btst_insn): Rename into...
29679 (*cbranch<mode>4_btst_insn): ...this and adjust.
29680 (trap): New define_insn.
29682 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29684 PR tree-optimization/79389
29685 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29686 (noce_reversed_cond_code): New function.
29687 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29688 reversed_comparison_code. Formatting fix.
29689 (noce_try_store_flag): Test rev_cond != NULL in addition to
29690 reversed_comparison_code.
29691 (noce_try_store_flag_constants): Likewise.
29692 (noce_try_store_flag_mask): Likewise.
29693 (noce_try_addcc): Use rev_cond if non-NULL instead of
29694 reversed_comparison_code.
29695 (noce_try_cmove_arith): Likewise. Formatting fixes.
29696 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29697 (noce_find_if_block): Initialize rev_cond.
29698 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29699 instead of false as last argument never attempt to reverse it
29702 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29704 PR tree-optimization/79663
29705 * tree-predcom.c (combine_chains): Process refs in reverse order
29706 only for ZERO length chains, and add explaining comment.
29708 2017-02-23 Jeff Law <law@redhat.com>
29710 PR tree-optimization/79578
29711 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29712 in call to operand_equal_p.
29714 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29717 * config/i386/cpuid.h: Fix another undefined behavior.
29719 2017-02-23 Richard Biener <rguenther@suse.de>
29721 PR tree-optimization/79683
29722 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29723 vector types for data-refs.
29725 2017-02-23 Martin Liska <mliska@suse.cz>
29727 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29729 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29731 PR middle-end/79665
29732 * internal-fn.c (get_range_pos_neg): Moved to ...
29733 * tree.c (get_range_pos_neg): ... here. No longer static.
29734 * tree.h (get_range_pos_neg): New prototype.
29735 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29736 are known to be in between 0 and signed maximum inclusive, try to
29737 expand both unsigned and signed divmod and use the cheaper one from
29740 2017-02-22 Jeff Law <law@redhat.com>
29742 PR tree-optimization/79578
29743 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29744 to compare base operands.
29746 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29749 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29750 gpc_reg_operand instead of fpr_reg_operand.
29752 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29754 * config/mips/mips.c (mips_return_in_memory): Force FP
29755 vector types to be returned in memory for o32 ABI.
29757 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29759 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29760 instead of DW_TAG_member for static data member declarations and don't
29761 set no_linkage_name for static inline data members.
29762 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29765 2017-02-22 Martin Liska <mliska@suse.cz>
29767 * doc/invoke.texi: Replace inequality signs with square brackets
29770 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29773 * lra-constraints.c (simplify_operand_subreg): Handle
29774 WORD_REGISTER_OPERATIONS targets.
29776 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29779 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29780 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29781 elimination by swapping fld*.
29783 2017-02-22 Richard Biener <rguenther@suse.de>
29785 PR tree-optimization/79673
29786 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29787 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29788 irrelevant address-space qualifiers and avoiding a
29789 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29791 2017-02-22 Richard Biener <rguenther@suse.de>
29793 PR tree-optimization/79666
29794 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29795 to not symbolically negate if that may introduce undefined
29798 2017-02-22 Martin Liska <mliska@suse.cz>
29801 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29802 * data-streamer-out.c (streamer_write_gcov_count_stream):
29804 * value-prof.c (stream_out_histogram_value): Make assert more
29805 precise based on type of counter.
29807 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29810 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29811 Use nonimmediate_operand instead of memory_operand for operand 1.
29812 (float-extend standard_x87sse_constant_load splitter): Ditto.
29814 2017-02-21 Jeff Law <law@redhat.com>
29816 PR tree-optimization/79621
29817 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29818 blocks with edges to themselves.
29820 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29823 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29824 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29825 Use gimple_call_builtin_p.
29828 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29829 on temporarily removed DEBUG_INSNs.
29831 PR tree-optimization/79649
29832 * tree-loop-distribution.c (classify_partition): Give up on
29833 non-generic address space loads/stores.
29835 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29837 * doc/loop.texi (Loop manipulation): Remove nonexistent
29838 tree_ssa_loop_version from the documentation.
29839 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29841 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29844 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29845 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29846 * config/rs6000/rs6000.c: Include except.h.
29847 (rs6000_expand_split_stack_prologue): Call
29848 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29850 2017-02-21 Martin Jambor <mjambor@suse.cz>
29853 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29854 have been analyzed.
29856 2017-02-21 Martin Jambor <mjambor@suse.cz>
29858 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29859 for backward compatibility only.
29860 * doc/invoke.texi (Option Summary): Remove all references to
29861 -fipa-cp-alignment.
29863 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29867 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29869 * lra-constraints.c (curr_insn_transform): Handle
29870 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29872 2017-02-21 Martin Liska <mliska@suse.cz>
29874 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29876 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29878 PR translation/79638
29879 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29881 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29884 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29885 (arm_function_ok_for_sibcall): Return false for an indirect call by
29886 descriptor if all the argument registers are used.
29887 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29888 alignment of the function.
29890 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29892 PR tree-optimization/61441
29893 * simplify-rtx.c (simplify_const_unary_operation): For
29894 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29895 the sNaN unmodified.
29897 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29899 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29900 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29901 instead of SYSTEM_HEADER_DIR.
29903 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29904 Martin Liška <mliska@suse.cz>
29906 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29907 Fix typos and grammar, use active voice, and clarify.
29909 2017-02-20 Marek Polacek <polacek@redhat.com>
29911 PR middle-end/79537
29912 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29915 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29917 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29920 * config/i386/i386.c (ix86_expand_builtin): Handle
29921 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29922 ix86_builtins_isa[fcode].isa as a requirement of those
29923 flags and any other flag in the bitmask.
29924 (ix86_init_mmx_sse_builtins): Use 0 instead of
29925 ~OPTION_MASK_ISA_64BIT as mask.
29926 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29927 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29928 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29929 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29931 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29934 * lra-constraints.c (split_reg): Check requested split mode
29935 is supported by the register.
29937 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29939 * lra-constraints.c (simplify_operand_subreg): Remove early
29942 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29945 * lra-constraints.c (curr_insn_transform): Tighten condition
29946 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29948 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29951 * lra-constraints.c (curr_insn_transform): Handle
29952 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29954 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29957 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29959 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29961 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29964 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29967 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29969 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29972 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29974 * final.c (last_columnnum, override_columnnum): New variables.
29975 (final_start_function): Set last_columnnum, pass it to begin_prologue
29976 hook and pass 0 to dwarf2out_begin_prologue.
29977 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29978 to source_line debug hook.
29979 (notice_source_line): Compute last_columnnum and for debug_column_info
29980 return true on column changes.
29981 * debug.h (struct gcc_debug_hooks): Add column argument to
29982 source_line and begin_prologue hooks.
29983 (debug_nothing_int_charstar_int_bool): Remove prototype.
29984 (debug_nothing_int_int_charstar,
29985 debug_nothing_int_int_charstar_int_bool): New prototypes.
29986 (dwarf2out_begin_prologue): Add column argument.
29987 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29988 begin_prologue hooks.
29989 (debug_nothing_int_charstar_int_bool): Remove.
29990 (debug_nothing_int_int_charstar,
29991 debug_nothing_int_int_charstar_int_bool): New functions.
29992 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29993 through to dwarf2out_source_line.
29994 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29995 (dwarf2out_source_line): Add column argument, emit it if requested.
29996 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29998 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29999 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30000 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30001 through to dwarf2out_begin_prologue.
30002 (vmsdbgout_source_line): Add column argument, pass it through to
30003 dwarf2out_source_line.
30004 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30005 dbxout_source_line caller.
30006 (dbxout_source_line): Add column argument.
30008 * common.opt (gno-column-info, gcolumn-info): New options.
30009 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30010 (check_die): Also test for multiple DW_AT_decl_column attributes.
30011 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30012 DW_AT_decl_column if requested.
30013 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30015 (gen_variable_die): Likewise.
30016 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30017 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30020 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30021 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30022 (ix86_handle_option): Handle OPT_m3dnowa.
30023 * doc/invoke.texi (-m3dnowa): Document.
30024 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30025 -m3dnowa instead of -m3dnow -march=athlon.
30028 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30029 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30031 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30034 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30035 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30036 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30037 generator for vsx_xxpermdi_<mode>_be.
30038 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30039 force big-endian semantics.
30040 (vsx_xxpermdi_<mode>_be): New define_expand with same
30041 implementation as previous version of vsx_xxpermdi_<mode>.
30043 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30045 PR tree-optimization/79327
30046 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30047 variable, its initialization and use.
30049 2017-02-17 Julia Koval <julia.koval@intel.com>
30051 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30052 (OPTION_MASK_ISA_PKU_UNSET): New.
30053 (ix86_handle_option): Handle -mrdpid.
30054 * config/i386/cpuid.h (bit_RDPID): New.
30055 * config/i386/driver-i386.c (host_detect_local_cpu):
30056 Detect RDPID feature.
30057 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30058 * config/i386/i386-c.c (ix86_target_macros_internal):
30060 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30061 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30062 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30063 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30064 * config/i386/i386.md (define_insn "rdpid"): New.
30065 * config/i386/i386.opt Add -mrdpid.
30066 * config/i386/immintrin.h (_rdpid_u32): New.
30068 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30070 PR rtl-optimization/79541
30071 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30072 instead of transforming it into USE.
30074 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30076 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30077 If HONOR_SNANS (SFmode) force the input to a register.
30078 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30079 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30080 an frsp or similar insn.
30082 2017-02-17 Martin Liska <mliska@suse.cz>
30084 PR rtl-optimization/79577
30085 * params.def (selsched-max-sched-times): Increase minimum to 1.
30087 2017-02-17 Martin Liska <mliska@suse.cz>
30089 PR rtl-optimization/79574
30090 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30092 2017-02-17 Martin Liska <mliska@suse.cz>
30094 PR tree-optimization/79529
30095 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30096 ssa_defined_default_def_p to handle cases which are implicitly
30098 * tree-ssa.c (ssa_defined_default_def_p): New function.
30099 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30100 which are implicitly defined.
30101 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30103 2017-02-17 Richard Biener <rguenther@suse.de>
30105 PR middle-end/79576
30106 * params.def (max-ssa-name-query-depth): Limit to 10.
30108 2017-02-17 Richard Biener <rguenther@suse.de>
30110 PR tree-optimization/79552
30111 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30114 2017-02-17 Richard Biener <rguenther@suse.de>
30117 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30119 2017-02-17 Marek Polacek <polacek@redhat.com>
30121 PR middle-end/79536
30122 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30123 (fold_negate_expr): New wrapper.
30125 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30127 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30128 Correct terminology and de-emphasize pre-standard behavior.
30130 2017-02-16 Alan Modra <amodra@gmail.com>
30132 PR rtl-optimization/79286
30133 * ira.c (def_dominates_uses): New function.
30134 (update_equiv_regs): Don't create an equivalence for insns that
30135 may trap where the register def does not dominate the use.
30137 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30139 PR rtl-optimization/78127
30140 * lra.c (lra): Call lra_eliminate before finish the loop after
30143 2017-02-16 Richard Biener <rguenther@suse.de>
30145 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30147 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30148 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30149 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30150 (isl_val_int_from_wi): New function.
30151 (extract_affine_gmp): Rename to ...
30152 (extract_affine_wi): ... this, take a widest_int.
30153 (extract_affine_int): Just wrap extract_affine_wi.
30154 (add_param_constraints): Use isl_val_int_from_wi.
30155 (add_loop_constraints): Likewise, and extract_affine_wi.
30157 2017-02-15 Jeff Law <law@redhat.com>
30159 PR middle-end/79521
30160 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30161 ira_init_register_move_cost_if_necessary.
30163 2017-02-15 Martin Sebor <msebor@redhat.com>
30165 PR middle-end/32003
30166 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30167 removed in a prior commit.
30169 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30171 PR tree-optimization/79347
30172 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30173 counters during peeling.
30175 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30177 * Makefile.in (site.exp): Remove "set ISLVER".
30179 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30182 * real.c (real_from_integer): Call real_convert even for decimal.
30184 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30187 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30189 2017-02-14 Andrew Pinski <apinski@cavium.com>
30191 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30192 cores and change the partno/implementer to be correct.
30193 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30194 the 'B" as the implementer.
30195 * config/aarch64/aarch64-tune.md: Regenerate.
30197 2017-02-14 Carl Love <cel@us.ibm.com>
30199 * config/rs6000/rs6000.c: Add case statement entry to make the
30200 xvcvuxdsp built-in argument unsigned.
30201 * config/rs6000/vsx.md: Fix the source and return operand types so they
30202 match the instruction definitions from the ISA document. Fix typo
30203 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30206 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30209 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30210 member early_clobber_alts.
30211 * lra-lives.c (reg_early_clobber_p): New.
30212 (process_bb_lives): Use it.
30213 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30214 (debug_operand_data): Initialize early_clobber_alts.
30215 (setup_operand_alternative): Set up early_clobber_alts.
30216 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30217 alternatives to new_insn_reg.
30218 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30220 (lra_update_insn_regno_info): Pass the new arg.
30222 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30224 PR middle-end/79505
30225 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30226 (new_oacc_loop_raw): Don't clear already cleared fields.
30229 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30230 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30231 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30233 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30236 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30238 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30241 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30242 the extra instruction to the right place to store 128-bit constant
30245 2017-02-14 Martin Sebor <msebor@redhat.com>
30247 PR middle-end/79448
30248 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30249 warning for strings of unknown length.
30251 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30253 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30255 2017-02-14 Jeff Law <law@redhat.com>
30258 * ira-costs.c (scan_one_insn): Initialize register move costs
30259 for pseudos seen in USE/CLOBBER insns.
30261 PR tree-optimization/79095
30262 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30263 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30264 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30265 if the operands are known to be not equal, then the resulting range
30267 (intersect_ranges): If the new range is ~[0,0] and the old range is
30268 wide, then prefer ~[0,0].
30269 * tree-vrp.c (overflow_comparison_p_1): New function.
30270 (overflow_comparison_p): New function.
30271 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30272 if NAME is used in an overflow test.
30273 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30274 overflow check that can be expressed as an equality test, then adjust
30275 ops to be that equality test.
30277 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30279 * config/s390/s390-builtin-types.def: Remove flags argument.
30280 * config/s390/s390.c (s390_init_builtins): Likewise.
30282 2017-02-14 Martin Liska <mliska@suse.cz>
30284 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30285 vector. Fix trailing white spaces.
30287 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30289 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30292 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30294 PR rtl-optimization/68664
30295 * config/arm/arm.c (arm_sched_can_speculate_insn):
30296 New function. Declare prototype.
30297 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30299 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30301 PR rtl-optimization/68664
30302 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30304 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30306 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30308 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30309 max skip bytes for function, loop and jump.
30311 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30313 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30314 ABS_EXPR for gimple dump.
30316 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30319 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30321 PR tree-optimization/79408
30322 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30323 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30324 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30325 also if rhs1 is INTEGER_CST.
30327 2017-02-14 Richard Biener <rguenther@suse.de>
30329 PR middle-end/79432
30330 * tree-into-ssa.c (insert_phi_nodes): When the function can
30331 have abnormal edges rewrite SSA names with broken use-def
30332 dominance out of SSA and register them for PHI insertion.
30334 2017-02-13 Martin Sebor <msebor@redhat.com>
30336 PR middle-end/79496
30337 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30338 clearing info.nowrite flag when snprintf size argument is a range.
30340 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30342 * cprop.c (cprop_jump): Add missing space in string literal.
30343 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30344 (get_constraint_for_component_ref): Likewise.
30345 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30346 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30347 * lra-constraints.c (process_alt_operands): Likewise.
30348 * ipa-inline.c (inline_small_functions): Likewise.
30349 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30350 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30351 * trans-mem.c (diagnose_tm_1_op): Likewise.
30352 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30353 (grid_parallel_clauses_gridifiable): Likewise.
30355 * config/nvptx/mkoffload.c (process): Add space in between
30358 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30359 "MOD4_SSE_REGS" and "ALL_REGS".
30361 * spellcheck.c (test_data): Add , in between "foo" and "food".
30363 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30366 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30367 boundary crossing check and subsequent code generation agree.
30369 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30371 * config/aarch64/aarch64.c (has_memory_op): Delete.
30372 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30375 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30377 PR rtl-optimization/79388
30378 PR rtl-optimization/79450
30379 * combine.c (distribute_notes): When removing TEM_INSN for which
30380 corresponding dest has last value recorded, invalidate that last
30383 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30385 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30386 of explicit '@'. Add missing assembly comment marker on branch costs
30389 2017-02-13 Nathan Sidwell <nathan@acm.org>
30391 * gengtype-lex.l (<in_struct>): Add '/'.
30393 2017-02-13 Martin Liska <mliska@suse.cz>
30396 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30398 2017-02-13 Richard Biener <rguenther@suse.de>
30400 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30402 * configure: Re-generate.
30403 * config.in: Likewise.
30404 * graphite-dependences.c: Simplify as if
30405 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30406 * graphite-isl-ast-to-gimple.c: Likewise.
30407 * graphite-optimize-isl.c: Likewise.
30408 * graphite-poly.c: Likewise.
30409 * graphite-sese-to-poly.c: Likewise.
30410 * graphite.h: Likewise.
30411 * toplev.c: Include isl/version.h and use isl_version () for
30412 printing the ISL version.
30413 * doc/install.texi: Update ISL requirement.
30415 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30417 * doc/standards.texi (Standards): Update reference to
30420 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30422 * doc/extend.texi (Named Address Spaces): sourceware.org now
30424 * doc/install.texi (Binaries): Ditto.
30427 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30429 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30430 standard terminology "stringize"/"stringizing" throughout.
30431 * doc/cppinternals.texi: Likewise.
30433 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30435 * doc/extend.texi: Fix some spelling mistakes and typos.
30436 * doc/invoke.texi: Likewise.
30438 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30441 * params.def (inline-min-speedup) Change from 10 to 8.
30443 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30445 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30448 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30451 * ipa-inline-analysis.c (get_minimal_bb): New function.
30452 (record_modified): Use it.
30453 (remap_edge_change_prob): Handle also ancestor functions.
30455 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30457 * doc/contrib.texi (Contributors): Remove broken link into
30458 the Mauve CVS repository.
30460 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30462 PR middle-end/79454
30463 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30464 result computation whenever lhs doesn't have vector mode, not
30465 just when it has BLKmode.
30467 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30469 * doc/makefile.texi (profiledbootstrap): Refer to the
30470 installation instructions only in textual form.
30472 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30475 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30477 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30479 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30480 (Specific): Update mingw-w64 reference.
30482 (Specific): Remove broken link to Renesas RX processor.
30484 2017-02-10 Richard Biener <rguenther@suse.de>
30486 * toplev.c (process_options): Do not mention obsolete graphite
30487 options when printing sorry message about missing graphite support.
30488 Mention -floop-nest-optimize.
30490 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30492 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30493 (vtst_p16): Likewise.
30494 (vtstq_p8): Likewise.
30495 (vtstq_p16): Likewise.
30497 (vtstq_p64): Likewise.
30498 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30499 (vset_lane_p64): New.
30500 (vsetq_lane_p64): New.
30502 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30504 PR tree-optimization/79411
30505 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30506 stmt operands are SSA_NAMEs used in abnormal phis.
30507 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30510 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30513 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30516 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30518 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30520 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30522 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30526 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30527 not arbitrary TREE_CONSTANT.
30530 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30531 "omp declare target link" attribute unless is_global_var.
30532 * omp-offload.c (find_link_var_op): Likewise.
30534 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30535 Chung-Lin Tang <cltang@codesourcery.com>
30537 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30539 (gimplify_adjust_omp_clauses): Don't delete TILE.
30540 (gimplify_omp_for): Deal with TILE.
30541 * internal-fn.c (expand_GOACC_TILE): New function.
30542 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30544 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30545 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30547 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30548 avoid DIV for outermost collapse var.
30549 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30550 Remove out of date comments, fix whitespace.
30551 * omp-general.c (omp_extract_for_data): Deal with tiling.
30552 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30553 adjust OLF_DIM_BASE value.
30554 (struct omp_for_data): Add tiling field.
30555 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30556 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30557 for auto loops. Remove default auto determining, moved to
30558 oacc_loop_fixed_partitions.
30559 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30560 stmts, add e_mask field.
30561 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30562 (oacc_thread_numbers): Use oacc_dim_call.
30563 (oacc_xform_tile): New.
30564 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30565 (finish_oacc_loop): Adjust for ifns vector.
30566 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30567 add case for GOACC_TILE fns.
30568 (oacc_loop_xform_loop): Delete.
30569 (oacc_loop_process): Iterate over call list directly, and add
30570 handling for GOACC_TILE fns.
30571 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30573 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30574 vector partitioning to outer loops. Assign 2 partitions to loops
30575 when available. Add TILE handling.
30576 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30577 (execite_oacc_device_lower): Process GOACC_TILE fns,
30578 ignore unknown specs.
30579 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30580 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30581 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30583 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30585 * configure.ac (ACX_BUGURL): Update.
30586 * configure: Regenerate.
30588 2017-02-09 Richard Biener <rguenther@suse.de>
30590 PR tree-optimization/69823
30591 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30592 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30594 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30596 * config/arc/arc-c.def: Add __NPS400__ definition.
30597 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30598 (TARGET_NPS400): Define.
30600 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30602 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30604 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30605 pointer, arch_info.
30606 (arc_cpu_types): Fill the arch_info field with a pointer into the
30607 arc_arch_types table.
30608 (arc_selected_cpu): Declare.
30609 * config/arc/arc.c (arc_selected_cpu): Make global.
30610 (arc_selected_arch): Delete.
30611 (arc_base_cpu): Delete.
30612 (arc_override_options): Remove references to deleted variables,
30613 update access to arch information.
30614 (ARC_OPT): Update access to arch information.
30615 (ARC_OPTX): Likewise.
30616 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30617 (TARGET_ARC600): Update access to arch information.
30618 (TARGET_ARC601): Likewise.
30619 (TARGET_ARC700): Likewise.
30620 (TARGET_EM): Likewise.
30621 (TARGET_HS): Likewise.
30622 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30625 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30628 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30629 condition/operands for integer GE/LE/GEU/LEU operations.
30631 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30633 PR translation/79397
30634 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30637 2017-02-08 Martin Jambor <mjambor@suse.cz>
30640 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30641 whether allocation happened.
30642 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30643 nothing was allocated.
30645 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30647 PR tree-optimization/79408
30648 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30649 constant, but SSA_NAME with a known integer range, use the minimum
30650 of that range instead of op1 to determine if modulo can be replaced
30651 with its first operand.
30653 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30655 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30657 2017-02-08 Richard Biener <rguenther@suse.de>
30659 PR tree-optimization/71824
30660 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30661 Check all loops contained in the merged region.
30663 2017-02-07 Andrew Pinski <apinski@cavium.com>
30665 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30667 2017-02-07 Andrew Pinski <apinski@cavium.com>
30669 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30670 (thunderxt88): Likewise.
30671 (thunderxt81): Disable LSE and change v8.1 to v8.
30672 (thunderxt83): Likewise.
30674 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30675 Richard Biener <rguenther@suse.de>
30677 PR middle-end/79399
30678 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30679 type from int to size_t.
30680 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30682 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30684 PR rtl-optimization/79386
30685 * cprop.c (bypass_conditional_jumps): Initialize
30686 bypass_last_basic_block already before splitting bbs after
30687 unconditional traps...
30688 (bypass_conditional_jumps): ... rather than here.
30691 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30692 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30693 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30694 fix -masm=intel patterns.
30696 2017-02-07 Richard Biener <rguenther@suse.de>
30698 PR tree-optimization/79256
30699 PR middle-end/79278
30700 * builtins.c (get_object_alignment_2): Use min_align_of_type
30701 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30702 and ADJUST_FIELD_ALIGN.
30704 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30706 * doc/tm.texi: Regenerate.
30707 * stor-layout.c (layout_decl): Adjust.
30708 (update_alignment_for_field): Likewise.
30709 (place_field): Likewise.
30710 (min_align_of_type): Likewise.
30711 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30712 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30713 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30714 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30715 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30716 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30717 * config/i386/i386.c (x86_field_alignment): Likewise.
30718 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30719 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30720 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30721 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30722 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30723 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30727 2017-01-30 Richard Biener <rguenther@suse.de>
30729 PR tree-optimization/79256
30730 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30731 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30734 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30736 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30737 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30738 builtins to SImode and emit a zero-extend, if necessary.
30740 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30742 * docs/invoke.texi (RISC-V Options): Alphabetize.
30744 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30746 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30749 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30751 * config/riscv/riscv.c: New file.
30752 * common/config/riscv/riscv-common.c: Likewise.
30753 * config.gcc: Likewise.
30754 * config/riscv/constraints.md: Likewise.
30755 * config/riscv/elf.h: Likewise.
30756 * config/riscv/generic.md: Likewise.
30757 * config/riscv/linux.h: Likewise.
30758 * config/riscv/multilib-generator: Likewise.
30759 * config/riscv/peephole.md: Likewise.
30760 * config/riscv/pic.md: Likewise.
30761 * config/riscv/predicates.md: Likewise.
30762 * config/riscv/riscv-builtins.c: Likewise.
30763 * config/riscv/riscv-c.c: Likewise.
30764 * config/riscv/riscv-ftypes.def: Likewise.
30765 * config/riscv/riscv-modes.def: Likewise.
30766 * config/riscv/riscv-opts.h: Likewise.
30767 * config/riscv/riscv-protos.h: Likewise.
30768 * config/riscv/riscv.h: Likewise.
30769 * config/riscv/riscv.md: Likewise.
30770 * config/riscv/riscv.opt: Likewise.
30771 * config/riscv/sync.md: Likewise.
30772 * config/riscv/t-elf-multilib: Likewise.
30773 * config/riscv/t-linux: Likewise.
30774 * config/riscv/t-linux-multilib: Likewise.
30775 * config/riscv/t-riscv: Likewise.
30776 * configure.ac: Likewise.
30777 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30778 Waterman as RISC-V maintainers.
30779 * doc/install.texi: Add RISC-V entries.
30780 * doc/invoke.texi: Add RISC-V options section.
30781 * doc/md.texi: Add RISC-V constraints section.
30782 * configure: Regenerated.
30784 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30787 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30788 false values to be constant vectors with all 0 or all 1 bits set.
30789 (vcondu<mode><mode>): Likewise.
30790 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30792 (fpmask_comparison_operator): Update comment.
30793 (vecint_comparison_operator): New predicate.
30794 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30795 vector conditionals when the true and false values are constant
30796 vectors with all 0 bits or all 1 bits set.
30798 2017-02-06 Martin Sebor <msebor@redhat.com>
30800 PR tree-optimization/79376
30801 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30803 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30805 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30806 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30807 to simplify split condition.
30809 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30811 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30812 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30815 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30817 PR rtl-optimization/68664
30818 * target.def (can_speculate_insn): New hook.
30819 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30820 * doc/tm.texi: Regenerate.
30821 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30822 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30823 (rs6000_sched_can_speculate_insn): New function.
30825 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30827 PR tree-optimization/79284
30828 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30829 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30830 vectorizable_mask_load_store, vectorizable_operation,
30831 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30832 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30833 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30834 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30835 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30836 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30837 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30838 is_gimple_assign (stmt). Replace another such test with
30839 is_gimple_assign (stmt).
30841 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30844 * config/avr/avr.c (rtl-iter.h): Include it.
30845 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30846 (avr_legitimate_combined_insn): ...and implementation.
30848 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30850 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30851 * config/s390/s390.c (s390_const_operand_ok)
30852 (s390_canonicalize_comparison, s390_extract_part)
30853 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30854 (s390_contiguous_bitmask_p, s390_rtx_costs)
30855 (legitimize_pic_address): Likewise.
30856 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30857 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30858 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30859 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30860 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30862 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30864 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30865 REGNO($0) == REGNO($1).
30867 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30869 * config/s390/linux.h(SIZE_TYPE): Add comment.
30871 2017-02-06 Julian Brown <julian@codesourcery.com>
30872 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30873 Virendra Pathak <virendra.pathak@broadcom.com>
30875 * config/aarch64/aarch64-cores.def: Change the scheduler
30877 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30878 * config/aarch64/thunderx2t99.md: New file.
30880 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30882 * doc/standards.texi (Go Language): Update link to language
30885 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30887 * tree-eh.c (lower_resx): Sanitize profile.
30888 (cleanup_empty_eh_move_lp): Likewise.
30890 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30893 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30895 * cfgloopmanip.h (loop_version): Update prototype.
30896 * modulo-sched.c (sms_schedule): Update call of loop_version.
30897 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30898 * tree-parloops.c (gen_parallel_loop): Likewise.
30899 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30900 * tree-ssa-loop-split.c (split_loop): Likewise.
30901 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30902 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30904 2017-02-05 Martin Liska <mliska@suse.cz>
30907 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30909 (print_operand_address): Initialize a struct to zero.
30911 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30913 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30914 garbage collector only in textual form.
30916 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30918 * doc/extend.texi (x86 specific memory model extensions for
30919 transactional memory): Simplify a phrase.
30921 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30924 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30925 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30926 (atomic_storedi_1): Likewise.
30928 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30930 PR tree-optimization/79338
30931 * tree-parloops.c (gather_scalar_reductions): Don't call
30932 vect_analyze_loop_form for loop->inner before destroying loop's
30935 2017-02-03 Martin Sebor <msebor@redhat.com>
30937 PR tree-optimization/79327
30938 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30939 when precision has resulted in leading zeros.
30940 (format_integer): Adjust the likely counter to assume an unknown
30941 argument that may be zero is non-zero.
30943 2017-02-03 Jason Merrill <jason@redhat.com>
30946 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30947 avoid copying non-taken branch.
30949 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30951 PR tree-optimization/79340
30952 * tree-vect-loop.c (vectorizable_reduction): Release
30953 vec_defs elements after safe_splicing them into other vectors.
30956 PR tree-optimization/79327
30957 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30958 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30960 (format_integer): Use wide_int_to_tree instead of build_int_cst
30961 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30962 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30963 of shortest and longest sequence.
30965 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30967 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30968 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30970 2017-02-03 Walter Lee <walt@tilera.com>
30973 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30974 after initial stackframe link reg save.
30975 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30977 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30980 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30981 wu for stxssp alternative.
30983 2017-02-03 Martin Sebor <msebor@redhat.com>
30985 PR tree-optimization/79352
30986 * gimple-fold.c (get_range_strlen): Add argument.
30987 (get_range_strlen): Change return type to bool.
30988 (get_maxval_strlen): Pass in a dummy argument.
30989 * gimple-fold.h (get_range_strlen): Change return type to bool.
30990 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30991 * tree.h (array_at_struct_end_p): Add argument.
30992 * tree.c (array_at_struct_end_p): Handle it.
30994 2017-02-03 Martin Liska <mliska@suse.cz>
30997 * multiple_target.c (create_dispatcher_calls): Redirect edge
30998 from a caller of a dispatcher.
30999 (expand_target_clones): Make the clones local.
31000 (ipa_target_clone): Do both target clones and resolvers.
31001 (ipa_dispatcher_calls): Remove the pass.
31002 (pass_dispatcher_calls::gate): Likewise.
31003 (make_pass_dispatcher_calls): Likewise.
31004 * passes.def (pass_target_clone): Put as very first IPA early
31007 2017-02-03 Martin Liska <mliska@suse.cz>
31009 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31010 in case of a function with ifunc attribute.
31012 2017-02-03 Martin Liska <mliska@suse.cz>
31014 * cgraph.c (cgraph_node::dump): Dump function version info.
31015 * symtab.c (symtab_node::dump_base): Add missing new line.
31017 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31019 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31020 (ifcombine_ifandif): Use it.
31022 2017-02-03 Martin Liska <mliska@suse.cz>
31024 * doc/invoke.texi: Document default value for
31025 use-after-scope-direct-emission-threshold.
31027 2017-02-03 Martin Liska <mliska@suse.cz>
31029 PR tree-optimization/79339
31030 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31031 (format_floating): Likewise.
31033 2017-02-03 Martin Liska <mliska@suse.cz>
31036 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31038 (ipa_node_params_t::remove): Likewise.
31039 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31040 initialization from removed ipa_node_params_t::insert.
31041 (ipa_node_params::~ipa_node_params): Move from removed
31042 ipa_node_params_t::release.
31043 * symbol-summary.h (symbol_summary::m_released): New member.
31044 Do not release a summary twice. Do not allow to call finalizer
31045 for types of a summary that live in GGC memory.
31047 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31049 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31052 2017-02-02 Martin Sebor <msebor@redhat.com>
31054 PR middle-end/79275
31055 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31056 (format_string): Tighten up the range of output for non-constant
31057 strings and correct the expected range for wide non-constant strings.
31059 2017-02-02 Martin Sebor <msebor@redhat.com>
31061 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31063 PR middle-end/32003
31064 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31066 (-fdump-tree-@var): Add to index and document how to come up
31067 with pass-specific option and dump file names.
31068 (-fdump-passes): Clarify where to look for output.
31070 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31072 PR middle-end/77445
31073 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31074 statistics of the analyzed path; allow threading for speed when
31075 any of BBs along the path are optimized for speed.
31077 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31079 PR middle-end/78468
31080 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31081 settings of the virtual registers.
31084 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31086 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31087 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31090 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31092 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31093 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31095 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31097 * config/s390/s390.md: Add missing comments with the expanded
31099 * config/s390/vector.md: Likewise.
31100 * config/s390/vx-builtins.md: Likewise.
31102 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31105 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31106 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31107 conditions on a single line.
31109 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31111 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31112 __S390_VX__ to __VX__.
31114 2017-02-01 Andrew Pinski <apinski@cavium.com>
31116 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31117 stmt_info to record_stmt_cost.
31118 (vect_get_known_peeling_cost): Pass stmt_info if known to
31120 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31121 cpu_vector_cost field into
31122 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31123 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31124 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31125 splitting of scalar_stmt_cost and vec_stmt_cost.
31126 (thunderx_vector_cost): Likewise.
31127 (cortexa57_vector_cost): LIkewise.
31128 (exynosm1_vector_cost): Likewise.
31129 (xgene1_vector_cost): Likewise.
31130 (thunderx2t99_vector_cost): Improve after the splitting of the two
31132 (aarch64_builtin_vectorization_cost): Update for the splitting of
31133 scalar_stmt_cost and vec_stmt_cost.
31135 2017-02-01 Torvald Riegel <triegel@redhat.com>
31136 Richard Henderson <rth@redhat.com>
31138 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31139 conditional on existance of a fast atomic load.
31140 * optabs-query.c (can_atomic_load_p): New function.
31141 * optabs-query.h (can_atomic_load_p): Declare it.
31142 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31143 no fast atomic load is available for the particular size of access.
31144 (expand_atomic_compare_and_swap): Likewise.
31145 (expand_atomic_load): Likewise.
31146 (expand_atomic_store): Likewise.
31147 (expand_atomic_fetch_op): Likewise.
31148 * testsuite/lib/target-supports.exp
31149 (check_effective_target_sync_int_128): Remove x86 because it provides
31150 no fast atomic load.
31151 (check_effective_target_sync_int_128_runtime): Likewise.
31153 2017-02-01 Richard Biener <rguenther@suse.de>
31155 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31156 (graphite_transform_loops): Provide opt-info for optimized nests.
31157 * tree-parloop.c (parallelize_loops): Provide opt-info for
31158 parallelized loops.
31160 2017-02-01 Richard Biener <rguenther@suse.de>
31162 PR middle-end/79315
31163 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31164 was not set before.
31166 2017-02-01 Richard Biener <rguenther@suse.de>
31168 PR tree-optimization/71824
31169 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31170 Verify the loops are valid in the merged SESE region.
31171 (scop_detection::can_represent_loop_1): Check analyzing the
31172 evolution of the number of iterations in the region succeeds.
31174 2017-01-31 Ian Lance Taylor <iant@golang.org>
31176 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31177 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31179 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31181 PR preprocessor/79210
31182 * input.c (get_substring_ranges_for_loc): Replace line_width
31183 assertion with error-handling.
31185 2017-01-31 Richard Biener <rguenther@suse.de>
31187 PR tree-optimization/77318
31188 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31189 (create_pw_aff_from_tree): Take loop parameter.
31190 (add_condition_to_pbb): Pass loop of the condition to
31191 create_pw_aff_from_tree.
31193 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31195 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31196 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31198 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31202 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31204 (convert_int_to_float128): Likewise.
31205 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31206 (convert_int_to_float128): Likewise.
31207 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31208 (UNSPEC_IEEE128_CONVERT): Likewise.
31209 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31210 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31211 Use local variables for IBM extended format.
31212 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31213 (fix_trunc<mode>si2_fprs): Likewise.
31214 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31215 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31216 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31217 to know that we can now have integers of all sizes in vector
31219 (fix<uns>_<mode>di2_hw): Likewise.
31220 (float<uns>_<mode>si2_hw): Likewise.
31221 (fix_<mode>si2_hw): Likewise.
31222 (fixuns_<mode>si2_hw): Likewise.
31223 (float<uns>_<mode>di2_hw): Likewise.
31224 (float_<mode>di2_hw): Likewise.
31225 (float_<mode>si2_hw): Likewise.
31226 (floatuns_<mode>di2_hw): Likewise.
31227 (floatuns_<mode>si2_hw): Likewise.
31228 (xscvqp<su>wz_<mode>): Delete, no longer used.
31229 (xscvqp<su>dz_<mode>): Likewise.
31230 (xscv<su>dqp_<mode>): Likewise.
31231 (ieee128_mfvsrd_64bit): Likewise.
31232 (ieee128_mfvsrd_32bit): Likewise.
31233 (ieee128_mfvsrwz): Likewise.
31234 (ieee128_mtvsrw): Likewise.
31235 (ieee128_mtvsrd_64bit): Likewise.
31236 (ieee128_mtvsrd_32bit): Likewise.
31238 2017-01-31 Martin Liska <mliska@suse.cz>
31241 * ipa-prop.c (ipa_free_all_node_params): Call release method
31242 instead of ~sumbol_summary to not to trigger double times
31245 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31247 PR tree-optimization/71691
31248 * bitmap.h (class auto_bitmap): New.
31249 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31250 is_maybe_undefined instead of ssa_undefined_value_p.
31252 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31254 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31255 __S390_ARCH_LEVEL__ to __ARCH__.
31257 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31259 PR tree-optimization/79267
31260 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31261 if should_remove_lhs_p is true.
31263 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31266 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31267 (add_alignment_attribute): New.
31268 (base_type_die): Add alignment attribute.
31269 (subrange_type_die): Likewise.
31270 (modified_type_die): Likewise.
31271 (gen_array_type_die): Likewise.
31272 (gen_descr_array_type_die: Likewise.
31273 (gen_enumeration_type_die): Likewise.
31274 (gen_subprogram_die): Likewise.
31275 (gen_variable_die): Likewise.
31276 (gen_field_die): Likewise.
31277 (gen_ptr_to_mbr_type_die): Likewise.
31278 (gen_struct_or_union_type_die): Likewise.
31279 (gen_subroutine_type_die): Likewise.
31280 (gen_typedef_die): Likewise.
31281 (base_type_cmp): Compare alignment attribute.
31283 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31286 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31287 (setb_unsigned) New pattern for setb with CCUNS.
31288 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31289 subfc./subfe sequence to avoid overflow problems. Generate a
31290 shorter sequence with cmpld/setb for power9.
31291 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31292 for generating subfc. instruction.
31293 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31294 now uses this instruction.
31296 2017-01-30 Ian Lance Taylor <iant@google.com>
31299 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31300 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31302 2017-01-30 Martin Sebor <msebor@redhat.com>
31304 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31305 Move constant to the right of a relational operator.
31306 (get_mpfr_format_length, format_character, format_string): Ditto.
31307 (should_warn_p, maybe_warn): Same.
31309 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31311 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31314 * asan.c (get_translation_unit_decl): Remove function.
31315 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31317 2017-01-30 Martin Liska <mliska@suse.cz>
31319 PR gcov-profile/79259
31320 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31321 -fprofile-generate.
31323 2017-01-30 Martin Liska <mliska@suse.cz>
31326 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31327 Initialize variables with NULL value.
31329 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31332 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31334 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31336 2017-01-30 Richard Biener <rguenther@suse.de>
31338 PR tree-optimization/79276
31339 * tree-vrp.c (process_assert_insertions): Properly adjust common
31340 when removing a duplicate.
31342 2017-01-30 Richard Biener <rguenther@suse.de>
31344 PR tree-optimization/79256
31345 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31346 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31348 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31350 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31353 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31354 ("*r<noxa>sbg_<mode>_sll_bitmask")
31355 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31356 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31357 Use contiguous_bitmask_nowrap_operand.
31359 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31362 * config/rs6000/altivec.h (vec_xl): Revise #define.
31363 (vec_xst): Likewise.
31365 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31367 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31369 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31371 PR rtl-optimization/79194
31372 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31373 traps before call to bypass_conditional_jumps.
31375 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31377 PR tree-optimization/71374
31378 * lra-constraints.c (check_conflict_input_operands): New.
31379 (match_reload): Use it.
31381 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31384 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31385 account to calculate conflict_set.
31387 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31389 PR rtl-optimization/78559
31390 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31391 other_insn in combine.
31393 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31395 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31396 uint16_type_node for BT_UINT16.
31398 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31400 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31401 "RTL Tests" to menu.
31402 (GIMPLE Tests): New node.
31403 (RTL Tests): New node.
31405 2017-01-27 Richard Biener <rguenther@suse.de>
31407 PR tree-optimization/79245
31408 * tree-loop-distribution.c (distribute_loop): Apply cost
31409 modeling also to detected patterns.
31411 2017-01-27 Richard Biener <rguenther@suse.de>
31413 PR tree-optimization/71433
31414 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31415 (compare_assert_loc): New function.
31416 (process_assert_insertions): Sort and optimize assert locations
31417 to remove duplicates and push down identical assertions on
31418 edges to their destination block.
31420 2017-01-27 Richard Biener <rguenther@suse.de>
31422 PR tree-optimization/79244
31423 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31424 out SSA names even if abnormal.
31426 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31428 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31429 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31430 instead of MPFR_RNDN.
31432 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31435 * arm.c (arm_option_override): Don't call build_target_option_node
31436 until after doing all option overrides.
31437 (arm_valid_target_attribute_tree): Likewise.
31439 2017-01-27 Martin Liska <mliska@suse.cz>
31441 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31442 for {cd}tors and C++ {cd}tors.
31444 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31446 * config/s390/s390.md ("*setmem_long_and")
31447 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31449 2017-01-26 Martin Sebor <msebor@redhat.com>
31451 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31454 2017-01-26 Martin Sebor <msebor@redhat.com>
31456 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31457 HAVE_DFmode before using XFmode or DFmode.
31458 (parse_directive): Avoid using the z length modifier to avoid
31459 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31461 PR middle-end/78703
31462 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31463 to accept adjustment as an array.
31464 (get_int_range): New function.
31465 (struct directive): Make width and prec arrays.
31466 (directive::set_width, directive::set_precision): Call get_int_range.
31467 (format_integer, format_floating): Handle width and precision ranges.
31468 (format_string, parse_directive): Same.
31470 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31473 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31474 comdat_type_p set, just clone them, but keep the children in the
31478 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31479 which have direct callers with -fvar-tracking-assignments enabled
31481 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31482 inside of type units.
31484 2017-01-26 Martin Sebor <msebor@redhat.com>
31486 PR middle-end/78703
31487 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31489 (struct format_result): Replace number_chars, number_chars_min,
31490 and number_chars_max with a single member of struct result_range.
31492 (format_result::operator+=): Adjust.
31493 (struct fmtresult): Remove bounded. Handle likely and unlikely
31495 (fmtresult::adjust_for_width_or_precision): New function.
31496 (fmtresult:type_max_digits): New function.
31497 (bytes_remaining): Handle likely and unlikely counters.
31498 (min_bytes_remaining): Remove.
31499 (format_percent): Simplify.
31500 (format_integer, format_floating): Set likely and unlikely counters.
31501 (get_string_length, format_character, format_string): Same.
31502 (format_plain, should_warn_p): New function.
31503 (maybe_warn): Call should_warn_p. Update diagnostic messages
31504 and handle those for all directives, including plain strings.
31505 (format_directive): Handle likely and unlikely counters.
31506 Remove unnecessary quoting from diagnostics. Add an informational
31508 (add_bytes): Remove.
31509 (pass_sprintf_length::compute_format_length): Simplify.
31510 (try_substitute_return_value): Handle likely and unlikely counters.
31512 2017-01-26 Carl Love <cel@us.ibm.com>
31514 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31515 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31517 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31520 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31521 endianess for subregs into account.
31522 * lra-constraints.c (lra_constraints): Do risky transformations
31523 always on the first iteration.
31524 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31525 last_call_used_reg_set.
31526 (process_bb_lives): Define and use last_call_used_reg_set.
31527 * lra.c (lra): Always continue after lra_constraints on the first
31530 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31532 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31534 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31536 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31538 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31539 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31540 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31541 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31542 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31543 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31544 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31545 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31546 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31548 2017-01-26 Marek Polacek <polacek@redhat.com>
31551 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31552 for the third operand.
31554 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31556 PR middle-end/79236
31557 * omp-low.c (struct omp_context): Add simt_stmt field.
31558 (scan_omp_for): Return omp_context *.
31559 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31560 context to the _simt_ SIMD stmt.
31561 (lower_omp_for): For combined SIMD with sibling _simt_
31562 SIMD, make sure to use the same decls in _looptemp_
31563 clauses as in the sibling.
31565 2017-01-26 David Sherwood <david.sherwood@arm.com>
31567 PR middle-end/79212
31568 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31571 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31574 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31575 emit fld b; fld a; if possible.
31577 * brig-builtins.def: Update copyright years.
31578 * config/arm/arm_acle_builtins.def: Update copyright years.
31580 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31583 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31584 constraint instead of o for the stxsd instruction.
31586 2017-01-25 Carl Love <cel@us.ibm.com>
31588 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31589 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31591 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31593 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31595 2017-01-25 Richard Biener <rguenther@suse.de>
31597 PR tree-optimization/69264
31598 * target.def (vector_alignment_reachable): Improve documentation.
31599 * doc/tm.texi: Regenerate.
31600 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31602 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31603 earlier changes with respect to TYPE_USER_ALIGN.
31604 (vector_alignment_reachable_p): Likewise. Improve dumping.
31606 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31609 * config/arm/arm.md (xordi3): Force constant operand into a register
31612 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31614 * doc/invoke.texi (-fstore-merging): Correct default optimization
31615 levels at which it is enabled.
31616 (-O): Move -fstore-merging from list to...
31619 2017-01-25 Richard Biener <rguenther@suse.de>
31622 * omp-expand.c: Include debug.h.
31623 (expand_omp_taskreg): Make sure to generate early debug before
31624 outlining anything from a function.
31625 (expand_omp_target): Likewise.
31626 (grid_expand_target_grid_body): Likewise.
31628 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31631 * asan.c (get_translation_unit_decl): New function.
31632 (asan_add_global): Extract modules file name from globals
31633 TRANSLATION_UNIT_DECL name.
31635 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31638 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31639 for long calls with APCS frame and VFP.
31641 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31643 * cfg.c (original_copy_tables_initialized_p): New function.
31644 * cfg.h (original_copy_tables_initialized_p): New decl.
31645 * cfgrtl.c (relink_block_chain): Guard the call to
31646 free_original_copy_tables with a call to
31647 original_copy_tables_initialized_p.
31648 * cgraph.h (symtab_node::native_rtl_p): New decl.
31649 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31650 (symtab_node::needed_p): Don't assert for early assembly output
31651 for __RTL functions.
31652 (cgraph_node::finalize_function): Set "force_output" for __RTL
31654 (cgraph_node::analyze): Bail out early for __RTL functions.
31655 (analyze_functions): Update assertion to support __RTL functions.
31656 (cgraph_node::expand): Bail out early for __RTL functions.
31657 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31659 * function.h (struct function): Update comment for field
31661 * gimple-expr.c: Include "tree-pass.h".
31662 (gimple_has_body_p): Return false for __RTL functions.
31663 * Makefile.in (OBJS): Add run-rtl-passes.o.
31664 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31666 (gcc::pass_manager::get_clean_slate): New accessor.
31667 * passes.c: Include "insn-addr.h".
31668 (should_skip_pass_p): Add logging. Update logic for running
31669 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31670 property-provider override so it is only done for gimple passes.
31672 (skip_pass): New function.
31673 (execute_one_pass): Call skip_pass when skipping passes.
31674 * read-md.c (md_reader::read_char): Support filtering
31675 the input to a subset of line numbers.
31676 (md_reader::md_reader): Initialize fields
31677 m_first_line and m_last_line.
31678 (md_reader::read_file_fragment): New function.
31679 * read-md.h (md_reader::read_file_fragment): New decl.
31680 (md_reader::m_first_line): New field.
31681 (md_reader::m_last_line): New field.
31682 * read-rtl-function.c (function_reader::create_function): Only
31683 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31684 curr_properties. Set DECL_INITIAL to a dummy block.
31685 (read_rtl_function_body_from_file_range): New function.
31686 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31688 * run-rtl-passes.c: New file.
31689 * run-rtl-passes.h: New file.
31691 2017-01-24 Jeff Law <law@redhat.com>
31693 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31696 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31698 PR tree-optimization/79159
31699 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31700 (record_nonwrapping_iv): Improve boundary using above function if no
31701 value range information.
31703 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31704 Martin Jambor <mjambor@suse.cz>
31706 * brig-builtins.def: New file.
31707 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31708 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31709 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31710 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31711 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31712 * builtin-types.def (BT_INT8): New.
31713 (BT_INT16): Likewise.
31714 (BT_UINT8): Likewise.
31715 (BT_UINT16): Likewise.
31716 (BT_FN_ULONG): Likewise.
31717 (BT_FN_UINT_INT): Likewise.
31718 (BT_FN_UINT_ULONG): Likewise.
31719 (BT_FN_UINT_LONG): Likewise.
31720 (BT_FN_UINT_PTR): Likewise.
31721 (BT_FN_ULONG_PTR): Likewise.
31722 (BT_FN_INT8_FLOAT): Likewise.
31723 (BT_FN_INT16_FLOAT): Likewise.
31724 (BT_FN_UINT32_FLOAT): Likewise.
31725 (BT_FN_UINT16_FLOAT): Likewise.
31726 (BT_FN_UINT8_FLOAT): Likewise.
31727 (BT_FN_UINT64_FLOAT): Likewise.
31728 (BT_FN_UINT16_UINT32): Likewise.
31729 (BT_FN_UINT32_UINT16): Likewise.
31730 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31731 (BT_FN_INT_PTR_INT): Likewise.
31732 (BT_FN_UINT_PTR_UINT): Likewise.
31733 (BT_FN_LONG_PTR_LONG): Likewise.
31734 (BT_FN_ULONG_PTR_ULONG): Likewise.
31735 (BT_FN_VOID_UINT64_UINT64): Likewise.
31736 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31737 (BT_FN_INT8_INT8_INT8): Likewise.
31738 (BT_FN_INT16_INT16_INT16): Likewise.
31739 (BT_FN_INT_INT_INT): Likewise.
31740 (BT_FN_UINT_FLOAT_UINT): Likewise.
31741 (BT_FN_FLOAT_UINT_UINT): Likewise.
31742 (BT_FN_ULONG_UINT_UINT): Likewise.
31743 (BT_FN_ULONG_UINT_PTR): Likewise.
31744 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31745 (BT_FN_UINT_UINT_UINT): Likewise.
31746 (BT_FN_VOID_UINT_PTR): Likewise.
31747 (BT_FN_UINT_UINT_PTR: Likewise.
31748 (BT_FN_UINT32_UINT64_PTR): Likewise.
31749 (BT_FN_INT_INT_UINT_UINT): Likewise.
31750 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31751 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31752 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31753 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31754 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31755 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31756 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31757 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31758 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31759 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31760 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31761 * doc/frontends.texi: List BRIG FE.
31762 * doc/install.texi (Testing): Add BRIG tesring requirements.
31763 * doc/invoke.texi (Overall Options): Mention BRIG.
31764 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31766 2017-01-24 Richard Biener <rguenther@suse.de>
31768 PR translation/79208
31769 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31771 2017-01-24 Martin Jambor <mjambor@suse.cz>
31774 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31775 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31776 and known_contexts.
31778 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31780 PR middle-end/79123
31781 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31782 casts from signed to unsigned really don't have a range.
31784 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31786 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31787 GMP_RNDx for compatiblity.
31789 2017-01-24 Martin Liska <mliska@suse.cz>
31792 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31793 that would prevent us to call alloca with -1 as argument.
31795 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31797 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31798 Avoid -Wformat-security warning.
31800 2017-01-23 Andrew Pinski <apinski@cavium.com>
31802 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31805 2017-01-23 Martin Sebor <msebor@redhat.com>
31807 PR middle-end/78703
31808 * gimple-ssa-sprintf.c (warn_level): New global.
31809 (format_integer): Use it here and throughout the rest of the file.
31810 Use the same switch to compute sign as base.
31811 (maybe_warn): New function.
31812 (format_directive): Factor out warnings into maybe_warn.
31813 Add debugging output. Use warn_level.
31814 (add_bytes): Use warn_level.
31815 (pass_sprintf_length::compute_format_length): Add debugging output.
31816 (try_substitute_return_value): Same.
31817 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31819 PR middle-end/78703
31820 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31821 (struct fmtresult, format_integer, format_floating): Adjust.
31822 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31823 (get_string_length, format_string,format_directive): Same.
31824 (pass_sprintf_length::compute_format_length): Same.
31825 (try_substitute_return_value): Simplify slightly.
31827 PR middle-end/78703
31828 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31829 (fmtresult::operator+=): Outlined.
31830 (struct fmtresult): Add ctors.
31831 (struct conversion_spec): Rename...
31832 (struct directive): ...to this. Add and remove data members.
31833 (directive::set_width, directive::set_precision): New functions.
31834 (format_percent): Use fmtresult ctor.
31835 (get_width_and_precision): Remove.
31836 (format_integer): Make naming changes. Avoid computing width and
31838 (format_floating): Same. Adjust indentation.
31839 (format_character, format_none): New functions.
31840 (format_string): Moved character handling to format_character.
31841 (format_directive): Remove arguments, change return type.
31842 (parse_directive): New function.
31843 (pass_sprintf_length::compute_format_length): Move directive
31844 parsing to parse_directive.
31846 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31848 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31849 (assign_assembler_name_if_needed): ... this.
31850 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31851 (assign_assembler_name_if_needed): ... this.
31852 (free_lang_data_in_cgraph): Adjust callers.
31853 * cgraphunit.c (cgraph_node::analyze): Likewise.
31854 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31857 2017-01-23 Richard Biener <rguenther@suse.de>
31859 PR tree-optimization/79088
31860 PR tree-optimization/79188
31861 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31862 resetting loop bounds after last path deletion. Reset loop
31863 bounds of the target loop, make code match the comments.
31864 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31865 Make sure loops need no fixups.
31867 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31869 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31870 exponent support with double type for first argument.
31871 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31872 type returned by __builtin_vec_extract_sig,
31873 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31874 functions from "vector int" to "vector unsigned int" or from
31875 "vector long long int" to "vector unsigned long long int".
31876 Changed type returned by __builtin_vec_extract_exp,
31877 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31878 functions from "vector int" to "vector unsigned int" or from
31879 "vector long long int" to "vector unsigned long long int".
31880 Changed return type of __builtin_vec_test_data_class,
31881 __builtin_vec_test_data_class_sp, and
31882 __builtin_vec_test_data_class_dp from "vector int" to
31883 "vector bool int" or from "vector long long int" to "vector bool
31884 long long int" and changed second argument type from "unsigned
31885 int" to "int". Added new overloaded function forms "vector float
31886 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31887 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31888 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31889 double, vector unsigned long long int)" and "vector double
31890 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31891 long int)". Changed return type of
31892 __builtin_scalar_test_data_class and
31893 __builtin_scalar_test_data_class_sp and
31894 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31895 int" and changed second argument from "unsigned int" to "int".
31896 Changed type returned by __builtin_scalar_test_neg,
31897 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31898 from "int" to "bool int". Added new overloaded function form
31899 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31900 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31901 exponent double-precision with floating point first argument.
31902 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31903 documentation of scalar_test_data_class, scalar_test_neg,
31904 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31905 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31906 vec_test_data_class built-in functions to reflect refinements in
31907 their type signatures.
31909 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31911 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31913 (aarch64_elf_asm_destructor): Likewise.
31915 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31917 PR rtl-optimization/78634
31918 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31919 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31920 * ifcvt.c (noce_try_cmove): Add missing cost check.
31922 PR rtl-optimization/71724
31923 * combine.c (if_then_else_cond): Look for situations where it is
31924 beneficial to undo the work of one of the recursive calls.
31926 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31928 PR tree-optimization/70754
31929 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31930 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31931 combined stmt before it if not NULL.
31932 (combine_chains): Process refs reversely and compute dominance point
31935 2017-01-23 Martin Liska <mliska@suse.cz>
31937 PR tree-optimization/79196
31938 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31939 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31941 (strlen_optimize_stmt): Call the renamed function.
31943 2017-01-23 Michael Matz <matz@suse.de>
31945 PR tree-optimization/78384
31946 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31948 2017-01-23 Richard Biener <rguenther@suse.de>
31950 PR tree-optimization/79186
31951 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31952 both incoming edges before moving an assert.
31954 2017-01-23 Martin Jambor <mjambor@suse.cz>
31956 * ipa-prop.c (load_from_param_1): Removed.
31957 (load_from_unmodified_param): Bits from load_from_param_1 put back
31959 (load_from_param): Removed.
31960 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31961 with stmt. Reverted back to use of load_from_unmodified_param.
31963 2017-01-23 Martin Jambor <mjambor@suse.cz>
31966 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31967 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31968 field a pointer to garbage collected vector, mark lattices and
31969 ipcp_orig_node with GTY((skip)).
31970 (ipa_get_param_count): Adjust to descriptors being a pointer.
31971 (ipa_get_param): Likewise.
31972 (ipa_get_type): Likewise.
31973 (ipa_get_param_move_cost): Likewise.
31974 (ipa_set_param_used): Likewise.
31975 (ipa_get_controlled_uses): Likewise.
31976 (ipa_set_controlled_uses): Likewise.
31977 (ipa_is_param_used): Likewise.
31978 (ipa_node_params_t): Move into garbage collector. New methods insert
31980 (ipa_node_params_sum): Annotate wth GTY(()).
31981 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31983 (ipa_load_from_parm_agg): Adjust declaration.
31984 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31985 * ipa-profile.c (ipa_profile): Likewise.
31986 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31987 (ipa_populate_param_decls): Make descriptors parameter garbage
31989 (ipa_dump_param): Adjust to descriptors being a pointer.
31990 (ipa_alloc_node_params): Likewise.
31991 (ipa_initialize_node_params): Likewise.
31992 (load_from_param_1): Make descriptors parameter garbage collected.
31993 (load_from_unmodified_param): Likewise.
31994 (load_from_param): Likewise.
31995 (ipa_load_from_parm_agg): Likewise.
31996 (ipa_node_params::~ipa_node_params): Removed.
31997 (ipa_free_all_node_params): Remove call to delete operator.
31998 (ipa_node_params_t::insert): New.
31999 (ipa_node_params_t::remove): Likewise.
32000 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32001 copy known_csts and known_contexts vectors.
32002 (ipa_read_node_info): Adjust to descriptors being a pointer.
32003 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32005 (ipcp_transform_function): Make descriptors variable garbage
32008 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32010 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32011 * config/i386/avx512dqintrin.h: Ditto.
32012 * config/i386/avx512fintrin.h: Ditto.
32013 * config/i386/i386.c: Handle new builtins.
32014 * config/i386/i386-builtin.def: Add new builtins.
32015 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32016 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32018 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32019 Martin Liska <mliska@suse.cz>
32021 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32022 * asan.c (asan_expand_poison_ifn): Support stores and use
32023 appropriate ASAN report function.
32024 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32025 * internal-fn.def (ASAN_POISON_USE): Declare.
32026 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32027 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32028 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32029 ASAN_POISON calls w/o LHS.
32030 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32031 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32032 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32033 * gimplify.c (asan_poison_variables): Add attribute
32034 use_after_scope_memory to variables that really needs to live
32036 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32037 having the attribute.
32039 2017-01-23 Martin Liska <mliska@suse.cz>
32041 * asan.c (create_asan_shadow_var): New function.
32042 (asan_expand_poison_ifn): Likewise.
32043 * asan.h (asan_expand_poison_ifn): New declaration.
32044 * internal-fn.c (expand_ASAN_POISON): Likewise.
32045 * internal-fn.def (ASAN_POISON): New builtin.
32046 * sanopt.c (pass_sanopt::execute): Expand
32047 asan_expand_poison_ifn.
32048 * tree-inline.c (copy_decl_for_dup_finish): Make function
32050 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32051 * tree-ssa.c (is_asan_mark_p): New function.
32052 (execute_update_addresses_taken): Rewrite local variables
32053 (identified just by use-after-scope as addressable) into SSA.
32055 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32057 * doc/install.texi (Specific): opensource.apple.com uses https
32058 now. Remove trailing slash.
32060 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32062 * README.Portability: Remove note on an Irix compatibility issue.
32064 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32066 * gcov.c (INCLUDE_ALGORITHM): Define.
32067 (INCLUDE_VECTOR): Define.
32068 No longer include <vector> and <algorithm> directly.
32070 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32072 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32074 * doc/invoke.texi (Code Gen Options): Ditto.
32076 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32079 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32081 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32083 rtl-optimization/79125
32084 * cprop.c (local_cprop_pass): Handle cases where we make an
32085 unconditional trap.
32087 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32091 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32092 read from, for big endian.
32094 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32096 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32097 register pauth builtins for LP64 only.
32099 2017-01-20 Marek Polacek <polacek@redhat.com>
32102 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32105 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32107 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32109 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32110 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32111 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32113 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32116 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32117 in big-endian mode when they are not a single duplicated value.
32119 2017-01-20 Richard Biener <rguenther@suse.de>
32121 * BASE-VER: Bump to 7.0.1.
32123 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32125 * omp-low.c (omplow_simd_context): New struct. Use it...
32126 (lower_rec_simd_input_clauses): ...here and...
32127 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32128 references to idx, lane, max_vf, is_simt.
32130 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32132 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32135 2017-01-20 Martin Jambor <mjambor@suse.cz>
32137 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32138 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32140 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32141 (GTFILES): Rename hsa.c to hsa-common.c.
32142 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32143 * hsa-dump.c: Likewise.
32144 * hsa-gen.c: Likewise.
32145 * hsa-regalloc.c: Likewise.
32146 * ipa-hsa.c: Likewise.
32147 * omp-expand.c: Likewise.
32148 * omp-low.c: Likewise.
32149 * toplev.c: Likewise.
32151 2017-01-20 Marek Polacek <polacek@redhat.com>
32154 * doc/invoke.texi: Document -Wduplicated-branches.
32155 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32156 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32157 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32158 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32159 return 0 only when not OEP_LEXICOGRAPHIC.
32160 (fold_build_cleanup_point_expr): Use the expression
32161 location when building CLEANUP_POINT_EXPR.
32162 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32163 * tree.c (add_expr): Handle error_mark_node.
32165 2017-01-20 Martin Liska <mliska@suse.cz>
32168 * tree-profile.c (init_ic_make_global_vars): Do not call
32170 (gimple_init_gcov_profiler): Likewise.
32172 2017-01-20 Martin Liska <mliska@suse.cz>
32175 * cgraph.h (maybe_create_reference): Remove argument and
32177 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32179 * ipa-cp.c (create_specialized_node): Likewise.
32180 * symtab.c (symtab_node::maybe_create_reference): Handle
32181 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32183 2017-01-20 Martin Liska <mliska@suse.cz>
32185 * read-rtl-function.c (function_reader::create_function): Use
32186 build_decl instread of build_decl_stat.
32188 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32190 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32191 * config/i386/avx512dqintrin.h: Ditto.
32192 * config/i386/avx512fintrin.h: Ditto.
32193 * config/i386/i386-builtin-types.def: Add new types.
32194 * config/i386/i386.c: Handle new types.
32195 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32196 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32197 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32198 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32199 (__builtin_ia32_kshiftridi): New.
32200 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32202 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32206 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32207 define to rs6000_init_stack_protect_guard.
32208 (rs6000_init_stack_protect_guard): New function.
32210 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32211 Yunqiang Su <yunqiang.su@imgtec.com>
32213 * config.gcc (supported_defaults): Add madd4.
32214 (with_madd4): Add validation.
32215 (all_defaults): Add madd4.
32216 * config/mips/mips.opt (mmadd4): New option.
32217 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32219 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32221 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32222 (ISA_HAS_FUSED_MADD4): Likewise.
32223 * doc/invoke.texi (-mmadd4): Document the new option.
32224 * doc/install.texi (--with-madd4): Document the new option.
32226 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32228 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32229 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32230 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32231 (aarch64_init_pauth_hint_builtins): New.
32232 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32233 (aarch64_expand_builtin): Expand new builtins.
32235 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32237 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32238 * combine-stack-adj.c (no_unhandled_cfa): Handle
32239 REG_CFA_TOGGLE_RA_MANGLE.
32240 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32241 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32242 info for return address signing.
32243 (aarch64_expand_epilogue): Likewise.
32245 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32247 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32248 * config/aarch64/aarch64-protos.h
32249 (aarch64_return_address_signing_enabled): New declaration.
32250 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32252 (aarch64_expand_prologue): Sign return address before it's pushed onto
32254 (aarch64_expand_epilogue): Authenticate return address fetched from
32256 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32257 (aarch64_attributes): New function attributes for "sign-return-address".
32258 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32259 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32260 ("*do_return"): Generate combined instructions according to key index.
32261 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32262 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32264 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32265 * config/aarch64/aarch64.opt (msign-return-address=): New.
32266 * doc/extend.texi (AArch64 Function Attributes): Documents
32267 "sign-return-address=".
32268 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32270 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32272 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32273 overall option summary.
32275 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32277 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32278 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32279 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32280 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32282 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32284 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32285 -mpower9-minmax by default for -mcpu=power9.
32286 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32287 128-bit floating point.
32289 2017-01-20 Alan Modra <amodra@gmail.com>
32291 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32292 optimizing for size.
32294 2017-01-20 Alan Modra <amodra@gmail.com>
32297 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32298 for strcmp and strncmp from corresponding builtin decl.
32300 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32302 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32303 instead of i386/rtems-64.h.
32304 * config/i386/rtems-64.h: Remove.
32306 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32310 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32312 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32314 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32316 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32317 Change int to HOST_WIDE_INT.
32318 * config/aarch64/aarch64-protos.h
32319 (aarch64_simd_gen_const_vector_dup): Likewise.
32320 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32322 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32324 * langhooks-def.h (lhd_type_for_size): New decl.
32325 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32326 * langhooks.c (lhd_type_for_size): New function, taken from
32329 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32331 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32332 define_bypass for CR latency.
32333 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32334 (power9-alu2): Add define_bypass for CR latency.
32336 (power9-mul): Update insn latency.
32337 (power9-mul-compare): Update insn latency, bypass latency and remove
32340 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32342 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32344 * config/aarch64/aarch64.md
32345 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32346 aarch64_nopcrelative_literal_loads.
32347 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32349 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32351 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32352 TARGET_LOONGSON_3A.
32353 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32355 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32358 * config.gcc (supported_defaults): Add lxc1-sxc1.
32359 (with_lxc1_sxc1): Add validation.
32360 (all_defaults): Add lxc1-sxc1.
32361 * config/mips/mips.opt (mlxc1-sxc1): New option.
32362 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32364 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32365 __mips_no_lxc1_sxc1.
32366 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32367 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32368 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32370 2017-01-19 Richard Biener <rguenther@suse.de>
32372 PR tree-optimization/72488
32373 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32374 sure to restore SSA info.
32375 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32377 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32379 PR rtl-optimization/79121
32380 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32381 of the inner type when shifting an extended value.
32383 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32386 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32387 interposable aliases.
32389 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32392 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32393 Use the evmergelohi instruction.
32394 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32395 (mov_sitf_e500_subreg8_2_be): Likewise.
32396 (mov_sitf_e500_subreg12_2_le): Likewise.
32397 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32398 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32399 (mov_sitf_e500_subreg8_2_le): Likewise.
32400 (mov_sitf_e500_subreg12_2_be): Likewise.
32402 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32404 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32405 attribute from vecsimple to vecperm.
32406 (altivec_vbpermq2): Likewise.
32408 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32411 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32413 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32414 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32415 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32416 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32417 case where N arg is SIZE_MAX.
32418 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32419 (cmpstrsi): Add pattern.
32421 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32423 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32424 __builtin_vec_revb builtins.
32425 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32426 built-in functions to support generation of the ISA 3.0 XXBR<x>
32427 vector byte reverse instructions.
32428 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32429 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32430 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32431 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32432 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32433 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32434 (P9V_BUILTIN_VEC_REVB): Likewise.
32435 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32436 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32437 (p9_xxbrq_v16qi): Likewise.
32438 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32439 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32440 (p9_xxbrh_v8hi): Likewise.
32441 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32442 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32443 vec_revb built-in functions.
32445 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32447 PR rtl-optimization/78952
32448 * config/i386/i386.md (any_extract): New code iterator.
32449 (*insvqi_2): Use any_extract for source operand.
32450 (*insvqi_3): Use any_shiftrt for source operand.
32452 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32454 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32456 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32458 2017-01-18 Matthias Klose <doko@ubuntu.com>
32460 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32462 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32464 * config/rs6000/altivec.h (vec_bperm): Change #define.
32465 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32466 (altivec_vbpermq2): New define_insn.
32467 (altivec_vbpermd): Likewise.
32468 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32469 function interface.
32470 (VBPERMD): Likewise.
32471 (VBPERM): New polymorphic function interface.
32472 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32473 Add entries for P9V_BUILTIN_VEC_VBPERM.
32474 * doc/extend.texi: Add interfaces for vec_bperm.
32476 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32478 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32479 first letter of error messages.
32480 (s390_resolve_overloaded_builtin): Likewise.
32481 * config/s390/s390.c (s390_expand_builtin): Likewise.
32482 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32483 (s390_valid_target_attribute_inner_p): Likewise.
32484 * config/s390/s390.md ("tabort"): Likewise.
32486 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32488 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32489 (ISA_AVOID_DIV_HILO): New macro.
32490 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32491 (ISA_HAS_DDIV): Likewise.
32493 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32495 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32497 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32499 * doc/invoke.texi (fabi-version): Spelling fix.
32501 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32504 * doc/invoke.texi (fabi-version): Mention mangling fix for
32507 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32510 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32512 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32515 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32516 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32518 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32520 (stack_protect_test): Ditto.
32521 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32522 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32524 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32525 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32526 -mstack-protector-guard-offset=.
32527 (RS/6000 and PowerPC Options): Ditto.
32529 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32531 * config/i386/i386.h (MASK_CLASS_P): New define.
32532 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32533 there are no registers from different register sets also when
32534 mask registers are used. Update function comment.
32535 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32536 to (*k/*r) and (*k/*km) alternatives.
32538 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32540 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32541 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32542 (EH_RETURN_HANDLER_RTX): New define.
32543 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32544 Force frame pointer in EH return functions.
32545 (aarch64_expand_epilogue): Add barrier for eh_return.
32546 (aarch64_final_eh_return_addr): Remove.
32547 (aarch64_eh_return_handler_rtx): New function.
32548 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32550 (aarch64_eh_return_handler_rtx): New prototype.
32552 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32554 * config/rs6000/altivec.h (vec_rlmi): New #define.
32555 (vec_vrlnm): Likewise.
32556 (vec_rlnm): Likewise.
32557 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32558 (UNSPEC_VRLNM): Likewise.
32559 (VIlong): New mode iterator.
32560 (altivec_vrl<VI_char>mi): New define_insn.
32561 (altivec_vrl<VI_char>nm): Likewise.
32562 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32564 (VRLDNM): Likewise.
32565 (RLNM): New polymorphic function entry.
32566 (VRLWMI): New monomorphic function entry.
32567 (VRLDMI): Likewise.
32568 (RLMI): New polymorphic function entry.
32569 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32570 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32571 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32574 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32577 * dwarf2out.c (field_byte_offset): Restore the
32578 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32579 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32580 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32583 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32586 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32588 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32591 * dwarf2out.c (add_data_member_location_attribute): For constant
32592 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32593 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32594 DW_AT_byte_size attributes.
32596 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32598 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32599 after forcing to constant memory when the code model is medium.
32601 2017-01-17 Julia Koval <julia.koval@intel.com>
32604 * config/i386/avx512fintrin.h
32605 (_mm512_i32gather_ps): Change __addr type to void const*.
32606 (_mm512_mask_i32gather_ps): Ditto.
32607 (_mm512_i32gather_pd): Ditto.
32608 (_mm512_mask_i32gather_pd): Ditto.
32609 (_mm512_i64gather_ps): Ditto.
32610 (_mm512_mask_i64gather_ps): Ditto.
32611 (_mm512_i64gather_pd): Ditto.
32612 (_mm512_mask_i64gather_pd): Ditto.
32613 (_mm512_i32gather_epi32): Ditto.
32614 (_mm512_mask_i32gather_epi32): Ditto.
32615 (_mm512_i32gather_epi64): Ditto.
32616 (_mm512_mask_i32gather_epi64): Ditto.
32617 (_mm512_i64gather_epi32): Ditto.
32618 (_mm512_mask_i64gather_epi32): Ditto.
32619 (_mm512_i64gather_epi64): Ditto.
32620 (_mm512_mask_i64gather_epi64): Ditto.
32621 (_mm512_i32scatter_ps): Change __addr type to void*.
32622 (_mm512_mask_i32scatter_ps): Ditto.
32623 (_mm512_i32scatter_pd): Ditto.
32624 (_mm512_mask_i32scatter_pd): Ditto.
32625 (_mm512_i64scatter_ps): Ditto.
32626 (_mm512_mask_i64scatter_ps): Ditto.
32627 (_mm512_i64scatter_pd): Ditto.
32628 (_mm512_mask_i64scatter_pd): Ditto.
32629 (_mm512_i32scatter_epi32): Ditto.
32630 (_mm512_mask_i32scatter_epi32): Ditto.
32631 (_mm512_i32scatter_epi64): Ditto.
32632 (_mm512_mask_i32scatter_epi64): Ditto.
32633 (_mm512_i64scatter_epi32): Ditto.
32634 (_mm512_mask_i64scatter_epi32): Ditto.
32635 (_mm512_i64scatter_epi64): Ditto.
32636 (_mm512_mask_i64scatter_epi64): Ditto.
32637 * config/i386/avx512pfintrin.h
32638 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32639 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32640 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32641 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32642 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32643 (_mm512_prefetch_i32scatter_ps): Ditto.
32644 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32645 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32646 (_mm512_prefetch_i64scatter_pd): Ditto.
32647 (_mm512_prefetch_i64scatter_ps): Ditto.
32648 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32649 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32650 * config/i386/avx512vlintrin.h
32651 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32652 (_mm_mmask_i32gather_ps): Ditto.
32653 (_mm256_mmask_i32gather_pd): Ditto.
32654 (_mm_mmask_i32gather_pd): Ditto.
32655 (_mm256_mmask_i64gather_ps): Ditto.
32656 (_mm_mmask_i64gather_ps): Ditto.
32657 (_mm256_mmask_i64gather_pd): Ditto.
32658 (_mm_mmask_i64gather_pd): Ditto.
32659 (_mm256_mmask_i32gather_epi32): Ditto.
32660 (_mm_mmask_i32gather_epi32): Ditto.
32661 (_mm256_mmask_i32gather_epi64): Ditto.
32662 (_mm_mmask_i32gather_epi64): Ditto.
32663 (_mm256_mmask_i64gather_epi32): Ditto.
32664 (_mm_mmask_i64gather_epi32): Ditto.
32665 (_mm256_mmask_i64gather_epi64): Ditto.
32666 (_mm_mmask_i64gather_epi64): Ditto.
32667 (_mm256_i32scatter_ps): Change __addr type to void*.
32668 (_mm256_mask_i32scatter_ps): Ditto.
32669 (_mm_i32scatter_ps): Ditto.
32670 (_mm_mask_i32scatter_ps): Ditto.
32671 (_mm256_i32scatter_pd): Ditto.
32672 (_mm256_mask_i32scatter_pd): Ditto.
32673 (_mm_i32scatter_pd): Ditto.
32674 (_mm_mask_i32scatter_pd): Ditto.
32675 (_mm256_i64scatter_ps): Ditto.
32676 (_mm256_mask_i64scatter_ps): Ditto.
32677 (_mm_i64scatter_ps): Ditto.
32678 (_mm_mask_i64scatter_ps): Ditto.
32679 (_mm256_i64scatter_pd): Ditto.
32680 (_mm256_mask_i64scatter_pd): Ditto.
32681 (_mm_i64scatter_pd): Ditto.
32682 (_mm_mask_i64scatter_pd): Ditto.
32683 (_mm256_i32scatter_epi32): Ditto.
32684 (_mm256_mask_i32scatter_epi32): Ditto.
32685 (_mm_i32scatter_epi32): Ditto.
32686 (_mm_mask_i32scatter_epi32): Ditto.
32687 (_mm256_i32scatter_epi64): Ditto.
32688 (_mm256_mask_i32scatter_epi64): Ditto.
32689 (_mm_i32scatter_epi64): Ditto.
32690 (_mm_mask_i32scatter_epi64): Ditto.
32691 (_mm256_i64scatter_epi32): Ditto.
32692 (_mm256_mask_i64scatter_epi32): Ditto.
32693 (_mm_i64scatter_epi32): Ditto.
32694 (_mm_mask_i64scatter_epi32): Ditto.
32695 (_mm256_i64scatter_epi64): Ditto.
32696 (_mm256_mask_i64scatter_epi64): Ditto.
32697 (_mm_i64scatter_epi64): Ditto.
32698 (_mm_mask_i64scatter_epi64): Ditto.
32699 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32700 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32701 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32702 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32703 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32704 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32705 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32706 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32707 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32708 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32709 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32710 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32711 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32712 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32713 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32714 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32715 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32716 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32717 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32718 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32719 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32720 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32721 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32722 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32723 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32724 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32725 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32726 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32727 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32728 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32729 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32730 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32731 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32732 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32733 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32734 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32735 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32736 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32737 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32738 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32739 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32740 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32741 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32742 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32743 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32744 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32745 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32746 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32747 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32748 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32749 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32750 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32751 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32752 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32753 definitions accordingly.
32755 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32756 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32759 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32762 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32765 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32766 hard regs for inner regno.
32768 2017-01-17 Martin Liska <mliska@suse.cz>
32771 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32772 assumption and add comment.
32774 2017-01-17 Nathan Sidwell <nathan@acm.org>
32776 * ipa-visibility.c (localize_node): New function, broken out of ...
32777 (function_and_variable_visibility): ... here. Call it.
32779 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32781 PR middle-end/77445
32782 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32783 correctly set frequency of oudgoing edge.
32784 (duplicate_thread_path): Fix profile updating.
32786 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32789 * configure.ac: Add GCC_BASE_VER.
32790 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32791 version from BASE-VER file.
32792 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32793 (gcc.o): Depend on $(BASEVER).
32794 * common.opt (dumpfullversion): New option.
32795 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32796 * doc/invoke.texi: Document -dumpfullversion.
32797 * doc/install.texi: Document --with-gcc-major-version-only.
32798 * configure: Regenerated.
32800 2017-01-17 Richard Biener <rguenther@suse.de>
32802 PR tree-optimization/71433
32803 * tree-vrp.c (register_new_assert_for): Merge same asserts
32804 on all incoming edges.
32805 (process_assert_insertions_for): Handle insertions at the
32808 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32810 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32811 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32813 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32816 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32819 2017-01-17 Alan Modra <amodra@gmail.com>
32822 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32823 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32824 symbolic stack limit when pic.
32826 2017-01-16 Martin Sebor <msebor@redhat.com>
32828 PR tree-optimization/78608
32829 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32831 2017-01-16 Jeff Law <law@redhat.com>
32834 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32835 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32836 for several include directories that may be relative to sysroot.
32837 * config/i386/x-mingw32 (gplus_includedir): Define.
32838 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32839 (native_system_includedir): Likewise.
32840 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32841 override if TARGET_SYSTEM_ROOT is defined.
32842 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32844 PR tree-optimization/79090
32845 PR tree-optimization/33562
32846 PR tree-optimization/61912
32847 PR tree-optimization/77485
32848 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32849 and computed trims into the dump file.
32851 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32853 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32855 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32858 * gimplify.c (gimplify_init_constructor): If want_value and
32859 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32863 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32864 sequence. Formatting fixes.
32865 (doloop_optimize): Formatting fixes.
32868 * gcc.c (debug_level_greater_than_spec_func): New function.
32869 (static_spec_functions): Add debug-level-gt spec function.
32870 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32872 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32873 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32874 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32875 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32876 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32877 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32879 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32881 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32882 QImode fixups to general and mask registers only.
32884 2017-01-16 Carl Love <cel@us.ibm.com>
32886 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32887 for built-in functions
32888 vector signed char vec_nabs (vector signed char)
32889 vector signed short vec_nabs (vector signed short)
32890 vector signed int vec_nabs (vector signed int)
32891 vector signed long long vec_nabs (vector signed long long)
32892 vector float vec_nabs (vector float)
32893 vector double vec_nabs (vector double)
32894 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32896 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32897 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32898 * doc/extend.texi: Update the documentation file for the new built-in
32901 2017-01-16 Martin Sebor <msebor@redhat.com>
32903 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32906 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32908 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32909 UNSPEC_VSX__XXSPLTD to require special splat handling.
32911 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32914 * system.h: Poison strndup.
32916 2017-01-16 Alan Modra <amodra@gmail.com>
32919 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32922 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32924 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32926 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32928 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32929 call recog here. Assert that INSN_CODE (insn) is non-negative.
32931 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32934 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32936 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32937 in the currently scheduled RTL fragment.
32939 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32941 PR rtl-optimization/78751
32942 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32945 2017-01-14 Jeff Law <law@redhat.com>
32947 PR tree-optimization/79090
32948 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32949 variable length stores.
32950 (compute_trims): Delete dead assignment to *trim_tail.
32951 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32954 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32956 PR rtl-optimization/78626
32957 PR rtl-optimization/78727
32958 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32959 of a block, and split such blocks after everything else is finished.
32961 2017-01-14 Alan Modra <amodra@gmail.com>
32964 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32965 target legitimate_combined_insn.
32966 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32967 (rs6000_legitimate_combined_insn): New function.
32968 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32970 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32971 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32972 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32974 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32976 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32978 2017-01-13 Jeff Law <law@redhat.com>
32980 PR tree-optimization/33562
32981 PR tree-optimization/61912
32982 PR tree-optimization/77485
32983 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32985 (delete_dead_assignment): Likewise.
32986 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32987 statement to delete_dead_call and delete_dead_assignment.
32989 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32992 * substring-locations.c (format_warning_va): Strengthen case 1 so
32993 that both endpoints of the substring must be within the format
32994 range for just the substring to be printed.
32996 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32998 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32999 * config/i386/i386.c (ix86_target_string): Add missing options
33000 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33001 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33002 flags_other and ix86_target_other to flags2_other. Display unknown
33004 (ix86_valid_target_attribute_inner_p): Add missing options and
33005 reorder options by implied ISAs, as in ix86_target_string.
33007 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33009 * hash-table.h (hash_table::too_empty_p): New function.
33010 (hash_table::expand): Use it.
33011 (hash_table::traverse): Likewise.
33012 (hash_table::empty_slot): Use sizeof (value_type) instead of
33013 sizeof (PTR) to convert bytes to elements. Shrink the table
33014 if the current size is excessive for the current number of
33017 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33019 * ira-costs.c (record_reg_classes): Break from the inner loop
33020 early once alt_fail is known to be true. Update outer loop
33021 handling accordingly.
33023 2017-01-13 Jeff Law <law@redhat.com>
33025 * tree-ssa-dse.c (decrement_count): New function.
33026 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33027 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33028 when we know the partially dead statement is a mem* function.
33030 PR tree-optimization/61912
33031 PR tree-optimization/77485
33032 * tree-ssa-dse.c: Include expr.h.
33033 (maybe_trim_constructor_store): New function.
33034 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33036 PR tree-optimization/33562
33037 PR tree-optimization/61912
33038 PR tree-optimization/77485
33039 * doc/invoke.texi: Document new dse-max-object-size param.
33040 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33041 * tree-ssa-dse.c: Include params.h.
33042 (dse_store_status): New enum.
33043 (initialize_ao_ref_for_dse): New, partially extracted from
33045 (valid_ao_ref_for_dse, normalize_ref): New.
33046 (setup_live_bytes_from_ref, compute_trims): Likewise.
33047 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33048 (maybe_trim_partially_dead_store): Likewise.
33049 (maybe_trim_complex_store): Likewise.
33050 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33051 Track what bytes live from the original store. Return tri-state
33052 for dead, partially dead or live.
33053 (dse_dom_walker): Add constructor, destructor and new private members.
33054 (delete_dead_call, delete_dead_assignment): New extracted from
33056 (dse_optimize_stmt): Make a member of dse_dom_walker.
33057 Use initialize_ao_ref_for_dse.
33059 PR tree-optimization/33562
33060 PR tree-optimization/61912
33061 PR tree-optimization/77485
33062 * sbitmap.h (bitmap_count_bits): Prototype.
33063 (bitmap_clear_range, bitmap_set_range): Likewise.
33064 * sbitmap.c (bitmap_clear_range): New function.
33065 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33067 2017-01-13 Martin Liska <mliska@suse.cz>
33070 * function.c (set_cfun): Add new argument force.
33071 * function.h (set_cfun): Likewise.
33072 * ipa-inline-transform.c (inline_call): Use the function when
33073 strict alising from is dropped for function we inline to.
33075 2017-01-13 Richard Biener <rguenther@suse.de>
33077 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33078 for dumping GIMPLE INTEGER_CSTs.
33080 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33082 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33083 to 201112L since C++17.
33085 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33088 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33089 if -fsanitize=kernel-address is present.
33091 2017-01-13 Richard Biener <rguenther@suse.de>
33093 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33094 as _Literal ( type ) number in case usual suffixes do not
33095 preserve all information.
33097 2017-01-13 Richard Biener <rguenther@suse.de>
33099 PR tree-optimization/77283
33100 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33101 and ssa-iterators.h.
33102 (is_feasible_trace): Implement a cost model based on joiner
33105 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33108 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33109 char or short to __float128/_Float128 directly.
33111 2017-01-12 Martin Sebor <msebor@redhat.com>
33113 to -Wformat-overflow.
33114 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33115 (min_bytes_remaining): Same.
33116 (get_string_length): Same.
33117 (format_string): Same.
33118 (format_directive): Same.
33120 (pass_sprintf_length::handle_gimple_call): Same.
33122 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33124 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33125 info.nowrite calls with no lhs that can't throw. Return bool
33126 whether gsi_remove has been called or not.
33127 (pass_sprintf_length::handle_gimple_call): Return bool whether
33128 try_substitute_return_value called gsi_remove. Formatting fix.
33129 (pass_sprintf_length::execute): Don't use gsi_remove if
33130 handle_gimple_call returned true.
33133 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33134 be removed due to side-effects, don't remove following barrier nor
33135 turn the successor edge into fallthru edge.
33137 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33140 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33141 element-reversing loads and stores as not swappable.
33143 2017-01-12 Nathan Sidwell <nathan@acm.org>
33144 Nicolai Stange <nicstange@gmail.com>
33146 * combine.c (try_combine): Don't ignore result of overlap checking
33147 loop. Combine overlap & asm check into single loop.
33149 2017-01-12 Richard Biener <rguenther@suse.de>
33151 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33152 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33154 2017-01-12 Richard Biener <rguenther@suse.de>
33156 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33157 and TS_TARGET_OPTION directly derive from TS_BASE.
33158 * tree-core.h (tree_optimization_option): Derive from tree_base.
33159 (tree_target_option): Likewise.
33161 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33163 * config/i386/i386.c (memory_address_length): Increase len
33164 only when rip_relative_addr_p returns false.
33166 2017-01-11 Julia Koval <julia.koval@intel.com>
33168 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33169 (OPTION_MASK_ISA_SGX_SET): New.
33170 (ix86_handle_option): Handle OPT_msgx.
33171 * config.gcc: Added sgxintrin.h.
33172 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33173 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33174 * config/i386/i386.c (ix86_target_string): Add -msgx.
33176 (ix86_option_override_internal): Handle new options.
33177 (ix86_valid_target_attribute_inner_p): Add sgx.
33178 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33179 * config/i386/i386.opt: Add msgx.
33180 * config/i386/sgxintrin.h: New file.
33181 * config/i386/x86intrin.h: Add sgxintrin.h.
33183 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33186 * fold-const.c (maybe_nonzero_address): Return 1 for function
33188 (tree_single_nonzero_warnv_p): Don't handle function local objects
33192 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33195 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33198 * opts.c: Include "spellcheck.h"
33199 (struct string_fragment): New struct.
33200 (struct edit_distance_traits<const string_fragment &>): New
33202 (get_closest_sanitizer_option): New function.
33203 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33205 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33207 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33209 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33210 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33211 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33212 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33213 for initial die_offset if dwarf_split_debug_info.
33214 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33215 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33217 (output_skeleton_debug_sections): Formatting fix. Use
33218 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33219 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33221 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33223 * config/arm/cortex-a53.md: Add bypasses for
33224 cortex_a53_r2f_cvt.
33225 (cortex_a53_r2f): Only use for transfers.
33226 (cortex_a53_f2r): Likewise.
33227 (cortex_a53_r2f_cvt): Add reservation for conversions.
33228 (cortex_a53_f2r_cvt): Likewise.
33230 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33232 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33233 to all inlined functions, change static to extern.
33235 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33238 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33240 (arm_assemble_integer): Likewise.
33242 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33244 * config.gcc: Use new awk script to check CPU, FPU and architecture
33245 parameters for --with-... options.
33246 * config/arm/parsecpu.awk: New file
33247 * config/arm/arm-cpus.in: New file.
33248 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33250 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33252 * config/arm/t-arm: Update dependency rules.
33253 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33254 of processing .def files.
33255 * config/arm/genopt.sh: Deleted.
33256 * config/arm/gentune.sh: Deleted.
33257 * config/arm/arm-cores.def: Deleted.
33258 * config/arm/arm-arches.def: Deleted.
33259 * config/arm/arm-fpus.def: Deleted.
33260 * config/arm/arm-tune.md: Regenerated.
33261 * config/arm/arm-tables.opt: Regenerated.
33262 * config/arm/arm-cpu.h: New generated file.
33263 * config/arm/arm-cpu-data.h: New generated file.
33264 * config/arm/arm-cpu-cdata.h: New generated file.
33266 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33269 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33271 (input_varpool_node): Unpack dynamically_initialized bit.
33273 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33275 PR rtl-optimization/79032
33276 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33277 the alignment of the adjusted memory reference against that of MODE,
33278 instead of the alignment of the original memory reference.
33280 2017-01-11 Martin Jambor <mjambor@suse.cz>
33282 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33284 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33285 decorated functions.
33287 2017-01-11 Richard Biener <rguenther@suse.de>
33289 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33290 set range/nonnull info for PHI results. Do not set it on
33291 stmts marked for removal.
33293 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33295 * expr.c (store_field): In the bitfield case, fetch the return value
33296 from the registers before applying a single big-endian adjustment.
33297 Always do a final load for a BLKmode value not larger than a word.
33299 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33302 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33303 that we correctly handle column numbers greater than
33304 LINE_MAP_MAX_COLUMN_NUMBER.
33306 2017-01-10 Martin Sebor <msebor@redhat.com>
33308 PR middle-end/78245
33309 * gimple-ssa-sprintf.c (get_destination_size): Call
33310 {init,fini}object_sizes.
33311 * tree-object-size.c (addr_object_size): Adjust.
33312 (pass_through_call): Adjust.
33313 (pass_object_sizes::execute): Adjust.
33314 * tree-object-size.h (fini_object_sizes): Declare.
33316 2017-01-10 Martin Sebor <msebor@redhat.com>
33318 PR tree-optimization/78775
33319 * builtins.c (get_size_range): Move...
33320 * calls.c: ...to here.
33321 (alloc_max_size): Accept zero argument.
33322 (operand_signed_p): Remove.
33323 (maybe_warn_alloc_args_overflow): Call get_size_range.
33324 * calls.h (get_size_range): Declare.
33326 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33328 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33329 from TI's devices.csv file as of September 2016.
33330 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33332 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33334 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33335 * doc/invoke.texi: Likewise.
33336 * doc/md.texi: Likewise.
33337 * doc/objc.texi: Likewise.
33339 2017-01-10 Joshua Conner <joshconner@google.com>
33341 * config/arm/fuchsia-elf.h: New file.
33342 * config/fuchsia.h: New file.
33343 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33344 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33346 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33348 2016-01-10 Richard Biener <rguenther@suse.de>
33350 PR tree-optimization/79034
33351 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33352 Propagate out degenerate PHIs in the joiner.
33354 2017-01-10 Martin Liska <mliska@suse.cz>
33356 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33357 (sort_congruence_classes_by_decl_uid): Likewise.
33358 (sort_congruence_class_groups_by_decl_uid): Likewise.
33359 (sem_item_optimizer::merge_classes): Sort class, groups in these
33360 classes and members in the groups by DECL_UID of declarations.
33361 This would make merge operations stable.
33363 2017-01-10 Martin Liska <mliska@suse.cz>
33365 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33366 usage of m_classes_vec.
33367 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33368 (sem_item_optimizer::get_group_by_hash): Likewise.
33369 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33370 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33371 (sem_item_optimizer::verify_classes): Likewise.
33372 (sem_item_optimizer::process_cong_reduction): Likewise.
33373 (sem_item_optimizer::dump_cong_classes): Likewise.
33374 (sem_item_optimizer::merge_classes): Likewise.
33375 * ipa-icf.h (congruence_class_hash): Rename from
33376 congruence_class_group_hash. Remove declaration of m_classes_vec.
33378 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33380 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33381 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33382 * config.gcc: Add avx512vpopcntdqintrin.h.
33383 * config/i386/avx512vpopcntdqintrin.h: New.
33384 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33385 * config/i386/i386-builtin-types.def: Add new types.
33386 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33387 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33388 __builtin_ia32_vpopcountq_v8di_mask): New.
33389 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33390 __AVX512VPOPCNTDQ__.
33391 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33392 (PTA_AVX512VPOPCNTDQ): Define.
33393 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33394 TARGET_AVX512VPOPCNTDQ_P): Define.
33395 * config/i386/i386.opt: Add mavx512vpopcntdq.
33396 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33397 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33399 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33401 PR middle-end/77484
33402 * predict.def (PRED_CALL): Set to 67.
33404 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33406 * expr.c (store_field): In the bitfield case, if the value comes from
33407 a function call and is of an aggregate type returned in registers, do
33408 not modify the field mode; extract the value in all cases if the mode
33409 is BLKmode and the size is not larger than a word.
33411 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33414 * config/i386/cpuid.h: Fix undefined behavior.
33416 2017-01-04 Jeff Law <law@redhat.com>
33418 PR tree-optimization/79007
33419 PR tree-optimization/67955
33420 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33421 conservative for pt.null when flag_non_call_exceptions is on.
33423 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33425 PR translation/79019
33426 PR translation/79020
33427 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33428 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33429 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33431 * config/avr/avr.opt (maccumulate-args): Likewise.
33432 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33433 * common.opt (freport-bug): Likewise.
33434 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33435 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33436 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33437 translatable string.
33438 * config/i386/i386.c (function_value_32): Likewise.
33439 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33440 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33442 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33443 * common/config/msp430/msp430-common.c (msp430_handle_option):
33445 * symtab.c (symtab_node::verify_base): Likewise.
33446 * opts.c (set_debug_level): Likewise.
33447 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33448 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33449 missing whitespace to translatable strings.
33450 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33451 * config/sh/superh.h: Likewise.
33452 * config/i386/xopintrin.h: Likewise.
33453 * config/i386/znver1.md: Likewise.
33454 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33455 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33456 * double-int.h (struct double_int): Likewise.
33457 * double-int.c (div_and_round_double): Likewise.
33458 * wide-int.cc: Likewise.
33459 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33460 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33461 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33462 (crossjumps_occurred): ... this.
33463 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33466 PR tree-optimization/78899
33467 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33468 returning bool return struct loop *, NULL for failure and the new
33470 (versionable_outer_loop_p): Don't version outer loop if it has
33471 dont_vectorized bit set.
33472 (tree_if_conversion): When versioning outer loop, ensure
33473 tree_if_conversion is performed also on the inner loop of the
33474 non-vectorizable outer loop copy.
33475 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33476 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33477 prevent vectorization of it.
33478 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33479 the outer loop vectorization of the non-scalar version is attempted
33480 before vectorization of the inner loop in scalar version. If
33481 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33482 vectorization of its inner loop.
33483 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33484 has 2 inner loops, rename also on edges from bb whose single pred
33485 is outer_loop->header. Fix typo in function comment.
33487 2017-01-09 Martin Sebor <msebor@redhat.com>
33490 * asan.c (asan_emit_stack_protection): Increase local buffer size
33491 to avoid snprintf truncation warning.
33493 2017-01-09 Andrew Pinski <apinski@cavium.com>
33495 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33496 to reference thunderx2t99 for the tuning structure
33497 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33499 (thunderx2t99_extra_costs): This.
33500 * config/aarch64/aarch64-tune.md: Regenerate.
33501 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33502 (vulcan_addrcost_table): This.
33503 (vulcan_regmove_cost): Rename to ...
33504 (thunderx2t99_regmove_cost): This.
33505 (vulcan_vector_cost): Rename to ...
33506 (thunderx2t99_vector_cost): this.
33507 (vulcan_branch_cost): Rename to ...
33508 (thunderx2t99_branch_cost): This.
33509 (vulcan_tunings): Rename to ...
33510 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33511 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33513 2017-01-09 Martin Jambor <mjambor@suse.cz>
33517 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33518 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33519 (propagate_vr_accross_jump_function): Use the above function for all
33520 value range computations for pass-through jump functions and type
33521 converasion from explicit value range values.
33522 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33523 parameters from TYPE_ARG_TYPES.
33524 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33525 (ipa_write_node_info): Stream type of the actual argument.
33526 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33528 2017-01-09 Martin Liska <mliska@suse.cz>
33531 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33532 (lookup_compiler): Do not show error message with have_E.
33534 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33536 PR tree-optimization/78938
33537 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33538 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33539 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33540 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33543 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33545 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33548 2017-01-09 Richard Biener <rguenther@suse.de>
33550 PR tree-optimization/78997
33551 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33552 name condition properly.
33554 2017-01-09 Richard Biener <rguenther@suse.de>
33557 * dwarf2out.c (is_cxx): New overload with context.
33558 (is_naming_typedef_decl): Use it.
33560 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33562 * invoke.texi (Option Summary): Correct spacing in option lists
33563 and add line breaks to fix over-long lines.
33565 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33567 PR middle-end/17660
33569 * extend.texi (Common Variable Attributes): Add xref to GCC
33570 Internals manual to explain mode attribute keywords.
33572 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33575 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33576 and Preprocessor Options.
33577 (Options for Linking): Document -pthread here....
33578 (RS/6000 and PowerPC Options): ...not here.
33579 (Solaris 2 Options): ...or here.
33580 * doc/cppopts.texi: Document -pthread.
33582 2017-01-08 Martin Sebor <msebor@redhat.com>
33584 PR middle-end/77708
33585 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33586 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33587 New member functions.
33588 (format_directive): Used them.
33590 (pass_sprintf_length::handle_gimple_call): Same.
33591 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33592 to avoid truncation for any argument.
33593 (extract_affine_mul): Same.
33594 * tree.c (get_file_function_name): Same.
33596 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33598 PR middle-end/77484
33599 * predict.def (PRED_INDIR_CALL): Set to 86.
33601 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33603 PR preprocessor/54124
33604 * doc/cppopts.texi: Reformat -d subtable to list the full name
33605 of the options. Add cross-reference to the docs for the general
33606 compiler -d options.
33607 * doc/invoke.texi (Developer Options): Add cross-reference to the
33608 preprocessor-specific -d option documentation.
33610 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33612 PR preprocessor/13498
33613 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33614 redudant material, and reflect new command-line options.
33615 (System Headers): Likewise.
33617 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33619 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33620 -isystem, and -idirafter. Copy-edit.
33621 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33622 default for -ftrack-macro-expansion. Delete obsolete and
33623 badly-formatted implementation details about -fdebug-cpp output.
33624 * doc/cppwarnopts.texi: Copy-edit.
33626 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33629 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33630 that the transition from a max line width >= 1<<10 to narrower
33631 lines works correctly.
33633 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33635 * doc/options.texi (PerFunction): New.
33636 * opt-functions.awk (switch_flags): Map both Optimization and
33637 PerFunction to CL_OPTIMIZATION.
33638 * opth-gen.awk: Test for PerFunction flag along with
33640 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33641 it only when the latter is present. Skip those that don't in
33642 the hash function generator.
33643 * common.opt (fvar-tracking): Mark as PerFunction instead of
33645 (fvar-tracking-assignments): Likewise.
33646 (fvar-tracking-assignments-toggle): Likewise.
33647 (fvar-tracking-uninit): Likewise.
33649 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33651 PR translation/79018
33652 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33655 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33658 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33659 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33660 TARGET_LONG_JUMP_TABLE_OFFSETS.
33661 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33662 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33663 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33664 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33665 * config/m68k/m68k.md (tablejump expander): Likewise.
33666 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33667 TARGET_LONG_JUMP_TABLE_OFFSETS.
33668 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33669 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33671 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33672 David Holsgrove <david.holsgrove@xilinx.com>
33674 * common/config/microblaze/microblaze-common.c
33675 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33676 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33678 * config/microblaze/microblaze.c (microblaze_must_save_register)
33679 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33681 (microblaze_eh_return): New function.
33682 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33683 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33684 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33685 * config/microblaze/microblaze.md (eh_return): New pattern.
33687 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33689 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33690 GCC_DIAGNOSTIC_STRINGIFY): Define.
33692 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33694 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33696 * config/arm/arm.md (<mcrr>): New.
33698 * config/arm/arm.c (arm_arch5te): New.
33699 (arm_option_override): Set arm_arch5te.
33700 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33702 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33703 (arm_mcrr_qualifiers): ... this. New.
33704 (MRRC_QUALIFIERS): Define to...
33705 (arm_mrrc_qualifiers): ... this. New.
33706 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33708 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33709 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33710 (MRRCI, mrrc, MRRC): New.
33711 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33712 VUNSPEC_MRRC2): New.
33714 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33716 * config/arm/arm.md (<mcr>): New.
33718 * config/arm/arm.c (arm_coproc_builtin_available): Add
33719 support for mcr, mrc, mcr2 and mrc2.
33720 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33721 (arm_mcr_qualifiers): ... this. New.
33722 (MRC_QUALIFIERS): Define to ...
33723 (arm_mrc_qualifiers): ... this. New.
33724 (MCR_QUALIFIERS): Define to ...
33725 (arm_mcr_qualifiers): ... this. New.
33726 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33728 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33729 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33730 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33731 VUNSPEC_MRC2): New.
33733 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33735 * config/arm/arm.md (*ldc): New.
33739 * config/arm/arm.c (arm_coproc_builtin_available): Add
33740 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33741 (arm_coproc_ldc_stc_legitimate_address): New.
33742 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33743 'qualifier_const_pointer'.
33744 (LDC_QUALIFIERS): Define to...
33745 (arm_ldc_qualifiers): ... this. New.
33746 (STC_QUALIFIERS): Define to...
33747 (arm_stc_qualifiers): ... this. New.
33748 * config/arm/arm-protos.h
33749 (arm_coproc_ldc_stc_legitimate_address): New.
33750 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33751 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33752 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33753 stc2, stcl, stc2l): New.
33754 * config/arm/constraints.md (Uz): New.
33755 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33756 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33757 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33758 VUNSPEC_STC2L): New.
33760 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33762 * config/arm/arm.md (<cdp>): New.
33763 * config/arm/arm.c (neon_const_bounds): Rename this ...
33764 (arm_const_bounds): ... this.
33765 (arm_coproc_builtin_available): New.
33766 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33767 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33768 (CDP_QUALIFIERS): Define to...
33769 (arm_cdp_qualifiers): ... this. New.
33771 (arm_expand_builtin_args): Add case for 6 arguments.
33772 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33773 (arm_const_bounds): ... this.
33774 (arm_coproc_builtin_available): New.
33775 * config/arm/arm_acle.h (__arm_cdp): New.
33777 * config/arm/arm_acle_builtins.def (cdp): New.
33779 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33780 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33781 'arm_const_bounds'.
33782 * config/arm/types.md (coproc): New.
33783 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33784 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33785 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33786 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33788 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33790 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33791 (UBINOP_QUALIFIERS): New.
33793 (acle_builtin_data): New. Change comment.
33794 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33795 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33796 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33797 arm_acle_builtins.def.
33798 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33799 (arm_init_acle_builtins): New.
33800 (CRC32_BUILTIN): Remove.
33801 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33802 crc32cb, crc32ch and crc32cw.
33803 (arm_init_crc32_builtins): Remove.
33804 (arm_init_builtins): Use arm_init_acle_builtins rather
33805 than arm_init_crc32_builtins.
33806 (arm_expand_acle_builtin): New.
33807 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33808 * config/arm/arm_acle_builtins.def: New.
33810 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33812 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33813 (arm_builtin_datum): ... this.
33814 (arm_init_neon_builtin): Rename to ...
33815 (arm_init_builtin): ... this. Add a new parameters PREFIX
33816 and USE_SIG_IN_NAME.
33817 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33818 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33819 'arm_builtin_datum'.
33820 (arm_init_vfp_builtins): Likewise.
33821 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33822 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33823 (arm_expand_neon_args): Rename to ...
33824 (arm_expand_builtin_args): ... this. Rename builtin_arg
33825 enum values and differentiate between ARG_BUILTIN_MEMORY
33826 and ARG_BUILTIN_NEON_MEMORY.
33827 (arm_expand_neon_builtin_1): Rename to ...
33828 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33829 values, arm_expand_builtin_args and add bool parameter NEON.
33830 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33831 (arm_expand_vfp_builtin): Likewise.
33832 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33834 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33836 PR middle-end/77484
33837 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33838 * predict.c (tree_estimate_probability_bb): Reverse direction of
33839 polymorphic call predictor.
33841 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33843 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33844 (determine_pass_name_match): ...this new function and...
33845 (should_skip_pass_p): ...this new function.
33847 2017-01-06 Nathan Sidwell <nathan@acm.org>
33849 * ipa-visibility.c (function_and_variable_visibility): Reformat
33850 comments and long lines. Remove extrneous if.
33851 * symtab.c (symtab_node::make_decl_local): Fix code format.
33852 (symtab_node::set_section_for_node): Fix comment typo.
33854 2017-01-06 Martin Liska <mliska@suse.cz>
33857 * lra-constraints.c: Rename invariant to lra_invariant.
33858 * predict.c (set_even_probabilities): Initialize e to NULL.
33860 2017-01-05 Martin Sebor <msebor@redhat.com>
33862 PR tree-optimization/78910
33863 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33864 (format_integer): Correct off-by-one error in the handling
33865 of precision with negative numbers in signed conversions..
33867 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33869 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33871 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33873 PR tree-optimization/71016
33874 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33875 factor_out_conditional_conversion. Formatting fix.
33876 (factor_out_conditional_conversion): Add cond_stmt argument.
33877 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33878 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33881 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33883 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33884 read-rtl-function.o, and selftest-rtl.o.
33885 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33886 (selftest::aarch64_test_loading_full_dump): New function.
33887 (selftest::aarch64_run_selftests): New function.
33888 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33889 selftest::aarch64_run_selftests.
33890 * config/i386/i386.c
33891 (selftest::ix86_test_loading_dump_fragment_1): New function.
33892 (selftest::ix86_test_loading_call_insn): New function.
33893 (selftest::ix86_test_loading_full_dump): New function.
33894 (selftest::ix86_test_loading_unspec): New function.
33895 (selftest::ix86_run_selftests): Call the new functions.
33896 * emit-rtl.c (maybe_set_max_label_num): New function.
33897 * emit-rtl.h (maybe_set_max_label_num): New decl.
33898 * function.c (instantiate_decls): Guard call to
33899 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33900 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33902 * gensupport.c (gen_reader::gen_reader): Pass "false"
33903 for new "compact" param of rtx_reader.
33904 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33905 rather than an empty string for NULL strings.
33906 * read-md.c: Potentially include config.h rather than bconfig.h.
33907 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33908 (have_error): New global, copied from errors.c.
33909 (md_reader::read_name): Rename to...
33910 (md_reader::read_name_1): ...this, adding "out_loc" param,
33911 and converting "missing name or number" to returning false, rather
33913 (md_reader::read_name): Reimplement in terms of read_name_1.
33914 (md_reader::read_name_or_nil): New function.
33915 (md_reader::read_string): Handle "(nil)" by returning NULL.
33916 (md_reader::md_reader): Add new param "compact".
33917 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33918 (md_reader::read_file): New method.
33919 * read-md.h (md_reader::md_reader): Add new param "compact".
33920 (md_reader::read_file): New method.
33921 (md_reader::is_compact): New accessor.
33922 (md_reader::read_name): Convert return type from void to file_location.
33923 (md_reader::read_name_or_nil): New decl.
33924 (md_reader::read_name_1): New decl.
33925 (md_reader::m_compact): New field.
33926 (noop_reader::noop_reader): Pass "false" for new "compact" param
33928 (rtx_reader::rtx_reader): Add new "compact" param.
33929 (rtx_reader::read_rtx_operand): Make virtual and convert return
33930 type from void to rtx.
33931 (rtx_reader::read_until): New decl.
33932 (rtx_reader::handle_any_trailing_information): New virtual function.
33933 (rtx_reader::postprocess): New virtual function.
33934 (rtx_reader::finalize_string): New virtual function.
33935 (rtx_reader::m_in_call_function_usage): New field.
33936 (rtx_reader::m_reuse_rtx_by_id): New field.
33937 * read-rtl-function.c: New file.
33938 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33939 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33940 (selftest::verify_three_block_rtl_cfg): New decl.
33941 * read-rtl-function.h: New file.
33942 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33943 For host, include function.h, memmodel.h, and emit-rtl.h.
33944 (one_time_initialization): New function.
33945 (struct compact_insn_name): New struct.
33946 (compact_insn_names): New array.
33947 (find_code): Handle insn codes in compact dumps.
33948 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33949 (bind_subst_iter_and_attr): Likewise.
33950 (add_condition_to_string): Likewise.
33951 (add_condition_to_rtx): Likewise.
33952 (apply_attribute_uses): Likewise.
33953 (add_current_iterators): Likewise.
33954 (apply_iterators): Likewise.
33955 (initialize_iterators): Guard usage of apply_subst_iterator with
33956 #ifdef GENERATOR_FILE.
33957 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33958 (md_reader::read_mapping): Likewise.
33959 (add_define_attr_for_define_subst): Likewise.
33960 (add_define_subst_attr): Likewise.
33961 (read_subst_mapping): Likewise.
33962 (check_code_iterator): Likewise.
33963 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33965 (one_time_initialization): New function.
33966 (rtx_reader::read_until): New method.
33967 (read_flags): New function.
33968 (parse_reg_note_name): New function.
33969 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33970 Handle reuse_rtx ids.
33971 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33972 Add parsing support for RTL dumps, mirroring the special-cases in
33973 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33974 values, and calling handle_any_trailing_information.
33975 (rtx_reader::read_rtx_operand): Convert return type from void
33976 to rtx, returning return_rtx. Handle case 'e'. Call
33977 finalize_string on XSTR and XTMPL fields.
33978 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33979 "(nil)" values were omitted. Call the postprocess vfunc on the
33981 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33982 class ctor. Initialize m_in_call_function_usage. Call
33983 one_time_initialization.
33984 * rtl-tests.c (selftest::test_uncond_jump): Call
33985 set_new_first_and_last_insn.
33986 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33987 * selftest-rtl.c: New file.
33988 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33989 (selftest::get_insn_by_uid): New decl.
33990 * selftest-run-tests.c (selftest::run_tests): Call
33991 read_rtl_function_c_tests.
33992 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33993 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33996 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33998 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33999 operands in a special way. Assert that pos+len <= mode precision.
34001 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34003 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34004 3 argument Alias with unlimited for the negative form.
34005 (fno-vect-cost-model): Removed.
34007 2017-01-05 Martin Liska <mliska@suse.cz>
34009 * hsa-gen.c (gen_hsa_divmod): New function.
34010 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34012 2017-01-05 Martin Liska <mliska@suse.cz>
34015 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34018 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34020 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34021 small constant length operands.
34023 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34025 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34026 between loop iterations.
34028 2017-01-05 Martin Liska <mliska@suse.cz>
34031 * gimplify.c (gimplify_decl_expr): Compare to
34032 asan_poisoned_variables instread of checking flags.
34033 (gimplify_target_expr): Likewise.
34034 (gimplify_expr): Likewise.
34035 (gimplify_function_tree): Conditionally initialize
34036 asan_poisoned_variables.
34038 2017-01-04 Jeff Law <law@redhat.com>
34040 PR tree-optimizatin/78812
34041 * rtl.h (contains_mem_rtx_p): Prototype.
34042 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34043 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34044 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34045 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34046 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34048 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34050 * input.c (assert_char_at_range): Default-initialize actual_range.
34052 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34054 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34055 to match the caller.
34057 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34059 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34060 insns after final jump in test to emit dummy move.
34062 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34064 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34065 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34067 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34069 * multiple_target.c (create_dispatcher_calls): Init e_next.
34070 * tree-ssa-loop-split.c (split_loop): Init border.
34071 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34074 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34079 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34080 (altivec_register_operand): Do not return true if the operand
34081 contains a SUBREG mixing SImode and SFmode.
34082 (vsx_register_operand): Likewise.
34083 (vsx_reg_sfsubreg_ok): New predicate.
34084 (vfloat_operand): Do not return true if the operand contains a
34085 SUBREG mixing SImode and SFmode.
34086 (vint_operand): Likewise.
34087 (vlogical_operand): Likewise.
34088 (gpc_reg_operand): Likewise.
34089 (int_reg_operand): Likewise.
34090 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34091 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34092 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34094 (rs6000_emit_move_si_sf_subreg): New helper function.
34095 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34096 fixup SUBREGs involving SImode and SFmode.
34097 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34098 numbers for the new peephole2 optimization.
34099 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34100 the GLIBC math library that do AND/IOR/XOR operations on single
34101 precision floating point.
34102 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34103 target macros to say whether we need to avoid SUBREGs mixing
34105 (TARGET_ALLOW_SF_SUBREG): Likewise.
34106 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34107 (UNSPEC_SI_FROM_SF): Likewise.
34108 (iorxor): Change spacing.
34109 (and_ior_xor): New iterator for AND, IOR, and XOR.
34110 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34111 (movdi_from_sf_zero_ext): Likewise.
34112 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34113 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34114 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34115 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34116 (fms<mode>4): Likewise.
34117 (fnma<mode>4): Likewise.
34118 (fnms<mode>4): Likewise.
34119 (nfma<mode>4): Likewise.
34120 (nfms<mode>4): Likewise.
34122 2017-01-04 Marek Polacek <polacek@redhat.com>
34125 * doc/invoke.texi: Document -Wpointer-compare.
34127 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34129 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34132 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34133 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34136 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34139 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34140 documentation of the powerpc_popcntb_ok attribute.
34141 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34142 code to issue warning messages if a requested CPU configuration is
34143 not supported by the binary (assembler and loader) toolchain.
34144 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34145 made to define a built-in function that has been disabled.
34146 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34147 made to define a built-in function that has been disabled.
34148 (altivec_init_builtins): Add comment explaining why definition
34149 of the DST built-in functions is not preceded by an assertion
34150 check. Add assertions to prevent ICE if attempts are made to
34151 define an altivec predicate or an abs* built-in function that has
34153 (htm_init_builtins): Add comment explaining why definition of the
34154 htm built-in functions is not preceded by an assertion check.
34156 2017-01-04 Jeff Law <law@redhat.com>
34158 PR tree-optimizatin/67955
34159 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34160 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34161 the points-to solution does not include pt_null. Use DECL_PT_UID
34164 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34166 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34167 Use gen_int_mode instead of gen_lopwart for const_int operands.
34169 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34171 PR tree-optimization/71563
34172 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34173 out of range value - has low bits known to be zero.
34175 2017-01-04 Alan Modra <amodra@gmail.com>
34177 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34178 * configure: Regenerate.
34179 * config.in: Regenerate.
34181 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34184 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34185 a substring of the message, but strcmp with the whole message. Ifdef
34186 ENABLE_NLS, translate the message first using dgettext.
34188 2017-01-03 Jeff Law <law@redhat.com>
34190 PR tree-optimizatin/78856
34191 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34192 (mark_threaded_blocks): Remove code to truncate thread paths that
34193 cross multiple loop headers. Instead invalidate the cached loop
34194 iteration information and handle case of a thread path walking
34195 into an irreducible region.
34197 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34200 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34201 assertions. Add support for doing the signbit if the IEEE 128-bit
34202 floating point value is in a GPR.
34203 * config/rs6000/rs6000.md (Fsignbit): Delete.
34204 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34205 Update the length attribute if the value is in a GPR.
34206 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34207 the sign or zero extension instruction, since the value is always 0/1.
34208 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34211 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34212 extracting SImode to a GPR register so that we can generate a
34213 store, limit the vector to be in a traditional Altivec register
34214 for the vextuwrx instruction.
34216 2017-01-03 Ian Lance Taylor <iant@google.com>
34218 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34220 2017-01-03 Martin Sebor <msebor@redhat.com>
34222 PR tree-optimization/78696
34223 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34224 precision. Use MPFR for %f for greater fidelity. Correct handling
34226 (pass_sprintf_length::compute_format_length): Set width and precision
34227 specified by asrerisk to void_node for vararg functions.
34228 (try_substitute_return_value): Adjust dump output.
34230 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34232 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34234 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34236 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34237 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34238 -mlra/-mno-lra was passed to the compiler.
34240 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34242 PR rtl-optimization/65618
34243 * emit-rtl.c (try_split): Move initialization of "before" and
34244 "after" to just before the call to emit_insn_after_setloc.
34246 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34248 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34250 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34252 * dwarf2out.c (gen_enumeration_type_die): When
34253 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34255 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34257 PR tree-optimization/78965
34258 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34259 Change first argument from const call_info & to call_info &. For %n
34260 set info.nowrite to false.
34262 PR middle-end/78901
34263 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34264 possibly throwing calls.
34266 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34267 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34268 and fns handling, rather than in a separate case SSA_NAME.
34270 2017-01-02 Jeff Law <law@redhat.com>
34272 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34273 fixes for first_period and second_period variables.
34275 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34278 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34279 (*insvqi_1): New insn pattern.
34280 (*insvqi_1_mem_rex64): Ditto.
34281 (*insvqi_2): Ditto.
34282 (*insvqi_3): Rename from *insvqi.
34284 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34286 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34288 * doc/cfg.texi (Edges): Remove reference to Java.
34289 (Maintaining the CFG): Ditto.
34291 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34293 PR middle-end/77674
34294 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34295 transparent aliases.
34297 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34299 PR middle-end/77484
34300 * predict.def (PRED_CALL): Update hitrate.
34301 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34302 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34303 into direct/indirect/polymorphic variants.
34305 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34307 Update copyright years.
34309 * gcc.c (process_command): Update copyright notice dates.
34310 * gcov-dump.c (print_version): Ditto.
34311 * gcov.c (print_version): Ditto.
34312 * gcov-tool.c (print_version): Ditto.
34313 * gengtype.c (create_file): Ditto.
34314 * doc/cpp.texi: Bump @copying's copyright year.
34315 * doc/cppinternals.texi: Ditto.
34316 * doc/gcc.texi: Ditto.
34317 * doc/gccint.texi: Ditto.
34318 * doc/gcov.texi: Ditto.
34319 * doc/install.texi: Ditto.
34320 * doc/invoke.texi: Ditto.
34322 Copyright (C) 2017 Free Software Foundation, Inc.
34324 Copying and distribution of this file, with or without modification,
34325 are permitted in any medium without royalty provided the copyright
34326 notice and this notice are preserved.